AI Go-to-Market Playbooks Shift Toward Precision, Curiosity Hires

The panel that set the tone

At TechCrunch Disrupt in October 2025, three people with an unusually clear view of how AI startups sell sat down to talk about whether the old go-to-market rulebook still applies. They were Marc Manara, head of startups at OpenAI; Alison Wagonfeld, vice president of marketing at Google Cloud; and Max Altschuler, general partner at GTMfund.13 Nearly a year later, what they said reads less like a forecast and more like a description of how the market actually developed. The short version is that AI changed how companies find and qualify buyers much more than it changed what selling requires.

Altschuler started from the familiar line that companies can now do more with less.1 He then complicated it right away. Some startups have been hiring engineers fluent in AI and pointing them at classic go-to-market problems, but he argued that domain expertise still matters and that experienced advisors can pass on playbooks that remain useful.13 In his telling, founders still need a working grasp of why particular marketing tactics succeed.3

OpenAI's argument: focus, not just savings

Manara offered the most specific claim on the panel. He agreed that startups can work with smaller teams, but he said cost-cutting isn't the main reason they're adopting AI for go-to-market.1 What stands out to him is how far AI lets a company personalize outreach and act on buying signals, a level of precision he called genuinely differentiated.15

His examples were concrete. Lead generation used to mean running a simple database query. Now a startup can write AI prompts that find prospects matching a narrow, multi-part profile, and then use similar prompts to score inbound leads far more precisely than before.34 Coverage of the panel agreed on this point. Every account of his remarks treated personalization and signal-tracking as the core of OpenAI's view, not headcount reduction.45

The idea is easier to judge alongside how OpenAI built its own sales function. One third-party analysis says the company skipped a traditional sales development layer. In that account, each account executive finds and researches their own prospects, using enrichment workflows that pull in news, hiring data and firmographics, with GPT models summarizing the results.26 The same analysis describes OpenAI's revenue operations team as an engineering group. It says the team built internal assistants that answer inbound sales email, pull out contract terms and summarize calls inside Slack.26 Those figures come from an outside writer, not from OpenAI. Still, the picture matches what Manara described: AI supports the person doing the selling and doesn't replace them.

Max Greenbury, who works on OpenAI's startup sales team, made a similar point on a podcast this summer. He said tools like Codex take care of data pulls and CRM updates before a salesperson gets involved. He also predicted that AI will absorb the administrative side of selling within five years but won't replace the trust buyers place in people.24

Google's argument: the craft still matters

Wagonfeld put less weight on tooling and more on fundamentals. She said companies need AI knowledge, curiosity and technologists, but they also need people who understand what marketing is for: gathering customer insight, doing research and recognizing strong creative work.3 Where AI helps most, in her view, is speed. Teams can put out many more messages much faster and then step back to decide which metrics actually matter.3

Her clearest point was about hiring. Teams used to hire specialists, sometimes for sub-specialties within sales or marketing. Now, she said, curiosity and understanding have become close to the most important trait to hire for.13 Some write-ups described curiosity and adaptability as something both OpenAI and Google agreed on.5 The direct quotes, though, tie the hiring argument to Wagonfeld specifically.13

Where the coverage disagrees

The reports on the panel mostly agree. The disagreement shows up in how later commentary framed it. One long piece built on the panel claimed that OpenAI and Google hold sharply different go-to-market visions. It cast OpenAI as a developer-first, conversation-led company and Google as a vertically integrated ecosystem.2 That piece also added claims about autonomous sales agents and specific team-size ratios that don't appear in the panel reporting.225 The firsthand accounts describe something more modest. There was broad agreement that AI speeds up execution and sharpens targeting, and the speakers differed mainly in emphasis: Manara stressed precision, Wagonfeld stressed craft and Altschuler stressed advisors and domain knowledge.134

The divide between the two companies is real, but it's about corporate strategy, not marketing advice. Analysts last year described Google's Gemini Enterprise as an "enterprise-out" approach. It builds AI into Workspace and existing corporate workflows, with Workspace AI pricing cut by bundling.6 OpenAI, by contrast, was described as pursuing a "consumer-in" platform strategy through its Apps SDK and AgentKit, with plans for an app directory and agentic commerce.6 Those are two different business models. The advice to startups from both sides of the panel was close to identical.

The irony in the efficiency story

The data shows a tension in the "do more with less" message. OpenAI tells startups that small, AI-assisted teams can go far, yet its own go-to-market organization grew very quickly. Its first sales leader, Aliisa Rosenthal, ended up running a go-to-market organization of more than 300 people before she left for venture capital.22 One outside analysis puts the figure above 500,26 and an org-chart tracker lists about 600 people under go-to-market.29 In early 2026, OpenAI also put Barret Zoph in charge of its enterprise sales push, and CFO Sarah Friar named enterprise growth as a priority for the year.23

This looks like a difference in what's being sold, not hypocrisy. Greenbury has described enterprise and startup sales at OpenAI as nearly separate jobs. Enterprise deals mean working through many stakeholders and surfacing pain the buyer hasn't named yet. Startup accounts are more about tuning for latency and cost.24 AI-driven efficiency seems to help most in startup-style sales. Large enterprise contracts still need people.

The new channel: answer engines

The biggest change since the panel may be a new place where buyers find products. Greenbury has said that generative engine optimization, meaning getting a product named when someone asks ChatGPT for a recommendation, is becoming a real go-to-market priority.24 Google is pushing in the same direction. eMarketer reported in July that AI Mode reaches 1 billion monthly users and that Google's search redesign is pushing more searches toward zero clicks.10 Analyst Nate Elliott called predictions of the open web's death overstated, noting that more than a quarter of searches still lead to clicks on outside sites.10 Business Insider has reported that the search funnel, the time from first research to purchase, is getting shorter as AI gets people to answers faster.7

Founders should take this seriously. If an AI system is the first to describe your category to a buyer, then positioning, clear product language and solid evidence become distribution tools. Startup commentators are already telling founders to make their offers easy for machines to read and to check how they appear in AI answer surfaces.12 That fits with Manara's focus on signals and Wagonfeld's focus on craft. Precision matters because buyers arrive with more context, and craft matters because generic copy disappears.

What founders should take from it

The panel's central claim has held up: AI's main effect on go-to-market is better targeting, not cheaper headcount. GTMfund, Altschuler's firm, has since put it more directly. Its argument is that distribution is the last moat in AI, because building software keeps getting easier while reaching customers doesn't.19 GTMfund partner Paul Irving has said founders shouldn't try every channel at once and should get creative about reaching customers directly.19

Both OpenAI and Google seem to be backing that approach for their startup customers. Google's 2026 India accelerator, for example, pairs access to its AI stack with go-to-market mentorship for 20 AI startups chosen from about 2,500 applicants.1113 Rosenthal, now at Acrew Capital, says she plans to help founders with pricing, AI-native sales and the commercial decisions that determine what scales. She also argues there's still plenty of room for startups alongside the large labs.22

A year on, the practical takeaways from the panel are fairly clear. Use AI to find fewer, better-fit prospects. Keep people who understand why marketing works. Hire for curiosity rather than narrow specialization. And treat AI-generated answers as a sales channel in their own right.