A little good news before this week’s parade of AI-powered martech: Adobe says its Digital Academy has reached more than 3 million learners this year as it expands free creative, digital, and AI training around the world. The company’s goal is to reach 30 million learners by 2030.

That includes nearly 400,000 young people in India who have gained access to creative and AI-enabled digital skills through Adobe’s partnership with UNICEF Generation Unlimited. Adobe is also working with Indian state governments to bring digital creativity and AI skills to nearly 8 million students and teachers.

There’s something for marketers, too. Adobe and LinkedIn offer free AI training for marketers in 47 languages, covering digital marketing, content creation, social media, data analytics, and agentic workflows — a useful reminder that the AI boom isn’t only about building new tools but also about giving more people the skills to use them.

Here are this week’s other AI-powered martech releases and news:

September 17, 2026

Azoma set out criteria for evaluating generative engine optimization tools that measure brand recommendations across AI shopping agents. The software tracks prompt-level brand visibility across ChatGPT, Google Gemini, Perplexity, Amazon Rufus, and Walmart Sparky, and analyzes citation sources to guide updates to content and product data.

Beasley Media Group deployed Futuri’s TopLine Enterprise platform across all operating markets to support advertising sales workflows. AI agents perform prospecting research, draft outreach communications, build media plans, generate audio and video spec spots, and compile campaign performance reports while account executives retain approval control.

Brand Engagement Network, in partnership with Cataneo, introduced interactive commercial break technology for broadcast and streaming media. Embedded AI agents converse with viewers during ad breaks to answer product questions, generate personalized recommendations, and process purchases directly within the viewing interface.

Constant Contact launched an integration connector for Anthropic’s Claude to support conversational campaign creation. Claude drafts email content, generates social media posts, suggests subject lines, formats calls to action, and transfers campaign files directly into Constant Contact for sending and tracking.

Demandbase launched Mojo, an account-based marketing agent that automates cross-channel campaign operations. It defines target audiences, creates campaign briefs, manages ad distribution across connected platforms, detects tracking errors, and adjusts execution based on performance data from previous campaigns.

Fairing launched Advanced Attribution to track purchase drivers across podcasts, television, content creators, and AI search tools. It presents dynamic follow-up survey questions to post-purchase buyers, converts open-text responses into structured channel data via auto-suggest lists, and measures offline influence.

Informa TechTarget launched Buyer Intelligence to process person-level intent signals into GTM strategy insights. The tools identify active business pain points from content consumption patterns, generate account-specific sales pitch recommendations, and map active customer market segments.

Marvin launched Live Intercept to conduct automated customer research inside websites and digital products. An AI interviewer follows custom discussion guides, asks follow-up questions in real time, redacts sensitive participant data, and aggregates video recordings and transcripts into a centralized knowledge repository.

Nimble launched AI Sequences to automate multi-channel sales follow-up workflows. The software converts a single prompt into an outreach sequence, generates personalized email copy, schedules manual phone call reminders, and sets follow-up task timing based on account context.

PilotDeck launched a website growth platform to manage SEO, generative engine optimization, and content distribution workflows. It analyzes search intent, generates keyword-targeted articles and brand visuals, and evaluates site visibility across AI answer engines.

Pipedrive launched Nova, an automated meeting intelligence tool available across all subscription tiers. It records sales calls, transcribes meeting audio, extracts key conversational insights, updates CRM fields, and suggests next steps for sales.

RSVPify launched RSVPify Cue and the RSVPify MCP connector to automate event management tasks.RSVPify Cue processes natural language instructions to build registration forms, analyze guest attendance data, and summarize event revenue, while the MCP connector links event datasets to external AI tools.

Snipp Interactive launched a ChatGPT plugin and Claude connector to manage consumer loyalty and promotional programs. The tools process receipt validations, manage reward redemptions, analyze purchase patterns, and handle customer query responses through conversational interfaces.

Twilio integrated OpenAI’s GPT-Live-1 API into its telephony architecture to support full-duplex voice applications. The integration streams real-time audio between phone networks and AI models, supports voice interruptions, evaluates caller sentiment, and processes multi-language customer interactions.

Upfluence updated its creator marketing software into an autonomous platform powered by the Jaice engine. The software analyzes creator performance metrics, automates outreach, negotiates campaign fees, manages content approval steps, processes payouts, and screens video assets for brand safety compliance.

Zendesk introduced Specialized AI Agents pre-configured for specific industry service workflows. The software resolves customer service inquiries, executes multi-step business transactions, routes complex tickets to human representatives, and adapts responses using customer knowledge base records.

September 10, 2026

AI Mini Stores announced a managed ecommerce model that combines automated execution with human strategic management. AI agents perform product research, draft content, triage customer support inquiries, manage marketing campaigns, monitor inventory, summarize analytics, and detect fraud risks.

Augeo integrated SpaceXAI Grok reasoning and live X sentiment data into its THEO orchestration engine. The software generates personalized redemption offers for individual users and operates a Loyalty Media Network across rewards, promotions, and advocacy channels.

BounceBack AI launched a platform that processes hard email bounces to identify lost contacts and track career transitions. It maps job changes and highlights timing opportunities for sales teams to reconnect with former prospective buyers.

Certinia expanded its Veda suite by adding 14 autonomous agents and expanding its Intelligent Actions library to 135 tools. The platform coordinates professional services, customer success, and accounting processes across the enterprise lifecycle while executing general ledger actions through Model Context Protocol connections.

Fimo introduced an AI content management system that converts generated code into functional, self-updating websites. Software agents manage continuous site operations, including visual editing, site translations, content publishing, and search engine optimization.

Gupshup launched a self-serve Voice AI platform that lets users build, test, and deploy voice agents for phone interactions. AI agents conduct phone calls, resolve support requests, qualify inbound leads, schedule appointments, and send payment reminders across PSTN and WhatsApp telephony networks.

Klaviyo opened its marketing platform to external AI agents by making 260 MCP tools and 490 APIs accessible through third-party interfaces. The expansion introduces SQL capabilities within the Klaviyo Data Platform, translating natural-language prompts into technical queries that retrieve customer, catalog, and event metrics.

Madison Logic introduced AI Planner, an account-based marketing tool that converts campaign goals into audience segments, channel strategies, and budget allocations.It analyzes intent signals, identifies relevant topics, and maps existing content assets to target personas.

Markup AI expanded Markup AI for Marketers to Microsoft Word, ChatGPT, and Claude. Content Guardian Agents evaluate human and generated text against brand voice guidelines, flag outdated information, and check content structure for AI search citation readiness.

Observe.AI launched Performance Agents to automate contact center coaching workflows. The software evaluates conversation transcripts, identifies agent performance patterns, builds customized development plans, and tracks performance changes over time.

Qualtrics unveiled the XM Data & AI platform to model customer behavior and simulate business decisions. The software creates digital twins from customer data to test pricing adjustments, product changes, and operational policy shifts prior to implementation.

Seedtag launched the Seedtag Emotion Quotient to monitor emotional intensity across weekly online cultural moments. Its Neuro-Contextual AI, named Liz, analyzes digital content in real time to categorize layered audience emotions, interest, and intent without using personal data.

Sembly AI launched Sembly 3.0, an execution platform that converts documents, meeting data, and CRM records into branded deliverables. It researches target clients, extracts appropriate brand assets, and generates custom presentation decks, sales proposals, and case studies in over 45 languages.

Storyblok introduced Storyblok Agents, a conversational tool within its headlless content management system. The software processes natural language commands to update site content, run automated translation workflows, organize digital asset libraries, and generate multi-language copy across digital touchpoints.

Stravito released AI Persona Builder, a research platform that generates interactive buyer personas from internal brand studies. The system creates queryable customer profiles, scores individual traits against source documents, attributes statements to original files, and flags conflicting research data.

Target updated its app with digital tools that simplify product search, review synthesis, and order replenishment. Photo Search processes uploaded images to find matching products, Review Insights summarizes customer feedback into thematic groups, and Buy Again uses past shopping behavior to display relevant purchase recommendations.

