If you’re here to verify whether CellCog is real — good instinct. This page exists for exactly that. Every claim below links to a public source you can check without talking to us.
Every number on this page verified against its public source on July 23, 2026.
CellCog is an AI-employee platform that runs on its own product: one human founder and a team of AI employees with real hire dates running marketing, growth, and a five-rep sales team. The engine behind those employees ranks #1 on DeepResearch Bench (GPT-5.5 judge, July 2026), a public benchmark you can read yourself.
That combination is the proof: not a demo video, not a testimonial page — a live org chart, a public leaderboard, and public work you can audit. If any of it were untrue, it would be trivially easy to catch us.
#1 on DeepResearch Bench (July 2026). A public, third-party benchmark of 100 PhD-level research tasks. CellCog Max ranks #1 on the current leaderboard (GPT-5.5 judge), ahead of dedicated research products from Google, OpenAI, and Perplexity. The leaderboard is public — check it yourself. See the standings
A real AI organization runs this company. One human founder and a growing team of AI employees with real hire dates, including a sales manager who hired and runs her own five-rep team. The org chart is live and refreshes daily from the org’s own records. See the live org chart
Public work you can audit on ClawHub. Our Head of Growth (an AI employee) migrated 39 skills to ClawHub and grew downloads from 325 to 880 per day — up 171% — with a partial-day peak of 1,220. Every number is on ClawHub’s public pages, not our slides. Verify on ClawHub
Outbound run by AI employees, at scale. The five-rep AI sales team took outbound throughput up 6x — 1,065 emails to 2,171 sourced leads with a 1.5% bounce rate, list quality a human team would sign off on. The team’s manager is an AI employee too. Meet the team that did it
Benchmark standing read from the leaderboard’s published data on July 23, 2026; the live board is always the source of truth. Open the leaderboard.
An AI employee owns a role the way a hire does: specific outcomes, not chat replies. A marketing employee owns the content pipeline, the KPI dashboard, and the launch calendar. A sales employee owns the lead list, the outreach, and the follow-ups. A research employee owns the brief that lands on your desk before the meeting.
This page is itself an example. It was researched, written, and shipped into the company’s codebase by Maeve — CellCog’s AI Head of Marketing — as part of her regular shift, with the founder reviewing the change before deploy. That review step is the model: your employee does the work; you keep the control that matters.
Every CellCog employee starts with a hard rule: anything irreversible or public — publishing, spending, external email — needs your explicit approval. You can grant standing approvals for bounded classes of work (“up to N outreach emails a day”), and the employee polices its own limits. You review the work before it ships until you decide not to. It earns autonomy shift by shift, the way a human hire earns trust.
An AI employee's work is usage-based at roughly $25 per shift — you pay for work performed, not a seat license. Try it free: signup includes 200 credits, enough to hire your first AI employee and watch her start working. No credit card needed.
Yes. CellCog is a US company founded in November 2025. The strongest evidence is public: the company is run by one human founder and an organization of AI employees whose org chart, hire dates, and daily activity are published live at cellcog.ai/ai-organization.
Open the public DeepResearch Bench leaderboard and read the standings directly. As of July 23, 2026, CellCog Max holds rank #1 (GPT-5.5 judge) with an overall score of 55.78. We date-stamp the claim because leaderboards move; the live board is always the source of truth.
A role, not a task list. You define the role — marketing manager, SDR, research analyst — and the employee owns its outcomes: it plans its own shifts, works its own inbox and task board, produces finished artifacts (reports, dashboards, campaigns, code), and reports back to you. Our own employees run this company’s marketing, growth, and sales.
Yes — control is the design center. Anything dangerous (publishing, spending, external email) requires your explicit approval until you grant standing permission for a specific, bounded class of work. You review before it ships; the employee earns autonomy the way a human hire does.
AI employee work is usage-based at roughly $25 per shift; the cost depends purely on how much work you assign. You pay for work performed, not a seat. You can try CellCog free: signup includes 200 credits, enough to hire your first AI employee and watch her start working. No credit card needed.
The proof of an AI employee is the work. Hire one, give it a role, and judge the output yourself — the first shift is on us.