Orchestrator and Petrova.
Orchestrator runs your coding agents. Petrova is the account those agents can call. They do different things, so you could use both.
How Orchestrator works
A desktop app that runs many coding agents at once (orchestrator.inc).
- You install a free desktop app. It runs a small program on your own computer.
- You sign in to the coding agents you already use. It supports 38, such as Claude Code, Codex and Cursor.
- You tell a main agent what to build. It splits the work among worker agents, each with its own copy of the code.
- It watches your pull requests and sends failed tests and review comments back to the agent that wrote the code.
- A board shows what needs you. Paid plans run workers on cloud machines. It does not pass along your AI calls, so each AI company bills you directly.
How Petrova works
The account your agent calls. Today this all runs on test data.
- Your agent keeps running where it runs today. Petrova does not run it.
- You change two settings: the base URL and the key. Anthropic and OpenAI formats both work.
- Before each call, Petrova holds the most that call could cost.
- A supplier answers. If one fails before the first word, the call moves to another.
- You are charged what the call used, plus a 5.5% fee. The rest comes back. Never above the hold.
How they fit together
- Orchestrator starts a worker agent, such as Claude Code or Codex.
- That worker can be pointed at Petrova with a base URL and a key, as any Claude Code or Codex can.
- Each worker's calls are held, capped and receipted per agent and session, while Orchestrator keeps doing the planning and the reviews.
No partnership exists. We have not tested the two together, and Petrova is not live.
| Orchestrator | Petrova | |
|---|---|---|
| Live today | Yes | No. Preview on test data |
| Runs coding agents | Yes, many at once | No. Petrova never runs agents |
| Reviews pull requests and sends fixes back | Yes | No |
| Phone app | Yes | No |
| Open source | Yes, Apache-2.0 | No |
| One balance across AI companies | No. Each one bills you | Yes (test data) |
| Spend caps on model calls | Not found | Account, project, key, agent, session (test data) |
| Charge never above a hold set before the call | Not found | Yes (test data) |
| Failover between suppliers | Not found | Yes (test data) |
Green marks the stronger answer in a row, red the weaker. "Test data" means Petrova's code does it in tests, not as a service. "Not found" means we did not find it on their public pages on the check date. It may still exist.
Where Orchestrator is ahead
- Live today
- Runs coding agents
- Reviews pull requests and sends fixes back
- Phone app
- Open source
What we could not check
- Its Cloud in real use.
- Its cost view, which we only saw on a waitlist page.
Read on 27 October 2026, on their public pages: orchestrator.inc · orchestrator.inc/pricing · orchestrator.inc/llms.txt · docs.orchestrator.inc. Things may have changed since. Each claim of theirs is their own statement.