TechEsperto expanded operations beyond CRM integration to develop connected web applications, mobile software, and AI products. The company builds cross-platform software architectures that sync enterprise CRM data with mobile and web platforms.

Telesign partnered with B2Metric to integrate customer analytics with messaging channels.B2Metric analyzes customer behavior to predict engagement opportunities, while Telesign delivers cart reminders and promotional offers through WhatsApp and email.

Upfluence converted its creator marketing software into an autonomous platform powered by its Jaice engine. The software analyzes creator performance, conducts outreach, negotiates campaign fees, manages content approvals, processes payouts, and screens creator videos for brand safety risks.

Video Generator launched a free AI video creation platform that converts text inputs into visual content. Models analyze user instructions to format, assemble, and render complete video sequences for digital campaigns and social channels.

Videowise launched AI Visibility to index user-generated video content for AI tools. The tool transcribes video dialogue, labels product visuals, extracts product attributes, and structures visual data for AI search engines to discover and cite.

Vistatec launched Vistatec Data to manage global data services for model development. The division collects, annotates, evaluates, and validates dataset formats across text, image, audio, and video media to support model alignment and testing workflows.

September 3, 2026

ActiveCampaign launched Active Intelligence: Wavelength to customize email marketing campaigns and automated customer workflows. The system applies a context engine that continuously analyzes account history, past campaign performance, website activity, and over 500 business signals to draft targeted emails, construct segments, and identify broken customer journeys.

Alchemer launched Iris, a customer experience management platform designed to automate feedback management and operational responses. The software applies AI to analyze incoming customer feedback signals, surface root causes, and trigger automated follow-up tasks across enterprise systems.

Archive launched Archie, an AI agent for creator marketing campaigns. The software applies artificial intelligence to analyze brand descriptions, evaluate creator content histories, verify audience data, and return shortlists of pre-vetted creators for specific marketing campaigns.

Auth0 released new identity security capabilities to secure machine-to-machine interactions and autonomous processes. The updates use AI risk models to continuously monitor machine authentication traffic, detect anomalous access requests, and enforce identity governance rules across cloud applications.

AutoBacklinks launched an AI agent to automate search engine link-building operations. The software discovers target websites, evaluates domain relevance, generates personalized outreach messages, and tracks backlink placements.

Bazaarvoice introduced the AI Visibility Package to format brand product data and user-generated content for generative search engines. It structures customer reviews, product details, and visual media, ensuring LLMs index and cite brand information during conversational shopping queries.

BizzContacts launched the Free Disposable Email Checker tool to assess lead data quality. The software validates email syntax, extracts web domains, and flags temporary or disposable email addresses used on web forms.

Comcast Technology Solutions unveiled a suite of video workflow tools built on its VideoAI platform for broadcasters and streaming operators. The software converts landscape video to vertical formats, translates audio tracks with voice matching, identifies optimal advertising breakpoints, and performs automated quality checks.

Crescendo launched the Crescendo Customer Experience Platform to manage and automate customer support operations. It uses AI to resolve customer inquiries across channels, assist live support representatives, score conversation quality, and optimize staffing schedules.

CrunchJunkie expanded its search analytics software to monitor brand citations inside generative search engines. The platform measures brand visibility, sentiment, and positioning metrics across conversational platforms, including ChatGPT, Gemini, Perplexity, and Google AI Overviews.

Dreamdata released new analytics features to track how buyers use conversational platforms during the purchasing journey. It analyzes brand citations in large language models, calculates attribution signals, and evaluates the influence of artificial intelligence recommendations on digital pipeline conversion rates.

Findabl AI expanded its marketing capabilities as a unified platform for multi-format digital content generation. It uses AI to synthesize custom marketing images, produce synthetic voiceovers, generate marketing videos, and edit long-form promotional copy from unified creative prompts.

Gearset updated its Salesforce DevOps platform with new continuous deployment and automated testing tools. It scans release pipelines, predicts deployment risks, analyzes metadata dependencies, and suggests fixes for deployment errors.

Givsly extended its values-based advertising platform to political campaign operations. It uses AI to align digital advertisement delivery with specific voter values, track campaign spend allocations, and route designated advertising proceeds to partner non-profit organizations.

Helical Insight released an open-source business intelligence platform with integrated data analytics features. It uses AI to translate user text queries into SQL database commands, generate data visualizations, and construct interactive reporting dashboards.

Influencer launched Creator-First Answer Engine Optimization, a platform to track brand recommendations in AI search tools. It analyzes how conversational search engines cite creator content, measure social authority signals, and track brand visibility across generative answer platforms.

Innovid expanded its NIVO platform by integrating Meta Ads Model Context Protocol server capabilities. The platform automates connected television creative production, syncs dynamic ad variations with Meta social campaigns, and measures cross-channel campaign metrics.

Knorex expanded its digital advertising platform by releasing KAI Assist and integrating XPO Model Context Protocol server capabilities. The platform automates advertising campaign setups, generates audience segments, optimizes bid strategies, and adjusts channel budget allocations.

Local Media Consortium launched the AI Accelerator initiative to support local news publishing organizations. The initiative provides access to machine learning tools that summarize local news stories, automate content tagging, optimize subscription paywalls, and translate articles across digital publishing networks.

Operata launched the Control Layer platform to monitor the quality of customer service interactions across human and virtual support channels. It analyzes agent speech patterns, tracks system telemetry, monitors technical call quality, and flags communication friction points in real time.

Optimizely released purpose-built artificial intelligence models designed for enterprise digital marketing operations. They apply domain-specific machine learning models directly inside the marketing platform to generate campaign creative variations, predict content performance, and personalize website experiences.

Prospectr Digital launched a rebuilt lead generation platform designed for business-to-business sales teams. It uses AI to verify contact information, identify purchasing intent signals, and score inbound prospect accounts.

Sabio partnered with Gentoro to implement agentic artificial intelligence technologies across digital advertising operations. The system streamlines programmatic ad buying, automates creative assembly, processes audience data, and optimizes campaign delivery across connected television platforms.

SEOPulse launched an enterprise visibility platform designed to track brand positioning across generative answer engines. The application queries major LLMs, calculates brand share of voice, and identifies content gaps within conversational search results.

Sinch launched Mailgun Inspect on the Salesforce AgentExchange marketplace to streamline email campaign testing. It analyzes email rendering across inbox clients, detects spam trigger patterns, and evaluates sender domain reputation metrics before broadcast.

Skipio acquired RapidTalk to expand its customer engagement messaging software. The combined platform applies conversational AI models to respond to inbound customer text messages, schedule sales appointments, and execute automated follow-up sequences.

Webflow launched an agentic web development platform to automate website building and design workflows. It converts natural language descriptions into web layouts, writes site copy, optimizes page code, and maintains design systems across enterprise digital assets.

WizCommerce launched an AI CRM platform designed for wholesale product distributors. It can process purchase orders, analyze client buying histories, surface reorder opportunities, and generate tailored quotes for sales representatives.

Yext expanded its enterprise marketing platform to increase brand visibility across generative answer engines. It can format business data, update directory citations, and sync structured product information across conversational search engines.

August 27, 2026

3CLogic released the AI Agent Evaluator software to evaluate and score interactions with voice-based artificial intelligence agents. The software applies artificial intelligence to analyze spoken conversations, assess response quality, generate performance metrics, and trigger automated quality-assurance workflows within service management platforms.

Auxia launched Agent Studio, an operating system for marketing workflows. The platform applies artificial intelligence agents to analyze raw campaign data, map customer funnel drop-offs, generate creative briefs, and automate campaign changes across third-party marketing tools.

BioBrain introduced an AI-native operating system for market research and data intelligence. The system applies multimodal artificial intelligence algorithms to clean quantitative survey data, analyze qualitative voice and facial signals, and filter web conversation patterns across international markets.

Crescendo launched an artificial-intelligence customer-experience platform to manage enterprise customer service operations. The platform applies native artificial intelligence models across customer service interactions to handle incoming inquiries, route tasks, and update internal workflows.

