One cockpit for every coding agent, across every machine you own.
Claude, Codex, Gemini, and a browser agent — running on all your machines, commanded from a single board. Describe a goal and a CEO agent breaks it into tickets, dispatches the right worker to the right host, and each one ships a branch and a merge request. You bring the keys; you stay the operator.
From a sentence to a merge request in three moves.
No new workflow to learn. You stay the operator; the agents do the typing.
Describe
Type your intent into the Command Deck — "add rate limiting to the API and write tests." The CEO agent turns it into well-scoped tickets and assigns each a persona.
Dispatch
Approve the plan and workers spin up on your hosts. Each grabs a git worktree, does the work, and streams its terminal back to you live.
Review & merge
Workers push branches and open merge requests. You review the diff, discuss in a squad chat if needed, and merge — or send it back for another pass.
Everything you need to run a swarm, in one dense dashboard.
Built for someone who lives in it eight hours a day — calm, keyboard-first, information-rich.
Command Deck
Describe what you want in plain language. A CEO agent breaks it into tickets, picks the right persona for each, and dispatches workers to your machines — you approve before anything runs.
Ticket pipeline
Every unit of work flows Backlog → In progress → Review → Shipped. Watch it as a kanban board or a dense list, filtered by project, assignee, or status.
Agent fleet
See every worker across every host — studio-mbp, tower-01, rack-eu-1 — with live working / idle / offline state. Click any session to jump into its chat and terminal.
Personas
Reusable agent profiles: base CLI, model, system prompt, tool whitelist, and defaults like cost cap and idle timeout. Build a Backend Architect once, deploy it forever.
Cost tracking
Spend broken down by project and by provider, a 30-day graph, and per-day and per-month budget caps so a runaway loop never surprises you.
One-command hosts
Pair a new machine by running a single copyable curl command. The daemon phones home over an encrypted tunnel — no ports to open, no inbound firewall rules.
Your machines. Your keys. No lock-in.
Runs on machines you own
Agents execute on your desktops, laptops, and servers — not someone else’s cloud. The dashboard is just a remote control.
Your API keys and subscriptions
Bring your own Claude, OpenAI, and Gemini credentials. CallByte never proxies your model traffic or stores your keys.
Vendor-neutral — no model lock-in
Mix Claude, Codex, and Gemini on one board. When a better model ships, point a persona at a new CLI — your tickets, history, and workflow stay put.
Don’t guess which model is best. Race them.
Run the same prompt across all four CLIs in parallel and compare the results side by side — diff, rationale, time, and cost. Pick the winner and ship it. Or pull several agents into a squad room to debate a hard problem together.
useDeferredValue + key by id. Bench drops 312ms → 38ms p50 at 400 tickets.
Push virtualization into a Suspense boundary; drag handles stay outside the list.64 LoC.
Memoize the ticket→row map by id, move ordering into a reducer for atomic commits.
Traced it live: column re-renders 4× per update. Fix the map identity upstream.
Priced for operators — not per agent.
You bring the machines and the keys. We charge for the cockpit and the orchestration — never per agent you run.
Hacker
Run one agent at a time — free, on your own machine.
- 1 machine
- All 4 CLIs in one board
- Tickets + auto-run, 1 at a time
- Live terminal + squad chat
- Bring your own keys
Builder
Your personal swarm — many machines, agents in parallel.
- Unlimited machines
- 3 agents in parallel
- Bakeoff — race all 4 CLIs
- Cost caps + budget governance
- Telegram alerts
- Priority support
Team
Hand the wheel to the CEO orchestrator.
- Everything in Builder
- CEO orchestrator · 1 brief → many tickets
- Custom personas
- Unlimited parallel agents
- Shared fleet & org budgets
- Admin & audit log
You bring your own keys — CallByte never proxies your model traffic, so you only ever pay your own provider for tokens. Need SSO, on-prem, or an SLA? Talk to us about Enterprise.
Put your machines to work tonight.
Pair your first host with one command, hire an agent, and dispatch your first ticket in under five minutes.