Humans and AI agents.One company.
SingulaComp is a multiplayer, multi-agent platform. Your team and your AI agents share the same projects, skills, tools and memory. The agents do the work on real cloud computers, and your people decide what lands.
Free to start · No credit card required
Your people and your agents, on one org chart.
In SingulaComp the organization is your company. People join it with a role. Agents join it too: each one is a markdown persona with its own role, its own skills and a scoped reach into your tools. One permission model covers both, and the audit log names who did every action — a person, or an agent and the person it acted for.
People make the calls. Agents do the work.
You and your agents, before the first hire. A whole company without headcount.
Invite people as you need them. Roles cover people and agents alike, and only people take a seat.
- Founder
- Operations
- Design
- SSales
- EEngineering
- MMarketing
- FFinance
- PPeople
- DData
Work side by side in shared projects.
A project is one line of work, and its sessions share one cloud computer. Everyone works in their own session, so nobody waits for anybody, and what one session leaves on the machine is there for the next. Every session is open to the project’s members: read what an agent did, pick it up, carry it on.
- Common
- The project the whole organization shares. Every member can open every session in it.
- Personal
- Yours alone. Nobody else can open it, owners and admins included.
- Your projects
- One per line of work, each on a cloud computer of its own. Work that must not share files goes in a separate project.
An agent for every job. Set up once, for everyone.
An agent is a markdown persona with a prompt and a tightly scoped reach into tools. Install one in a click from the marketplace, or write your own. Agents, skills and connectors live in your organization’s Customize, so every project and every teammate works with the same agents. Skills ride into every session, so what the company learns compounds.
Bring your own API key for any major provider, or sign in with the ChatGPT subscription you already pay for. Set the models once for the whole organization, and switching provider is a setting rather than a migration.
- SalesResearch, drafts and CRM hygiene, held for approval
- Builds the outbound campaign
- Stops before anything is sent
- EngineeringReproduce it, patch it, open the change request
- Reads the logs and ships the fix
- Opens a change request against main
- MarketingProduction work that sounds like you
- Runs the SEO programme weekly
- Rewrites land as change requests
- FinanceThe close, the reconciliation and the variance note
- Reconciles the ledger against the bank
- Chases the missing receipts
- PeopleScheduling, kits and onboarding — never the decision
- Sources and scores the pipeline
- Drafts the first approach
- ProductFeedback synthesised into specs, with the evidence
- Groups feedback into themes
- Takes the top one to a change request
- Data ScienceA real machine, a real query, an analysis you can re-run
- Queries the warehouse on a schedule
- Posts the chart to Slack
- Your own agentA markdown file in .singulacomp/opencode/agents/
- Define the role and its tools
- Every project of the organization gets it
Agents propose. People decide.
When an agent wants to keep something, it commits and opens a change request toward main. Change requests, the connector calls you set to ask first, and the outputs agents submit for sign-off all wait in one Review inbox for the whole organization. Whoever holds the role reads it and decides. Merge is default-deny for an agent unless you grant it in singulacomp.yaml.
- In reviewFix the forecast cache expiryForecast API · +2 −1 · 118 tests passEEngineering
- Needs youSend the Harbor Point renewal proposalCommon · Ask set on send · 3 draftsSSales
- MergedWeekly API usage report for #metricsCommon · Posted to Slack · 06:00DData Science
Ask in the web app, or in a Slack thread.
Start a session in any project, or mention the agent in a Slack thread your team already uses — Slack connects once for the whole organization. Each session does the work on a cloud computer with the repo on it and comes back with the result: a report, a deck, a change request, a reply.
Ask in chat and get it now.
The agent works and checks in for the calls that matter.
A cron schedule or a webhook starts the session while you sleep.
- Web app
- Slack
Everything is code you own.
Every agent, skill, trigger and grant is a file in a git repository your company owns — the organization’s and each project’s. Read it, diff it, roll it back, and run it with any model and your own keys.
From sign-up to a working team.
All of it in the web app. Nothing to install, no laptop involved.
- 01DescribeCreate your organization: its name, its domain and what it does. The organization is your company.
- 02ConnectPick the models and connect the tools your company runs on — 3,000+ apps — once, for every project. A shared account serves everyone; anyone can add their own.
- 03InviteBring in your team. Everyone shares the Common project and gets a Personal one of their own.
- 04WorkStart sessions. Agents work on cloud computers, triggers fire in the night, and what they want to land waits in Review for a person.
Questions people ask before the first session.
Short answers, including the awkward ones.
- What does an agent actually run on?
- A real Linux cloud computer with your repo and your tools already on it. Every session of a project runs on that project’s one machine, each in its own conversation, so teammates work side by side and see the same files. Two sessions that edit the same file at once overwrite each other, so give parallel work different files; projects never share a machine. The agent can install, run and break anything on it; git is the only durable record, so what it commits and pushes is what survives.
- Who can see my sessions?
- Every member of the project it runs in: a project’s sessions share one machine, so they share one audience too. Common is open to the whole organization. Your Personal project is yours alone — nobody else, owners and admins included, can open it — so private work belongs there.
- Do I need a team to start?
- No. Start on your own with your agents — Free includes two organizations and unlimited projects, and a session you start without picking a project lands in your Personal one. Invite people when you need them: roles cover people and agents alike, and members are unlimited on every plan.
- What stops an agent doing something I did not want?
- You set allow, ask or block per action, once for the whole organization, and a rule can match on a glob, a regular expression, or the arguments of the call itself. Be aware that nothing is switched on for you: an action no rule matches runs without approval, so this is a thing you configure rather than a thing you inherit. Session work reaches main through a change request, and merge is default-deny for an agent — an admin has to grant it in singulacomp.yaml, and that grant is itself an edit somebody else has to merge.
- Can I see everything an agent did?
- It is all recorded, on every plan, and every record names who acted: a person, or an agent and the person it acted for. Recording is never gated — actions, approvals and the session transcript are captured whatever you pay. Reading, exporting and streaming that audit trail out to your own SIEM are not part of the Free and Team plans: the record exists, but the export does not.
- Is my company data yours?
- No. Your organization’s agents and skills and each project’s work live in git repositories — clone them and everything the company has learned comes with you.
- Can I use the models I already pay for?
- Yes. Bring your own API key for any major provider, or sign in with the ChatGPT subscription you already pay for; on Team, SingulaComp’s managed models need no key at all. You set the models once for the organization — which ones its projects may use, and the default — so switching provider is a setting rather than a migration.
- What does it cost?
- Free is $0 — two organizations with unlimited projects, and 200 credits a month of sandbox compute, with your own key for any premium model. Credits is $29.99 a month for 3,000 credits and Credits Plus is $99.99 for 10,000; members are unlimited, and people and agents draw on one shared pool.
Your people and your agents, in one place you own.
Start with one job and grow from there.
Free to start · No credit card required