The three biggest go-to-market keynotes in September agreed on two things: most GTM teams use AI, but few can say what it accomplished. #HubSpotUNBOUND shared a survey of 6,000 companies and found 90% are using AI yet only 6% seeing transformational results. Gong's #Celebrate26 conference shared only about a third of the companies buying AI can connect it to outcomes. At #Dreamforce2026 Anthropic's CEO Dario Amodei said "we're making use of maybe only 5% or 10% of what the possible value of the technology is." Understanding cost versus performance is the next big wave of AI investment, and it comes down to deploying AI with the right context engine for your GTM teams. And it all comes down to companies deploying it properly, with the right context engine for their GTM teams and AI. How are context, GTM, and AI ROI related? Many of us heard something like this back in the day that customer facing employees need: “The right context, for the right conversation, at the right time, in the right format.” The goal was always to drive more consistency across teams focused on revenue. This is still true, but the risks are higher and speeds are faster. All three companies agree that a GTM strategy with AI only succeeds when it knows your business and customers. When it doesn't it's a speedy and expensive, confident yet forgetful intern. HubSpot measured it and landed "AI, with bad context, is worse than no AI." The logic is bad context causes people stop to trusting it, then adoption goes with them, which erases any value it promises. Dreamforce Models alone cannot run the enterprise. Models alone are not going to show us what's possible. Marc Benioff CEO, Salesforce HubSpot UNBOUND We're all spending money on tokens, but the real question we have to ask is: Are we achieving something truly great? Yamini Rangan, CEO, HubSpot Gong Celebrate People bragging about how much they spend on tokens then CFOs crying about how much they spend. Amit Bendov, CEO, Gong Salesforce shared about how your business runs, your records, rules and permissions. HubSpot expanded a bit to talk about a context profile in its CRM. Gong means your engagement in calls and meetings. Of course this data is incredibly important. The challenge with a CRM-centric view is that its highest value is for sales teams. Your product, marketing and success teams' context about your business, products, competitors and brand lives elsewhere, and it's what you use to engage customers and run a great meeting. At the speed things are changing, sales can't work in a CRM context bubble. How we define GTM context At Loop, we break down content by what AI needs, and what your company versus GTM teams should provide to AI. Your company provides the foundation AI can't guess: your directives on voice, personas, guardrails and formatting, the structured data in your CRM and integrations to other systems, and who can see what, down to the file and field. The frameworks for how teams share templates, agents, and skills. Your product, marketing, sales and success teams provide most of the value, and it's where CRM stops: the skills, methodology and goals that keep everyone aligned, the research, positioning, launch plans and templates that win deals, what's happening in the market, and what's true right now, spread across drives, messages, calls and people's heads. AI context and memory fills in the rest as you work if set up properly, picking up the account or meeting you're in, remembering your decisions, learning which messages won, and connecting it all so it knows this customer relates to that deal and this research shaped that launch. That's the difference between AI that knows your CRM and AI that knows your business. Get the news and events that matter for GTM and AI, with a calendar of what's coming. Get Loop News What this means for your GTM team Teams who take context seriously and roll it out as a companywide initiative will see far more success than ones that give teams a set of licenses to pick from. What you can do before the next event: Pick four or five use cases that are relevant to the work you have right now that can drive real revenue or improvement. Write down the context each one needs, where it lives, and determine what home it should have going forward for humans and AI. Upgrade your company and team context, and if you can't migrate content/context now, just give it a temporary home for testing. Measure one team before and after, or split a team and A/B test the impact on productivity and revenue. Use this to drive a business case for the larger project on ensuring you're not fighting the context gap with AI any longer. We built Loop to solve this problem, and can help walk your teams through this process for free, and we'll even build one of these use cases with you using our own context engine and AI content architect. Conference Recaps Dreamforce and the case for structured data Enterprise context is anything that makes your business uniquely your business. Rohan Kumar, president of platform and engineering, made the argument for AI grounding via your systems of record. Salesforce's answer is AIforce, which brings its data, business rules and permissions into the tools people already use, starting with Claude, Slack and its Coworker assistant inside Salesforce. Patrick Stokes, president of applications and marketing, explained the bet: "our product is the trust" customers place in Salesforce to hold their data, workflows, permissions and security rules. I think that's right, and it's a big shift. The screen is becoming optional, and the record and the rules are the product. Salesforce has been on an acquisition spree, announcing "starting today, Fin, now part of