Two founders. One operating problem.
CommunityOS started as the answer to a question we kept being asked by Web3 projects and brand teams: "we have a follower list — how do we tell who actually matters?" Every answer in the market was vague, vanity-driven, or paid-influencer-shaped. We built the answer we wished existed.
Engineer plus operator.
The split is deliberate. The engine is built by an engineer who treats it as a real piece of software, not a feature. The product is shaped by an operator who has actually run the campaigns the tool is built for.
Built the scoring engine, the contract schemas, and the data infrastructure. Treats community as a software problem with auditable inputs and outputs, not as a marketing concept. Reviews every architectural decision.
Runs the product, the brand, the go-to-market, and the operator surfaces. Spent years on the receiving end of the dashboards CommunityOS is built to replace. Reviews every customer-facing surface.
What we kept seeing.
Every Web3 project we talked to in 2025 had the same shape of problem. They had spent real money on KOL deals, quest programs, and listening dashboards. None of them could answer the simplest question: which of the people in our follower list are actually the community, and which ones do we activate to ship our next launch.
Agencies had a parallel version of the same problem. They were running community for clients and reporting on it, but the reports they shipped were screenshots, vanity counts, and decks built the night before a QBR. The reporting layer was the weakest part of the engagement, and it was getting renewed on relationship, not on numbers.
The combined miss was clear. There was an operator-shaped product missing in the middle: rank the people you already have, activate the ones worth activating, verify what was done, and ship a defensible report. CommunityOS is that product.
How we decide what ships.
Deterministic over impressive.
The same inputs always produce the same outputs. Methodology you can audit, defend, and reproduce. We pass on features that improve the demo and weaken the audit.
Operator in the loop, always.
Nothing fires automatically. Every action is initiated by a person. The tool is intelligence and surface, not autopilot.
Proof, not promise.
Reports are built on the verified layer. Verified is the only state that counts. Claims without artifacts are not reportable.
One product, not a stack.
Intelligence, activation, and proof live in the same surface. The operator never leaves the screen that produced the signal to act on it.
Talk to the founders.
The pilot conversation happens directly with us. Leave a work email and we will reach out.