Dun & Bradstreet integrated the D&B Commercial Graph data set into the Perplexity artificial intelligence engine. Perplexity applies its conversational artificial intelligence models to extract structured corporate identity, financial risk, and market data from the graph in response to user search queries.

Findabl AI launched Search Engine Optimization services targeting artificial intelligence search engines. The service uses artificial intelligence prompt research tools to analyze brand visibility, track citation metrics, and identify content gaps across platforms such as ChatGPT and Perplexity.

Hey DAN released the Hey DAN AI voice entry platform for sales operations. The platform applies machine learning models and speech recognition to transcribe phone conversations, extract key buying signals, and input structured interaction records directly into CRM systems.

Impact.com partnered with Microsoft to launch an affiliate marketing program for the Minecraft game franchise. The platform applies artificial intelligence algorithms to match content creators with campaign offers, track referral metrics, and detect invalid referral traffic.

Informa TechTarget launched the B2B AI Authority Index to track corporate visibility across digital research channels. The tool applies artificial intelligence analytical models to evaluate how business software buyers discover, evaluate, and cite brands within artificial intelligence search engines.

KERV expanded its CTV partnership with LG Ad Solutions to launch interactive video ad formats across international markets. The platform applies computer vision artificial intelligence to analyze video frames, identify on-screen objects, and generate clickable product overlays on smart televisions.

LoopMe expanded its PurchaseLoop advertising platform to optimize lower-funnel campaign performance metrics. The platform applies over 2,000 machine learning models to analyze brand campaign engagement signals, process mobile SDK behavioral data, and adjust programmatic ad bidding to drive application installation targets.

OneIMS launched an AI Visibility Audit service for industrial supply and manufacturing companies. The service applies natural-language AI queries across multiple answer engines to evaluate how accurately platforms describe specific B2B vendor capabilities.

Optimove integrated the OptiGenie conversational agent into its native marketing platform. The platform applies natural language processing models to interpret plain-text commands, construct customer segments, design campaign logic, and generate reporting dashboards.

Practifi launched Sentir, a CRM platform built for wealth management firms. The platform applies specialized artificial intelligence agents directly inside the database to summarize client conversations, draft meeting briefs, and calculate relationship metrics.

Similarweb launched an artificial intelligence advertisement tracking system within its intelligence suite. The software applies machine learning scraping algorithms to monitor ad placements, calculate share of voice, and track brand citations inside conversational artificial intelligence models like ChatGPT and Google AI Overviews.

Solitics launched SAAI, an agentic artificial intelligence platform designed for retail banking institutions. The software applies autonomous artificial intelligence agents to analyze real-time customer transactional data, determine intent, and send personalized financial communications across digital banking channels.

Sprout Social released its Trellis artificial intelligence agent for enterprise social media management. The software applies conversational artificial intelligence models to query historical social performance data, surface shifts in audience sentiment, and execute custom reporting tasks based on user prompts.

System1 launched Test Your Ad Screen, a testing platform for early-stage video ad concepts. The system applies predictive artificial intelligence algorithms to evaluate raw ad animatics and forecast the long-term impact on brand equity before final video production.

Temu developed an internal artificial intelligence marketing platform to purchase programmatic media. The platform applies machine learning algorithms to evaluate customer purchasing history, dynamic ad creative performance, and bidding efficiency without third-party web tracking cookies.

Veuno released a free AI Visibility Checker tool for business-to-business organizations. The application uses natural language processing tools to query major large language models and generate reports on how AI assistants describe and recommend specific brands relative to competitors.

The post The latest AI-powered martech news and releases appeared first on MarTech.

How well does marketing technology empower the CMO? Does martech deliver something the boardroom asks for? We all agree that marketing technology has never been more powerful. We have more customer data, more automation, better measurement, and now AI. 

Yet while the technology keeps expanding, something rather strange is happening to the function using it: the CMO is disappearing. And some CEOs are taking marketing into their own hands. That creates an uncomfortable question for martech.

Will this shift make martech more or less relevant?

Will martech disappear with the CMO? Or become more important than ever? And can martech continue to do business as usual? If marketing itself is moving beyond the marketing department, who will decide which martech is needed, or what all that marketing technology is actually supposed to do?

1. Congrats, marketing got promoted. finally!

“The CEO, not the marketing team, is responsible for the brand.” That was the recent argument from former ING & UBS CEO Ralph Hamers. His reasoning is simple: a brand is not a logo, campaign, or tagline. It is the experience people have with a company. And that experience is shaped by strategy, behavior, investments, and trade-offs across the entire organization. The idea itself is not new.

“Marketing is too important to be left to the marketing department.”

David Packard, co-founder of Hewlett-Packard

In both cases, the point is not that we don’t need marketing. It is that marketing, with a capital M, is bigger than the marketing department.

Think about it, marketing’s traditional mandate has already eroded over the decades. Of the original four Ps, marketing often only holds on to Promotion. Price moved to Sales, Place to Trade marketing or Channel, and Product to Product Management.

Now that erosion appears to have reached the CMO role itself. Forrester’s 2026 analysis of the Fortune 500 found that only 36% of companies still use the chief marketing officer title, down sharply from 49% in 2025. Marketing representation at the top is falling too: only 52% have a marketing executive on the executive team and/or reporting to the CEO, down from 58%.

At first sight, that looks like marketing is losing.

Forrester’s Ian Bruce offers another interpretation: reinvention. Traditional marketing, sales, and customer success responsibilities are increasingly combined under broader roles such as chief growth officer, chief commercial officer, and chief customer officer. The mandate is moving beyond the traditional marketing promotional activities toward accountability for growth across the customer lifecycle (re)incorporating the other Ps, like Price, Place, and Product.

That sounds less like a demotion and more like a promotion. Marketing finally becomes a business responsibility, not a department. There is just one rather important question: When marketing gets promoted, who is equipped to lead it?

2. But who is going to lead marketing?

Promoting marketing to the enterprise level solves one problem. It creates another: Who is actually equipped to take the marketing initiative?

Option 1: The CEO? The CEO owns the business, so if marketing becomes a business responsibility, marketing ultimately belongs there too. There is just one problem: most CEOs have very little marketing experience. According to McKinsey, only 10% of Fortune 250 CEOs have marketing experience, and just 4% have previously held a CMO-like role. More than 70% of Fortune 100 CEOs come from operations or finance. Mind the gap.

Option 2: The CMO? There is a problem on this side, too. The McKinsey CMO Growth Research Survey found a striking difference in how CEOs and CMOs measure marketing. In 2024, 70% of CEOs measured marketing by revenue growth and margin, up from 50% a year earlier. Among CMOs, that number barely moved: from 33% to 35%. That is quite a gap.

The CEO increasingly looks at marketing and asks: Did we grow? Did we make money? The CMO often answers with ROAS, CTR, CAC, CLV, brand awareness, MQLs, and another three-letter acronym before breakfast. 

None of those marketing metrics are wrong. The problem is translation. Marketing metrics need to connect to the business outcomes the CEO ultimately owns.

So we have an interesting navigation problem:

• The CEO has the business experience, but often lacks the marketing experience, i.e., more operational and less customer-focused.
• The CMO has the marketing experience, but too often struggles to translate it into business outcomes.

Firing the CMO and handing marketing to the CEO, therefore, feels a little like firing your co-driver while racing because the driver ultimately decides where the car goes. The answer is not choosing between CEO and CMO. It is closing the gap between them.

3. The CMO needs to earn the promotion

Keeping the CMO in the room is not enough either. If marketing is being promoted from a department to a business responsibility, the CMO needs to make that same transition. And there is evidence that this is not happening fast enough.

The McKinsey CMO Growth Research Survey shows that 79% of CMOs say they understand how their marketing KPIs align with company KPIs. That sounds reassuring, until you ask a more fundamental question: What is the return? Only 30% say their organization has a clear definition of marketing ROI.

That is a 49-percentage-point accountability gap. 

CMOs believe their metrics are aligned with the business, but most organizations cannot clearly define what that alignment is worth. More concerning, the gap is getting worse. KPI alignment fell from 88% to 79% in one year, while clarity around marketing ROI fell from 40% to 30%. Both dropped by around 10 percentage points.

