For independent consultants and freelancers · business
Offer governed agent swarms to your clients, without building the plumbing.
You already do AI work. Your clients ask for more than a chat window: a team of agents that does a real piece of work, a record of what each agent did, and proof that nothing leaked. This is that, ready to put your name on.
Built on rUv's tools: Ruflo (MIT, by rUv / ruvnet), an agent meta-harness for Claude Code and Codex with a queen-led hive-mind and a policy engine that runs in legacy, observe or enforce mode and writes a hash-chained receipt for every decision. We build on it; we do not sell or own it. Ours: sgit vaults and Agent Behaviour Policies (Apache-2.0).
Who this is for
- Consultants who already sell AI workand want a swarm offering with evidence behind it.
- Freelancers who build for clientsand want a hive of agents to do a scoped piece of work in parallel.
- People who never had the time or the setupto run a Ruflo swarm themselves, and want to start with one run, done with someone who has.
The problem, in your words
"My client wants agents doing the work. They also want to know what the agents were allowed to do, what they actually did, and where the data went. I cannot spend three weeks wiring that up for every engagement."
Three ways to work with this
Copy everything
It is open. This site, the runner and the behaviour policies are Apache-2.0; rUv's tools are MIT. Clone the repository, follow the setup brief, and run it for your clients under your own name. No permission needed: anybody who implements this implements sgit vaults and Agent Behaviour Policies, which is what we want.
Your brand, our runs
You own the client and the conversation. We run the hive behind you and hand you the run's vault to deliver as yours.
An offering built with you
We shape the roles, the behaviour policy and the deliverables around your practice, then run the first engagements together.
How a typical engagement runs
Scope
You and your client write the brief: the objective, the materials (provided up front, no internet during the run), and one role per piece of work.
Mandate
Each role gets a written mandate (an Agent Behaviour Policy): what it can reach, what it may do, and what is stopped. It compiles into Ruflo's policy engine in
enforcemode.Run
A queen and workers build in parallel, each on its own branch of one encrypted vault. A person on our side starts the run and watches it.
Review
Expert time where it helps: a couple of hours to check a delivery, or days alongside you on a larger engagement.
Deliver
The run closes, is measured, and the vault goes to your client.
Not automated end to end yet. A person on our side starts each run and watches it to the end. The runs on this site used a local vault server on the machine that ran the hive, so their timings are local numbers; the workers were Claude Code subagents, with Ruflo holding the queen, the roles, the tasks and the broadcasts. Dedicated compute and the materials sandbox for customer uploads are not built yet, and SG Compute is not connected to runs: when we reply to an order we say what we can do today.
What your client receives
The run's vault: one place that holds the whole run.
- Every brief, and the feedback that started the run.
- Every message between the agents and every delivery, as commits on each worker's branch, merged.
- Every decision, with the receipt Ruflo's policy engine wrote for it.
- The run's ledger entry: commits, files, bytes, what the server held, and whether a fresh clone matched what was delivered.
Delivered as a zip, or published as a vault they can clone. Publishing shares the vault's read key for good, so it happens only when your client says so.
Why it is credible
Every run that made this site's materials is in the ledger, with its numbers:
One trace of 11 actions through Ruflo's policy engine: all allowed in its default legacy mode; in enforce mode with rules from the behaviour policy, 6 stopped (4 denied, 2 sent to a person), and both receipt ledgers valid. POC1 →
Offers that fit
Prices are in the shop. Mix and match: put what fits in the cart and send it; there is no payment step, the order comes to us by email and we reply.
Where this sits
Task runs: what you sell on top of, from the commodity (compute, models) to what your client sees. Read the map with the business model.
Talk to us
A live agent answers; a person follows up. Every contact and every order from the shop goes to this address.