djentic.ai Join early access
Now in early access

You're one person. Run a team of fifty.

The AI agent team you manage from one place.

Build your product and run your business with a lead for every team, all working in parallel, while you decide what matters.

Join with just your email. Bring your own model keys. Built by djentic agents.

You're on the list. We'll be in touch.

Start hereI build softwareI run a business

Works with ClaudeOpenAI CodexOpenRouterand hundreds more models, each task on the right one

Built by djentic

Open one and click around.

This site and Keystone, a full OKR app we built as a showcase, were made end to end by djentic agent teams. Both are working: open one and try it.

Open Keystone OKRs
Keystone OKRs, a showcase app built by djentic agents: the overview panel with the company score of 0.51, 21 objectives on track or done, pace against plan and key result mix.

Put a team of agents on your project.

  • Hands-on help to get started
  • Bring your own model keys
  • Your code stays on a machine you own

You're on the list. We'll be in touch.

What the team does

Same team. Two ways to use it.

Ship real apps.

djentic puts a lead and a team of helpers on your actual project: a real backend, real accounts, feature after feature. They work in parallel, each on its own branch. Every change passes your checks before it merges, and a live link updates after each merge.

  • Each agent works on its own branch, in its own copy of the code.
  • Every change merges only after your project's checks pass.
  • A live preview updates after every merge. Share it with anyone by email.
  • Claude, OpenAI Codex and hundreds of other models in one team, each task on the right one, so you get the most from every token.
  • Bring your own keys. Set a budget per team and watch spend live.
  • Agents run freely in a Docker container you control, and your code stays on a machine you own.

Watch a run

Pick a goal. Watch a team take it.

A replay: a scripted example, sped up. Watch it run.

  1. Pick a goal above to start the replay.

Bookings app

Replay

0 of 4 merged0 waiting on you

    djentic.ai is built by djentic.ai.

    Its own agent teams build the product you'd use. They made these too.

    See everything they built

    Put a team on your product, or your business.

    • A lead for every team, all working in parallel
    • One inbox for the decisions only you make
    • Hands-on help to get started

    You're on the list. We'll be in touch.

    How you manage them

    You manage the team.

    Dozens of agents, all moving at once. djentic gives you the controls: a lead for each team, one inbox for your decisions, and a live view of everything.

    • Leads and teamsA lead owns each goal: it splits the work, hires helpers and picks the model. You talk to leads.
    • The agents gridEvery agent at a glance: working, waiting on you, or finished.
    • One inboxEvery decision that needs you, on the web and on your phone.
    • A shared whiteboardAgents post what they're doing and what they decided. Each keeps its own to-do list.
    • Cost in viewSpend per agent, live, with daily or weekly budgets you set.
    • Your calendar backSchedules re-run a goal on a timer. Your phone pings only for decisions.

    Steer dozens of agents from one inbox.

    We'll email you when your spot opens.

    You're on the list. We'll be in touch.

    Smart by design

    Get the most from every token.

    With a team of agents working at once, djentic's brain decides who does what, on which model, with how much context. You set the budget and watch it work.

    • The right model for every taskEach task goes to the lightest model that does it well. Frontier models take the work that needs them.
    • Budgets you set, spend you seeGive each team a budget and watch spend live, agent by agent.
    • Your code stays yoursAgents work in a Docker container you control, on a machine you own.
    • Mix and matchClaude, OpenAI Codex and OpenRouter models in one project.
    • Lean from start to finishContext compaction keeps long runs focused: agents carry exactly what matters.
    • Right the first timeBuilt-in skills are how-to guides agents load first: less rework, fewer tokens.

    Put the brain to work on your project.

    We'll email you when your spot opens.

    You're on the list. We'll be in touch.

    For builders

    Under the hood.

    The short version, for whoever sets it up.

    How it works, in short
    Worker
    Runs on a worker you control: Mac, Linux or Docker (Windows via WSL). The cloud coordinates; your code stays on the worker.
    Isolation
    Each agent gets its own branch and git worktree. File claims show overlaps before they turn into conflicts.
    Checks
    Your project's check command runs on the worker before every push. Failures are re-run and bisected.
    Merges
    Serialized per project: one merge at a time. Docs-only changes go straight through.
    Models
    Claude, OpenAI Codex, and hundreds more through OpenRouter. The brain routes each task to the lightest model that does it well; pin a model per agent when you want one. Context compaction keeps long runs lean.
    Keys
    Bring your own API keys (Claude, OpenAI, OpenRouter and more).
    Integrations
    GitHub, plus any REST API or MCP server. Secrets are sealed, grants are per team, and writes are off by default.
    Containers
    Run the worker in Docker and agents work freely inside a container you control, away from your everyday machine.

    Run it on a machine you control.

    Early-access users get help with setup.

    You're on the list. We'll be in touch.

    FAQ

    Questions, answered.

    For builders

    Engineering teamsAI buildersSolo developersAgencies

    Ship in parallel. A lead splits each goal, helpers work side by side, and you review the result.

    Does it work on my real codebase?

    Yes. djentic works in your own GitHub repo, checks every change before it merges, and keeps a live preview up to date.

    How is this different from prototype builders, or running Claude Code myself?

    Prototype builders get you a demo fast. djentic works on your real project: each goal gets a lead that splits the work, runs helpers in parallel and asks you only for decisions.

    Which models and keys does it use?

    Claude, OpenAI's Codex models, and hundreds of others through OpenRouter, on your own API keys. The brain picks a model for each task, or you pin one per agent.

    What do I need to get started?

    A computer (Mac, Linux, or Windows via WSL) or Docker, a GitHub account, and a model key. A guided setup takes you through it.

    For business owners

    Founding CEOsSmall business ownersMarketing teamsOperators

    Run the business with a team behind you. Research, writing, planning and operations get done, and you make the calls.

    Can I use it if I'm not a developer?

    Yes. You set the goals, review live results and make the decisions; the agents do the building. Early-access users get hands-on help with setup.

    What can it do for my business?

    The same team does research, writing, planning and operations: tracking competitors, drafting docs, planning launches, preparing marketing. We run our own marketing that way.

    What will I see day to day?

    A live view of what each team is working on, finished results to review, and one inbox for the decisions only you can make.

    Will I get help getting started?

    Yes. Early-access users get hands-on help getting set up, and a guided setup takes you through it.

    One workspace for the whole team: builders ship the product, business owners run the business.

    For everyone

    Who is djentic for?

    Builders (engineering teams, AI builders, solo developers, agencies) who ship products, and business owners (CEOs, owners, marketing teams, operators) who run everything around them. Early-access users get hands-on help getting started.

    How many agents can I run?

    Dozens, across your teams, all working at the same time.

    What does it cost?

    Your agents use tokens from your own model keys, and you pay those providers directly. Set a budget per team, watch spend live agent by agent, and let the brain send each task to the lightest model that does it well.

    How does djentic make the most of my tokens?

    The brain routes each task to the lightest model that does it well and saves frontier models for the work that needs them. Context compaction keeps long runs focused, and built-in skills are curated how-to guides that agents load so they get it right the first time, with less rework. Mix Claude, OpenAI Codex and OpenRouter models in one project.

    Where does my code live?

    On your own machine or in a Docker container you control. djentic's servers coordinate the work while your code stays with you. We recommend the Docker worker: agents run freely inside a container you control.

    Is it available now?

    Yes, in early access. It's opening in waves, with hands-on help to get you started. Join the list below and we'll email you when your spot opens.

    Scale yourself. Join early access.

    Early access is opening in waves, with hands-on help to get you started.

    We'll email you when your spot opens.

    You're on the list. We'll be in touch.