This is not a stable weakness. It is active erosion.

And it helps explain the changes we saw in the previous sections. If marketing cannot demonstrate its contribution in terms the business understands, it should not be surprising when responsibility starts moving elsewhere. But replacing the CMO with a chief growth officer or chief commercial officer does not magically solve that problem. Changing the title does not create accountability. Neither does moving marketing underneath a CEO with little marketing experience.

The capability has to change. A broader marketing mandate requires a broader marketing leader: someone who understands customers and brands, but can also connect them to business outcomes like revenue, margin, growth, and ultimately enterprise value.

That is why the real question is not whether the CMO title survives. The real question is who takes responsibility for marketing, and whether that person can combine marketing expertise with business accountability.

Maybe the CMO isn’t being fired. Maybe the job description is being rewritten.

And that makes martech more relevant, not less. But its job description is being rewritten, too. If marketing becomes accountable for revenue, margin, growth, and customer value, martech can no longer justify itself through features, adoption, or marketing efficiency alone. It has to connect technology decisions to those same business outcomes.

Marketing got promoted. Now martech needs to earn the promotion too.

The post What happens to martech when the CMO disappears? appeared first on MarTech.

You probably have a drawer — or a box — at home where you put things that don’t quite belong anywhere else. This junk drawer isn’t a failure of your organizing skills, but a natural outcome of our ability to categorize things. That’s because things seldom fall into a small, finite number of hard categories but instead follow a long tail of smaller and smaller categories. So we choose a small number of the largest categories — pots, silverware, glasses, mugs, spatulas … and dedicate space to them — and everything else goes in the junk drawer.

The un-categorized category is called the residual category. Yes, it is a category, but we acknowledge that it is a category of the uncategorizable, which is in itself a kind of paradox!

In our marketing work, we may know this residual category as (direct)/(none), “Unassigned,” or (not set). In market research, it becomes that segment that just doesn’t fit, but we can’t describe, or the dreaded “Outliers.”

And all this is fine, and a natural, practical response to the long tail. But what happens when that category grows? If it represents 30% of our customers, can we really ignore it? What happens when a third of what we own is in the junk drawer?

Let’s take the case of the (direct)/(none) — analytics-speak for the group of people who just arrived at your platform with no provenance data. This is a growing and perplexing group. It used to be assumed that this was just people typing your address in directly, but no more. Nowadays, it is likely to be an artifact of privacy settings, modern browsers, or even from AI Agents visiting on someone’s behalf. These are your marketing ‘nones.’

Who is none, anyway?

To better understand how we handle the growing group of nones, we can turn to a parallel case in the social sciences, where the catch-all group has grown to be the largest category. 

For many years, researchers have been following US religious affiliations and beliefs — most notably Pew Research, which, in their 2024 report “Religious ‘Nones’ in America,” stated that 28% of people now check none as their affiliation. This compares to 16% back in 2007. 

The fascinating thing about this group is how diverse and surprising it is. It isn’t just a homogeneous group of the ungodly, as some might assume. The percentage of true atheists in here is small — just 17% (so less than 5% of the population) — the rest choose non-affiliation for a variety of different reasons. Lumping them together as a ‘none group’ misses the true insight into the group’s growth. That’s why Pew and others have worked to better understand the makeup of the nones and to better categorize them.

Pew did this by deeper, more nuanced research, going beyond the standard affiliation question and cross-referencing with other behavioral and attitudinal data. That let them discover some surprising things, such as the fact that 69% believed in God or some other higher power. Certainly not the picture that the simple none designation might have you believe.

Can we apply a similar approach to our nones? Yes.

How did they get here?

Start by hypothesizing what might be driving the category and why it is growing. Here are some of the ways that a customer might appear there:

• The Refuser: The active, conscious user who has made a deliberate choice to minimize tracking and visibility
• The Delegator: The person who outsourced the steps towards you, whether through agents, aggregators, AI search or communities
• The Defaulter: Someone who just accepted defaults in modern browsers or systems and became invisible
• The Typer: The person who just typed in your URL in a clean browser.

These four groups are vastly different, both in intent and in how we might reach them. Much industry discussion has moved from assuming that everyone in this group is a Typer to focusing on the Refuser.

However, the bulk of your nones will likely fall in either Defaulters (probably the largest group for most businesses) or the Delegators. While Defaulters are driven by changes in policy from tool providers (such as browsers), the growth in Delegators is a function of new technologies, such as AI.  As we see new types of intermediaries — especially agents — grow in usage, we will need to find better ways to size and understand this group. Ignoring this group could be a huge mistake.

The takeaway

So what can you do to better get to know your nones? Here are some general steps to drive better understanding and growth:

• Ask, “What is the residual category in my pipeline?” — Is it growing? Is it explainable?
• Stop treating the residual group as noise or a blind spot. Do the work to identify what it is made up of.
• When you’ve identified its components, build an attribution model. Find ways to size or discriminate between the subtypes. Maybe you need MMM or other research to identify what is really going on.
• Monitor growth in key groups, especially those driven by new technologies or behaviors.

It is imperative that we continually revisit our categorizations, especially when it comes to the residual category. Ignoring this group, especially as it grows, will mean lost opportunities and missed market movements. 

It should come as no surprise if new technologies mean you have to rethink your customers. Sometimes — as in the case of (direct)/(none) — you may need to get insight from other attribution methods, or review the tools being used. Whatever you do, you will need to understand that the nones are a real group of real customers — and understanding them may be critical to your business.

The post Why you need to know the ‘nones’ of your business appeared first on MarTech.

On the first day of Unbound26, HubSpot unveiled updates to change how people work with AI on its platform. Instead of choosing an agent or tool for each job, users tell Breeze Assistant what they want accomplished, and it selects the agents needed to do it.

This is part of a major overhaul of HubSpot built around what it calls “Growth Context,” which combines information about a company, its teams, and its customers. That context feeds the redesigned Breeze Assistant, a self-updating Smart CRM, and a new Context Home for identifying gaps in the information available to HubSpot’s AI.

Breeze Assistant now goes beyond answering questions. Give it a task, and it can assign specialized agents to do the work and produce campaign plans, reports, proposals, and other outputs using information from the CRM.

Keeping AI’s CRM context current

That approach depends on keeping the underlying customer data up to date. HubSpot’s new Smart CRM automatically captures and syncs calls, emails, and meetings rather than relying on employees to enter that information manually.

(Self-updating data is a sensitive topic for HubSpot and its customers. In July, customer backlash forced the company to reverse a terms of service change that let it take data from one customer and use it to improve another customer’s records.) 

Context Home gives teams a score showing how complete that information is and identifies gaps. Together, the two programs are designed to give HubSpot’s agents a continually updated picture of the business, rather than forcing users to assemble the necessary context whenever they assign a task to AI.

From finding a marketing problem to doing the work

Marketing Studio puts that approach into practice for marketers. It can surface something that needs attention, such as an AEO visibility score, an underperforming campaign segment, or leads that haven’t received follow-up. The marketer can then tell Breeze Assistant what needs fixing.

Breeze can bring in specialized content and campaign agents to plan the campaign, create content, and personalize nurture sequences. That moves the AI workflow beyond generating an individual piece of content or answering a question: the system can identify an issue and then coordinate agents to address it. 

HubSpot is also expanding the advertising channels marketers can manage from its CRM. It says it is the first CRM to integrate with ChatGPT Ads (Salesforce might disagree — it all depends on how you define “integrate.”) The integration sends leads generated in ChatGPT to HubSpot for follow-up and attribution alongside other channels. Microsoft Advertising is also being added. The plan is to connect ad engagement with contacts, pipeline, and revenue, and trigger follow-up in HubSpot.

The same model extends into sales

HubSpot is applying the same approach to sales. Its updated Prospecting Agent monitors more than 40 buying signals, assembles buying groups, and drafts personalized outreach. A new Mobile Notetaker captures conversations, while Deal Progression uses those transcripts to suggest CRM updates, draft follow-ups, and keep deal plans current with user approval.