Agentforce", and shared the Qualified acquisition is behind the Piper SDR agent. Salesforce says its outbound agent generated $500 million in pipeline last quarter. Adecco's CEO said his recruiters saved 35 to 40% of their time and placed 20,000 more people this year, with one condition: "AI had to happen with people and not to people." Rohan Kumar's platform section showed the other side of ROI, cost. Agent Fabric lists a company's agents from Salesforce, Microsoft, AWS and Google in one place and sets budgets for what they spend. During Dreamforce I went to a Claude for Sales event that dove deeper into use cases for CRM & Events with one of Anthropic's sales leaders sharing their own automation inside of Claude via Salesforce data, with powerful new ways to build Apps inside of Claude that are driven by customer and forecast data. Salesforce has the strongest answer for the structured half of context, the records, rules and permissions for larger organizations, especially with Slack filling in collaboration gaps in other CRM systems lack. HubSpot UNBOUND and turning context into action HubSpot UNBOUND The magic is not within the models. It is actually with the context of your business. Yamini Rangan, CEO, HubSpot HubSpot named the problem most GTM teams are living in. Yamini Rangan called it the "maxing phase": more tools, more agents, more tokens, and people working harder than before. A marketing leader she knows told her, "my work has actually tripled with AI." The chart that matters came next. HubSpot plotted fifty AI use cases by popularity against impact. The popular ones, like writing content, sending emails and meeting prep, sat low on impact, because without context anyone's AI produces the same output. The high-impact ones were analyzing campaigns, prioritizing pipeline and analyzing customer feedback, and each depends on what your company knows. Her line: "It is your data. It is your context. And, therefore, it is your advantage." HubSpot's definition of context was the broadest of the fall and the closest to ours: "the dynamic knowledge about your business, your customers and your team." Business is brand, voice, product and positioning. Customer is conversations, the ideal customer profile and buying signals. Team is roles, goals, approvals and methodology. Duncan Lennox, chief product and technology officer, showed how HubSpot plans to supply it: "A CRM only knows what someone remembers to tell it," so the new CRM updates itself from calls, emails and meetings, and a Context Home scores what's missing. The customer panel had the adoption lesson. A dental-software company's chief sales and marketing officer took her meeting booking rate from 11% to 43% in a year, after an early stretch where the prospecting agent targeted the wrong companies and her reps reviewed every email before it went out. HubSpot's innovation with their Smart CRM is strong around captures calls, emails and meetings automatically and scoring context completeness. Gong Celebrate: adoption is a people problem first If you don't fill the models with the right context, they'll confidently say the wrong thing. Amit Bendov, CEO, Gong Gong's stage was a mix of customers touting their achievements and lots of announcements, which I haven't figured out roadmap for yet. Cisco rolled out 18,000 seats in a year and gets 65% weekly use. Its head of AI seller experience was blunt that "skepticism is incredibly real." Mandates didn't move people. What worked was a champions program and the moment people "see the busy work evaporate from their day." It's the same lesson sales enablement has been teaching since I worked at an enablement vendor in 2007. The use I liked most came from Anthropic's own sales team. Travis Bryant's team reviews calls together in a "film room" every two weeks, with Claude suggesting which calls to watch, so reps teach each other. When Claude launched on Azure, his team searched past calls for prospects who'd said they wanted Claude on Azure and turned them into a campaign. His point: "every one of those conversations… is a data asset that isn't just for go to market but it's for the whole company." Gong launched Deep mode, which answers big questions like "why are we winning?" by building the right context for each question, instead of what Amit Bendov called "the transcript dumpster." It also previewed Activate, "an AI brain built from your wins." A few other September conferences Braze Forge. Marketers can run Braze from inside Claude, ChatGPT or Copilot (in beta), and an AI reviews campaigns for brand, link and compliance problems before they go out. It's another vendor coming to where people already work. WorkOS Init. A conference for builders, with tons of good content on what work looks like with AI, and most people becoming a manager of agents, how we securely let AI take action on our behalf, with or without a human in the loop. OpenAI DevDay. "Dots," always-on agents across 4,000+ apps, and ChatGPT Space, a shared workspace for teams and their agents. While everyone else plugs into the AI, OpenAI is building the team workspace itself, which tells you where collaboration is heading. LangChain Interrupt NYC. The engineers' version of the same story. Context Hub stores an agent's instructions, skills and examples as versioned files a team reviews together, like code. Context is becoming a team asset you govern, not a prompt you paste. Forrester B2B Forum EMEA. Forrester pointed AI at the buyer instead of the seller, starting from the finding that about nine in ten business buyers now use generative AI or conversational search when buying. Your buyer's AI reads you before your rep does, so what you publish has to be right.