What OpenAI Grove actually is (and what it isn't)
Grove is a five-week, in-person-heavy programme hosted at OpenAI's San Francisco HQ. Roughly fifteen participants are selected per cohort. The framing is deliberate: OpenAI calls the cohort “pre-idea” to signal that polished pitch decks are the wrong input. What they want is technical talent — engineers, researchers, tinkerers — who are close enough to frontier models to spot the next real company before anyone else does.
Participants get $50,000 in API credits, weekly office hours with OpenAI researchers, early access to unreleased tools and models, hands-on workshops, and — most valuably — a tight alumni network that keeps compounding after the cohort ends. There is no equity swap, no demo day, no forced pivot into a specific vertical. The programme explicitly welcomes healthcare, robotics, education and anything else founders want to try.
What Grove is not: it is not remote. Cohort 2 required in-person attendance during weeks one and five, with 4–6 hours of async work in between. It is not a grant programme — you are not paid. And it is not a soft-skills founder bootcamp — the content is heavily technical, which filters the applicant pool hard.
Inside Grove Cohort 2: January–February 2026
Cohort 2 ran from 22 January to 27 February 2026. Applications closed on 12 January. Fifteen builders were selected from a pool OpenAI has not publicly sized, but which applicants and community posts suggest ran to thousands.
The structure was tightened from Cohort 1 based on feedback: more in-person co-working time in week one, a mid-programme project review, and a final showcase where participants presented whatever they had actually shipped — not what they pitched. API-credit-wise, the $50K allocation held; participants also reported early access to frontier models ahead of public release, which is the programme's most-cited benefit on LinkedIn, per OpenAI Developers.
Cohort 1 alumni have already produced a handful of visible companies across developer tools, AI-assisted education and voice-agent infrastructure, several of which went on to close pre-seed rounds. OpenAI has not published a formal outcomes report, but the pattern — technical founder enters Grove, exits with a crisp problem statement and a working prototype, raises within six months — is tracking.
Cohort 3 has not been formally announced as of April 2026, but OpenAI's cadence suggests another window will open in late summer or autumn, following the same five-week structure.
Why Grove matters more in 2026 than it did at launch
When Grove launched in September 2025, it looked like a nice-to-have signalling exercise from OpenAI. Six months on, the context has shifted in three ways that make Grove's thesis sharper:
- AI startups are now capturing roughly 33% of global venture capital funding in 2026 — the highest concentration in any single category since the cloud wave. Technical founders with frontier-model fluency are the scarce resource venture capital is actively hunting.
- Solo-founded startups are now 36.3% of all new ventures, up from 23.7% in 2019. The “one-person unicorn” narrative has moved from cope post on Medium to boardroom assumption, and AI agents are the reason. Grove is explicitly solo-founder-friendly — small teams and individuals both qualify.
- The pre-idea gap is real. Most accelerators demand a formed thesis, a deck, a co-founder. Grove is one of the only well-resourced programmes that will take a single senior engineer who quit their job on Monday and wants to spend the next five weeks figuring out what to build. That gap existed in 2023. It is now commercially important, because the models move fast enough that the right idea in February is a shipped product by May.
Your tactical playbook for the next Grove application window
If you want to be in Cohort 3, the next 90-odd days matter more than the application itself.
- Ship something small, public, and technical — now. Grove reviewers read GitHub, X, Hugging Face and Replicate. A working demo of a weird agent, a custom eval harness, a novel fine-tune — any of these beats a polished application by a wide margin. Cohort 2 applicants who got in almost all had a public technical artefact within the last 60 days.
- Write one serious blog post about a frontier-model limitation you have actually hit. Not a trend post. A specific, reproducible failure mode and what you tried. This signals the kind of thinking Grove is designed to accelerate.
- Pre-idea does not mean unstructured. Applicants who framed their curiosity as a tight problem space — “I want to understand why current agent loops fail on long-horizon tasks in biology workflows” — outperformed those who wrote “I'm excited about AI and want to build something big.”
- Be realistic about the in-person requirement. If you cannot relocate to San Francisco for weeks one and five, Grove is not for you. OpenAI has not softened this, and community reports suggest they decline otherwise-strong candidates who flag scheduling issues.
- Start the alumni conversation early. Cohort 1 and 2 participants are publicly identifiable on LinkedIn. One coffee with an alum will tell you more about the application filter than any blog post, including this one.
What this means for marketers and early-stage founders
Here is the part nobody in the Grove-coverage ecosystem says out loud.
Grove solves the technical founder problem. It gives you frontier-model access, a credible network, and five weeks of compressed learning. What it does not solve — at all — is the distribution problem. The number of pre-seed AI startups in 2026 that die with good products and no users has never been higher. The bottleneck for a Grove alum in month six is almost never “can I build it.” It is “can anyone find it.”
The old answer was “hire a marketing team.” The 2026 answer cannot be, because the economics of a pre-seed AI startup do not support a three-person marketing hire, and the work a traditional marketing team does — brief writing, brand assets, SEO content, distribution, measurement — is now the kind of work AI agents are objectively good at. Anjin was built for exactly this gap — a Marketing Operating System that replaces the team the pre-seed economics will not fund.
This is the operational gap where Grove ends and something else has to start.
Anjin: the Marketing Operating System for pre-seed AI founders
Anjin is the Marketing Operating System — a single platform of AI agents that does the work a five-person marketing team used to do. Strategy, brand assets, SEO content, landing pages, email sequences, social, measurement. For a Grove alum sitting on $50K of API credits, a working prototype and no distribution, Anjin is the layer that turns week-six into month-six-with-real-users.
The thesis is symmetrical to Grove's. Grove says: pre-idea technical founders deserve frontier access and a tight network. Anjin says: pre-seed founders deserve a marketing stack that matches the pace at which they can now ship. You cannot raise a seed round in 2026 on a product nobody has seen, and you cannot afford a marketing team until after the seed round. The Marketing OS category exists to close that specific loop.
Agencies used to be the default answer. At £8k–£15k per month, for a pre-seed AI founder, they are not. A single lifetime licence for a Marketing OS — priced once, used forever — is.
The £888 Lifetime License — Offer Closing Soon
Lifetime access to Anjin for a one-time payment of £888. Not a subscription. Not a seat. Not a trial. One payment, unlimited use, for as long as Anjin exists.
The average marketing team spends £888 in about three working days on tooling, freelancers and coordination software. You're buying the platform that replaces most of it — once.
This price will not be offered again once we close our early-access cohort.
Claim your £888 Anjin lifetime license →Founders, agency owners and in-house marketers — this is how you run marketing at AI speed without the team, the burn, or another year of waiting.
Sources: OpenAI Grove, CNBC, OpenAI Developers on X, ShiningPens, StartupWired, EntrepreneurLoop, AI CERTs News, NxCode, Presta