Revenue Hub takes the automation further into the sales process by automatically creating quotes using deal information already in the CRM.

The larger change is in what HubSpot expects its users to manage. Rather than making marketers and sellers decide which AI agent should perform each part of a job, HubSpot wants them to specify the outcome and let Breeze Assistant coordinate the agents behind it. The Smart CRM and Growth Context form the foundation of that model, supplying the business and customer information agents need to act.

The post HubSpot rebuilds its platform around AI agents appeared first on MarTech.

The customer data platform (CDP) market has completely reinvented itself over the past 18 months, evolving rapidly in response to a landscape filled with fractured data silos and failed implementations. 

Historically, many organizations were left scarred by traditional CDP rollouts — enduring multi-million dollar, multi-year projects that focused too heavily on features rather than actual customer outcomes. 

Today, the integration of AI is reshaping how businesses manage customer experiences, flipping the old-school implementation model on its head to focus entirely on growth and tangible outcomes. 

In this episode of Conversations with MarTech, we sit down with Melissa Murray Bailey, CEO of BlueConic. Drawing on more than two decades of enterprise software and marketing experience from roles at LinkedIn and Hootsuite, Melissa outlines how a strong first-party data foundation coupled with AI can finally deliver on the decades-long promise of true personalization.

Key topics discussed in this episode 

The rise of agentic CDPs: Is “agent-washing” the new corporate tagline trend driven by executive AI mandates, or is there real transformative power in AI agents built from the ground up?

The holy grail of personalization: How AI allows B2C marketers to move past static automated email campaigns and manage infinite customer segments with real-time adjustments.

Creepy vs. helpful marketing: Where is the line for consumers when it comes to hyper-personalized experiences, and why being helpful is the ultimate test?

BlueConic’s strategic growth: The vision behind BlueConic’s recent acquisitions of Jebbit and Blueshift to build an end-to-end powerhouse for first-party data capture and customer engagement.

The industry’s “biggest lie”” Melissa pulls no punches when answering what the biggest lie the marketing industry tells itself today regarding customer experience.

Episode guide

00:00 Introduction
00:44 Meet Melissa Murray Bailey
01:40 Melissa takes us on a tour of the CDP market
03:41 CDPs as the ERP of the martech stack
04:15 Are people agent-washing the CDP market?
06:11 The good, the bad, and the ugly of AI agents
08:12 Are consumers ready for AI-level personalization?
10:32 BlueConic’s acquisition strategy in an M&A-fuled CDP space
14:30 What’s the biggest lie marketers tell themselves?

The post Does anyone understand the CDP market anymore? appeared first on MarTech.

OpenAI recently disclosed a fascinating, and more than a little unsettling, incident involving its AI models.

The models ran cybersecurity evaluations designed to see how well they could find and exploit software vulnerabilities. They were given difficult problems to solve in sandboxed environments, that is, isolated computer systems away from the main system. They weren’t given general internet access, and agents running separate evaluations weren’t supposed to communicate with each other.

But some of the agents found ways around those limitations. According to OpenAI, they created unauthorized communication channels, regained internet access, shared what they learned across separate evaluations, and exploited vulnerabilities that allowed them to access systems belonging to Hugging Face, an AI development platform.

The agents executed code on dozens of Hugging Face servers and obtained root access on one. Hugging Face later reconstructed roughly 17,600 actions associated with the intrusion.

That’s a fascinating cybersecurity story. But I’m not a cybersecurity expert. I’m a marketer. What caught my attention was how the agents pursued their objective: They pursued it extraordinarily well, even when doing so took them outside the boundaries their creators expected them to observe.

That’s the part marketers should pay attention to. As AI moves from generating things for us to making decisions and taking action on our behalf, we need to think carefully about the objectives we give it — and the boundaries we expect it to respect.

Be careful what you ask AI to do. Not because it might refuse, but because it might succeed.

The AI wasn’t trying to take over the world

Let’s dispense with the science-fiction version of this story. There’s no indication that these agents suddenly became evil or developed a secret desire for world domination. Instead, they were trying to solve a problem they’d been asked to solve.

OpenAI describes the agents as becoming hyper-focused on solving the cybersecurity evaluation. Some of the tasks were extraordinarily difficult. In fact, OpenAI later found that 198 of the 898 tasks had never been successfully solved by any of its models before the incident.

The agents didn’t give up when the obvious routes didn’t work. They kept looking for other ways to accomplish the objective. Unauthorized communication between agents made that even more powerful because agents could share discoveries and build on one another’s work.

In other words, the interesting part of this story isn’t that the AI failed. It’s that the AI got really, really good at pursuing the objective it was given. Marketers should be paying attention.

We’ve seen this movie before

Marketing has optimized toward objectives for decades, and we already know what can happen when the objective — or the metric we use as a proxy for it — is too narrowly defined.

For example, if you tell an email marketing team to maximize revenue, they may discover that sending more email generates more revenue. That sounds great until unsubscribe rates increase, engagement declines, deliverability suffers, and the long-term value of the email program starts heading in the wrong direction.

Or say you tell a demand generation team to maximize leads. You may get lots of leads. Unfortunately, they may not be the type of qualified leads sales is interested in calling.

Or perhaps you optimize digital advertising for clicks. You’re likely to discover that sensational headlines generate lots of them. Hello, clickbait, our old, worthless friend.

Or you might optimize solely for return on ad spend (ROAS) and then find that you’re extremely efficient at capturing customers who were already planning to buy from you, rather than generating incremental demand.

None of these are new problems. And in each case, the person or system doing the optimizing may be doing exactly what was asked of them.

The problem is that the definition of success was too narrow.

An objective isn’t a strategy

This becomes much more important as AI moves from generating things for us to doing things for us.

There’s a significant difference between asking an AI to “Write five subject lines for this email” and telling an AI agent to “Improve the performance of our email program.”

The first assignment has a relatively narrow scope. The second requires decisions. An AI agent pursuing that objective might analyze previous campaign performance, identify high-performing segments, adjust targeting, change cadence, generate creative, launch tests, and shift resources toward whatever appears to produce the best results.

That sounds wonderful. It’s one of the reasons marketers are excited about agentic AI.

But what exactly does improve performance mean? More opens? More clicks? More conversions? More immediate revenue? More incremental revenue? Greater customer lifetime value?

Just as important, what can’t the AI sacrifice in pursuit of that objective?

If you tell an AI agent to increase email revenue, for instance, the real job probably isn’t simply: Increase email revenue.

It’s closer to: Increase incremental revenue from email while maintaining healthy subscriber engagement, protecting deliverability, respecting customer preferences, and supporting long-term customer value.

Those are two very different assignments.

Define the whole job before you give it to an AI agent

I’ve written before about the importance of defining the goal before you decide how to measure success. If the job of an email isn’t to generate a click, don’t judge its success primarily on clicks. If the job is to drive registrations, measure registrations. If the job is to generate revenue, measure revenue.

I still believe that. But I think AI adds an important corollary: Measure the job. But make sure you’ve defined the whole job.

Most business objectives contain constraints we don’t bother stating because humans generally understand them from context. 

• When we say increase revenue, we mean increase revenue without destroying the customer relationship. 
• When we say generate more leads, we mean leads with a reasonable likelihood of becoming customers.
•  When we say reduce acquisition costs, we don’t necessarily mean eliminate every expensive acquisition source regardless of the lifetime value of the customers it produces.
• When we say complete this task, we generally assume that it means doing so without doing anything we wouldn’t approve of if a human employee did it.

With humans, those qualifications often go unstated. That usually works because experienced marketers can and do bring context to an assignment. We understand organizational norms, customer expectations, brand values, professional ethics, and the long-term consequences of our decisions.

As we delegate more decision-making to AI, I don’t think we can assume all of that context is implicit.

From prompting AI to managing AI agents

What does this mean if you’re using AI agents? It means that defining the objective isn’t nearly enough. You also need to specify:

• The constraints the AI should operate within.
• The metrics that define success.
• The decisions that still require human approval.

Much of the conversation about AI skills over the last few years has focused on prompting: how to write a better prompt, how to provide the right context, and how to give AI clearer instructions to get better output. All of that is still useful. But as AI becomes more agentic, I think another skill is at least as important: defining success.

For me, that comes down to four things:

• The objective: What outcome are we actually trying to produce?
• The guardrails: What can’t be sacrificed in pursuit of that outcome?
• The measures: How will we determine whether the result was genuinely successful?
• The escalation points: Which decisions require human judgment or approval?

That isn’t really prompting. It’s management.

It’s good marketing management whether the person — or thing — doing the work is human or artificial.

AI makes an old marketing problem bigger

The OpenAI incident is obviously an extreme example. Your marketing AI probably isn’t going to escape its sandbox and compromise a server because you asked it to increase conversion rates. At least I hope not.

But the underlying lesson applies to much more mundane marketing activities. Optimization has always had a weakness: The metric is usually a proxy for the outcome we really want. Humans compensate for imperfect proxies all the time. 

An experienced email marketer knows that a 10% increase in revenue isn’t necessarily good news if it required doubling send frequency and caused unsubscribes to spike. 

A demand generation marketer knows that a 40% increase in leads isn’t particularly exciting if none of them convert. 

A performance marketer should recognize that a spectacular ROAS isn’t necessarily spectacular if the advertising simply claimed credit for purchases that would have happened anyway.

AI can optimize faster than we can, across more variables than we can reasonably manage ourselves. That’s a tremendous opportunity. But if we give it an incomplete objective, it can optimize our mistakes faster, too.

As we hand more marketing execution and, eventually, more marketing decision-making over to AI, I think we need to ask a different question.

It’s no longer just, “How do I get AI to do what I want?” We also need to ask: “Have I defined what I want well enough that, if AI succeeds spectacularly, I’ll actually be happy with the result?”

Dig deeper: The AI era needs strategists grounded in expertise and guided by context

The post The problem with AI doing exactly what you ask appeared first on MarTech.

IAB Tech Lab is asking the programmatic advertising industry to weigh in on a proposed set of standard practices for how buyers and sellers conduct transactions.

Programmatic Standard Practices v1, released for public comment today, establishes shared business rules, norms, and principles for using existing programmatic standards. The goal is to create more consistency between buyers and sellers, reduce unnecessary transaction costs, and improve transparency across the programmatic ecosystem.

The proposal is the first deliverable from IAB Tech Lab’s Programmatic Governance Council, a senior-level industry group created to address business and governance issues in programmatic advertising. Rather than introducing another technical standard, the practices focus on how companies should use existing standards.

“Programmatic advertising depends on buyers and sellers having a common understanding of how transactions should work,” Anthony Katsur, CEO of IAB Tech Lab, said in a statement. “These practices are about making that understanding clearer and giving the industry a practical framework for using the standards that already exist.”

That distinction matters because technical standards alone don’t guarantee companies will implement or use them consistently. Shared practices could give buyers and sellers clearer expectations about how programmatic transactions should operate while reducing some of the friction created by differing interpretations and business practices.

The proposal is also an attempt to bring more formal governance to an ecosystem with numerous intermediaries and complicated transaction paths. IAB Tech Lab says greater consistency in how existing standards are used could reduce risk and strengthen trust between buyers and sellers.

Programmatic Standard Practices v1 is open for public comment through Oct. 16. Go here for more information.

The post IAB Tech Lab proposes new rules for programmatic appeared first on MarTech.

More money went into AI infrastructure in the last three years than was spent building the interstate highway system over four decades.

When I asked a former CEO and friend what he thought CEOs were actually losing sleep over these days, no surprise, his list came down to five questions, all revolving around AI: profitability, core strategy, leadership AI literacy, build/buy/partner decisions, and measuring impact.

In finding answers to those five questions, I discovered a sixth question that lies beneath them all: How can AI make your company more valuable to your customers?

I guess I’m not alone. New research from Epsilon and Forrester shows that most companies are approaching AI integration the wrong way without realizing it.

Most CEOs are asking how AI can make them more valuable, rather than how AI can make them more valuable to their customers.

Making your company more valuable to customers actually solves the first problem. Epsilon’s newly published 2026 benchmark study, a survey of 257 marketing decision-makers conducted with Fuld Inc., supports this.

Seventy-one percent said their primary use of AI is productivity and efficiency, aka doing their own job faster. Only 9% said revenue generation.

Yet when the same marketers were asked how they actually measure AI’s performance, 46% pointed to revenue. Umm… seems like companies are using AI to improve their workflows, then grading it against a scorecard built for customer value.

It’s like buying a dishwasher and then measuring whether using it has driven down your property taxes. That mismatch is the real story of AI in business right now. It’s why marketing, not operations or finance, is where the AI disruption is truly landing.

Marketing owns AI’s impact on customer value

In recent decades, marketing ran on a simple mechanic: buy attention, awareness, interest, desire, and action, mostly at the bottom of the funnel using paid digital tools.

AI answer engines break that mechanic. You can’t buy your way into an AI answer the way you bought a search results page. You have to earn it through reputational (brand meaning) proof that an AI system deems worthy of citing.

Marketers already sense this: GEO (AI-powered search optimization) is now the most widely used AI tool in marketing, with 54% adoption, edging out conversational AI and data analysis.

Content-generation tools, the No. 2 AI tool just a year ago, didn’t even crack this year’s top 10. If you believe a company’s value is directly linked to creating appreciative customers, then you’re likely on the right path to AI integration.

Indeed, it’s the lens through which I answered the five questions below.

1. How can AI improve my profitability?

First, from an operational perspective. Delta CEO Ed Bastian has said AI could lift the airline’s profitability by roughly 50% over time, moving operating margin from about 10% to 15% through better pricing and scheduling. To Bastian’s credit, he’s talking about AI in service to a better customer experience.

Second, and mostly ignored: How do you keep AI from taking your profitability?

• Gartner projected traditional search traffic will drop 25% by 2026 as AI absorbs product queries.
• Adobe found AI referral traffic to U.S. retail sites grew 693% year-over-year during the 2025 holiday season, converting 31% better than non-AI traffic. (Big stats. Double-checked.)
• Forrester‘s research finds that firms with high AI use report meaningful productivity gains, but cost savings of under 10% and revenue gains of under 5%.

Proof that AI usage and AI-driven revenue remain myopically separated from the power of customer experience.

The big point: If your brand isn’t visible inside the AI answer, none of that productivity converts to revenue.

2. How can AI enhance my core business strategy?

Depends which strategy you already have. A.G. Lafley and Roger Martin’s Playing to Win framework draws a clear line. Are you playing to win, or playing not to lose? The former suggests you have a powerful strategy. The latter suggests you’re playing it safe.

In that case, AI can help you shrink less fast and help you play better defense. But that’s not strategic leadership. If you’re playing to win, the job isn’t finding a new strategy. It’s finding where AI bears the most fruit for the one you already believe in.

The big point: Treating AI as a strategy substitute rather than an amplifier of a winning strategy is just expensive automation theater.

3. How do I get my leadership team AI-literate?

Short answer: Make them use it every day. 

The longer answer: Find someone who understands the technology and can find where it works synergistically with your strategy. Be careful, since feeding proprietary data into public AI tools can make it effectively public.

Most AI literacy conversations miss one important fact: What looks like a training gap is actually a top-down perception gap. Epsilon found 67% of C-level marketers rate their organization extremely mature in AI, compared to 33% of senior managers actually doing the work.

Leadership calls their AI tools extremely valuable at 73%, compared with 25% among senior managers. Epsilon’s own recommendation for this exact gap: “get leadership to start looking at the same scoreboard” as their teams.

The big point: Confidence at the top is running well ahead of what practitioners see on the ground. Literacy programs built to fix a skills gap are ignoring this.

4. Should I build, buy, or partner for my AI usage?

This leans more toward marketing than most CEOs expect, because it’s a question of differentiation. Menlo Ventures found 76% of AI use cases are already being purchased rather than built, so most companies have already answered this by default.

Seems to me the move is to buy the common stuff, then build or partner only where it touches something proprietary.

Epsilon’s own researchers reach the same conclusion from a different angle: With 100% of marketers now using AI, “simply ‘using AI’ doesn’t earn you an advantage anymore.”

Besides, as with any tool, the differentiator isn’t the tool itself. It’s what it’s pointed at. Using AI to mass-produce your marketing content is the opposite of differentiation. It’s a blend-in strategy hiding in the clothing of efficiency.

If your content output reads like your competitors’ AI output, you’ve automated your way onto what readers of my articles here will know as the Plateau of Indifference.

The big point: Judge every build/buy/partner decision by one question: Does this make us more distinct, or just cheaper at looking like everyone else?

5. How do I measure AI’s impact on my business?

Most companies measure AI activity instead of AI impact. Epsilon’s data shows marketers already default to revenue as their top measurement lens (46%, ahead of time savings at 36% and cost savings at 16%), so the instinct is there.

The problem is what those measurements reflect: an AI use case chosen for internal productivity rather than customer value.

For marketing specifically, the outcome that truly matters is AI visibility, because AI is indifferent to your media spend. (Oooh, there’s that plateau of indifference again.)

The big point: Measuring AI impact on internal AI optimization misses a huge factor that’ll only get huge-er. More customers equals more impact. Wasn’t it always thus?

The big through-line

Every one of these five questions runs into the same wall: A company can use AI constantly, measure it diligently, and still be optimizing for “are we better at this” instead of “are we more valuable to the people paying us.”

The data says most companies are doing the former and grading themselves on the latter without noticing the gap.

Use AI to help you get more customers by creating better, more appreciated experiences for them. That makes them feel more valued. That’s what every CEO should focus on because that’s where AI’s value will truly compound.

The post Companies are measuring AI against the wrong goal appeared first on MarTech.

Salesforce hopes its new tailored-for-CRM AI will be a boon for marketers and a moat against the general-purpose models of companies like OpenAI, Anthropic, and Google.

At Dreamforce, the company introduced Koa, its first CRM reasoning model, developed with Nvidia to work through complex, multistep sales, service, and customer workflows. Instead of simply answering questions or generating content, Koa is designed to determine which tools and actions are needed to complete a job.

For marketers, that distinction matters as AI agents take on more operational work. Writing an email is relatively straightforward. Deciding whether a lead qualifies, checking its account history, applying company rules, updating the CRM, and triggering the appropriate sales or nurture workflow requires the model to understand how the business operates.

Koa is Salesforce’s attempt to put some of that operational knowledge directly into the model.

“The most valuable thing Salesforce has built isn’t our platform — it’s the accumulated knowledge of how enterprise business actually works,” Marc Benioff, Salesforce’s chair and CEO, said in a statement. “With Koa, the knowledge is put inside the model itself. We trained a reasoning engine that understands the structure of a deal, the lifecycle of a service case, and the workflows that vary across industries. That’s a different kind of intelligence.” 

Koa learns how CRM work gets done

Salesforce built Koa on Nvidia’s Nemotron 3 Super and post-trained it using a proprietary synthetic dataset modeled on the company’s decades of experience with CRM deployments. No customer data was used to train the model.

Its training scenarios cover more than 14 industries and recreate workflows such as generating leads, qualifying opportunities, and resolving service cases. Each scenario maps the actions and tool calls needed to complete a task, teaching Koa to work through the steps required to reach an outcome rather than simply produce an answer.

Salesforce also controls the model weights and performs inference within its own infrastructure, so customer data doesn’t cross its trust boundary when Koa is being used. That’s particularly relevant when an agent moves beyond generating marketing assets and starts accessing customer information, changing records, or triggering workflows.

The company says it is “moving into customer pilots” with 1-800Accountant, Baxter Credit Union, Engine, Formula 1, UChicago Medicine, and Xero. General availability in U.S. regions is expected this winter.

Marketers may need more than one AI model

Koa isn’t replacing the general-purpose models Salesforce customers already use. Salesforce is expanding those choices at the same time it develops its own specialized model.

Agentforce customers can use Gemini models through Salesforce’s Google Cloud partnership. An expanded AWS integration adds models available through Amazon Bedrock, including models from Anthropic, Nvidia, and OpenAI.

That points toward a different way of thinking about AI in the martech stack. Instead of choosing a single model to handle everything, companies could use different models for different tasks.

A general-purpose model might be a good choice for brainstorming a campaign, analyzing research, or drafting content. A specialized model such as Koa could handle jobs that require knowledge of CRM processes, company rules, customer records, and the sequence of actions needed to complete a workflow.

For marketing operations teams, model selection could become another orchestration decision. Cost, accuracy, speed, access to customer data, governance requirements, and the consequences of an error could determine which model gets a particular job.

AIforce gives those models somewhere to work

Koa also fills in another piece of a Salesforce strategy already underway.

Last month, the company’s Claudeforce partnership made Salesforce’s interface optional by putting its data, business logic, and actions inside Claude. AIforce, formally unveiled at Dreamforce, expands that approach across more AI interfaces.

AIforce makes Salesforce data, workflows, semantics, permissions, security, governance, and actions available via API so they can be used in other AI environments. 

The company’s expanded partnerships with AWS and Google Cloud further extend that architecture. Salesforce capabilities can surface inside Amazon Quick and Gemini Enterprise, while models and agents from those ecosystems can operate with Salesforce data and workflows.

For marketers, the interface where work happens and the technology doing the work are becoming separate choices. Salesforce can provide customer context and business rules, Claude or Gemini can provide general-purpose reasoning, and Koa can handle work where specialized CRM knowledge is more useful.

The same shift is reaching customers

Salesforce’s expanded Google partnership shows what this separation can look like on the customer side.

Starting this fall, Commerce Cloud merchants will be able to surface products in Google Search, including AI Mode and Gemini. Customers can complete purchases through Google’s Universal Commerce Protocol while payments, compliance, and order management remain on the merchant’s Commerce Cloud infrastructure.

The customer can interact with Google while Salesforce operates underneath the experience.

That’s a significant change for marketers accustomed to thinking about customer journeys in terms of websites, apps, ecommerce stores, and other brand-controlled destinations. As AI interfaces become another place where discovery and transactions happen, the technology determining what customers see and what happens next is largely invisible to them.

Similarly, marketers may do less directly with the applications in their stack as agents handle more of the navigation between them.

That puts more weight on what’s underneath the interface: accurate customer data, consistent definitions, permissions, business rules, APIs, and governance. It also places greater emphasis on choosing the right reasoning for the job.

In a world where AI makes information cheap and expertise easier to access, Salesforce is betting that experience is harder to copy. 

The post Salesforce bets CRM experience can give its AI an edge appeared first on MarTech.

As software begins to interpret objectives, make decisions, and act across systems, the question is whether the environment around the stack contains enough context, rules, permissions, and accountability for both people and machines to operate reliably.

For years, marketing leaders have been taught to think about martech as a collection of capabilities. CRM handles customer data. DAM manages assets. Workflow coordinates work. CMS publishes. Automation executes repeatable processes. Analytics measures what happened.

The strategy followed naturally: build the right stack, connect it, get people to use it, and keep improving the architecture. That model still matters. But it rests on an assumption we rarely talk about — that a person is sitting somewhere in the middle, making sense of everything the technology doesn’t know.

A person knows which of the five assets in the DAM is the approved one. They know the CRM record is technically correct, but six months out of date. They know Legal objected to that phrase last time, even though nobody ever updated the guidelines. They know one market needs another review before anything goes live.

The stack has always had a human operating layer holding it together. AI is starting to expose what happens when that layer is no longer guaranteed.

The stack was built for human operators

Even the great waves of marketing automation didn’t fundamentally change this arrangement. Someone still defined the rule, configured the workflow, set the variables, and decided where automation started and stopped. The system executed within the boundaries that people had established in advance.

Where the technology fell short, people filled in the blanks. Because humans are good at working around ambiguity, marketing organizations have tolerated weak metadata, half-designed processes, inconsistent governance, local exceptions, and information scattered across inboxes, spreadsheets, meetings, and institutional memory.

That tolerance shaped the stack we have today. It also explains why so many organizations can have well-integrated technology and still depend on a handful of experienced people who just know how things work.

For a human-operated environment, it can remain inconvenient rather than catastrophic. A person can spot the inconsistency, phone somebody, check an old email, or know that the documented process isn’t quite how the process actually works. Software has no such luxury.

AI has no tribal knowledge

As intelligent systems move from assisting marketers to selecting, deciding, routing, generating, or acting, much of that invisible human context needs to be available somewhere the technology can access it.

A system needs to know not only that information exists, but which version is authoritative. It needs to know whether an application can perform an action and whether it’s allowed to do so in a given situation. It needs to know whether an asset is available, approved, current, correctly licensed, and appropriate for a particular market or audience.

This is why discipline is becoming more important than ever. Metadata, provenance, permissions, workflow state, rights, and ownership are starting to matter differently. None of them is new. Enterprise architects have been talking about them for years. The difference is what happens when they’re weak.

A bad taxonomy is used to make the DAM annoying to search. Now it can cause software to select the wrong asset. An ambiguous approval state is used to trigger another Teams message. Now it may determine whether a system believes it has permission to publish.

The weakness hasn’t changed. The actor has. That creates a requirement martech strategy has rarely had to consider: machine operability.

Human usability asks whether a marketer can work with the technology. Machine operability asks whether the information, rules, and capabilities within that environment are explicit enough for another system to understand and act on them reliably.

Gartner’s 2026 CMO Spend Survey suggests investment is already moving faster than operating maturity. CMOs are allocating an average of 15.3% of their marketing budgets to AI initiatives, yet only 30% report having mature AI readiness capabilities. Seventy percent say their internal marketing processes aren’t mature enough to implement and scale AI effectively.

That gap isn’t simply an AI problem. It’s an operating-environment problem.

CreativeOps is where the cracks show first

Creative production makes this easy to see because generative AI has dramatically increased the amount of marketing that can be produced.

At task level, the economics look fantastic. A first draft takes minutes. Visual routes multiply. Localization happens almost instantly. Work that once required more people, agency hours, or production budget suddenly appears available at a fraction of the effort.

Then it reaches the rest of the process. A tenfold increase in generation doesn’t create 10 times more useful marketing. If briefing, rights management, review, approval, localization, and publishing still run as they did before, very little changes. You haven’t removed the bottleneck. You’ve only moved it.

McKinsey found something similar at the enterprise level. Of 25 organizational attributes examined, workflow redesign showed the strongest relationship with reported EBIT impact from generative AI. Yet only 21% of organizations using the technology said they’d fundamentally redesigned at least some workflows.

Making the task faster doesn’t make the system better.

The hard part starts after generation

Imagine a global campaign producing thousands of variants. Creating them may now be straightforward. The difficult questions arrive afterward.

Which product information is current? Which claims have been approved? Which imagery can be used in which territory? What parts of the brand system can change? Which markets require additional review? What happens when something falls outside the normal process?

A seasoned reviewer carries much of that context instinctively. A machine doesn’t.

CreativeOps is becoming a proving ground for the next martech operating model. Its role has been framed in terms of briefing, resources, workflow, review, approval, technology, and delivery. As more production becomes automated, the organization faces the question of how human judgment, deterministic automation, and intelligent systems work together.

Adobe’s Workfront Content Reviewer is an early example of where this is heading. It can participate in projects and approval workflows in a similar way to a user, assessing work and making recommendations before a human makes the final call.

The clever part isn’t that AI can review an asset, but rather everything that has to exist before the review means anything.

Brand rules have to be explicit enough to assess. Approval criteria need to be clear. Rights and context need to be available. On-brand needs to mean more than “John from brand will know it when he sees it.”

If 10 years of institutional knowledge are still required to decide whether something is safe to ship, another agent doesn’t solve the problem. The platforms are connected while the judgment is often not.

Connected isn’t the same as operable

A smart martech reader could reasonably ask whether this is simply composable architecture, APIs, and interoperability repackaged for the AI era. It isn’t.

Integration solved an important problem. It allowed information and capabilities to move between systems without everything having to live in a single giant platform. But access isn’t understanding.

An API can expose an asset perfectly and give a machine no idea whether that asset is approved. It can expose customer data without explaining whether a particular use is permitted. It can expose a workflow status without telling the system whether that status grants authority to act.

Integration makes the information available. Operability makes the environment understandable enough to use. That second part doesn’t arrive free with the connector.

Somebody still has to intervene. Someone still has to define the rules, clean the metadata, structure the rights, clarify ownership, and decide what each workflow state means. It isn’t the sort of work that gets the glamorous transformation slide.

Unfortunately, it’s the work that determines whether the glamorous transformation survives contact with the organization.

It’s time to stop building tomorrow’s roadmap from today’s shopping list

This changes where martech strategy should begin. Most roadmaps start with the estate already in place. 

• What’s underperforming? 
• What’s duplicated? 
• What can be consolidated? 
• Which platform needs replacing? 
• What are we missing?

Those are sensible questions, but they also make today’s stack the starting point for tomorrow’s operating model. That’s backward. Start instead with the operating capability marketing wants to build: What do we expect people and intelligent systems to be able to do together?

You have to work backward into the technology. If the ambition is automated localization, the requirement isn’t simply a better generative model. Marketing needs structured assets, reliable rights information, useful metadata, market rules, approval logic, and an authoritative place for the final content to live.

If the ambition is more autonomous campaign optimization, the interesting question isn’t whether software can move budget. Of course it can. The question is when it’s allowed to, what information it should use, where the exceptions sit, and who owns the call when it gets one wrong.

An AI use case quickly becomes a martech requirement. That’s why AI strategy and martech strategy are becoming difficult to separate. AI strategy asks what greater intelligence makes possible. Martech strategy determines whether the organization is capable of supporting it.

Yesterday’s repositories become tomorrow’s infrastructure

There is another, slightly counterintuitive consequence. Some of the less fashionable parts of the stack may become more important as human interaction with them declines.

DAM is a good example. A poorly governed DAM, full of duplicates, weak metadata, and questionable rights information, doesn’t become strategic just because someone plugs AI into it. It becomes a faster way to find and use the wrong asset.

A DAM containing authoritative assets, strong metadata, clear rights, and meaningful relationships between content, products, and markets is a different proposition. Intelligent systems can consume that context directly.

Fewer people may eventually need to log into the DAM itself. That doesn’t mean it’s become less important. It may mean it’s graduated from “somewhere people go” to “somewhere the operating environment depends on.”

It also requires a rethink of procurement. Functionality, UX, implementation, and cost remain important, but buyers also need to ask whether the data, context, and actions inside a platform can participate in a wider environment that the organization controls.

A vendor can have the best AI demo in the room and still leave you with a very expensive dead end if everything useful is trapped inside its own ecosystem.

The environment must become the foundation of the strategy

The stack isn’t disappearing, and neither are people. This also isn’t an argument for handing the keys to autonomous agents and hoping for the best. Different organizations will move at different speeds and delegate different levels of authority to software.

What matters is preserving the ability to make those choices. The next martech roadmap needs to describe more than which platforms marketing intends to buy, replace, consolidate, or connect. It needs to describe the operating capability marketing intends to build.

That means machine operability sitting alongside human usability. It means turning more of the rules, permissions, context, and accountability carried by experienced people into infrastructure that the wider environment can use. It means recognizing that the next source of martech advantage may have very little to do with who owns the longest feature list.

For 20 years, the martech stack has largely existed to give marketers better tools with which to run marketing. Its next job is bigger: to create an environment in which people, automation, and intelligent systems can run marketing together without relying on somebody in the corner who just knows how this stuff works.

The next martech strategy will be decided by whether you’ve built an environment where both people and machines can be trusted to work.

The post The next martech strategy starts with the operating environment appeared first on MarTech.