Autonomous agent teams
A Project Leader agent hires and directs a team of specialists that plan, execute, and finish the work for each project — on a cadence you set.
Many projects, one workspace
Run every project inside a single workspace. Shared engines, per-project memory, credentials, and context — isolated where it matters.
You stay in control
Every plan, draft, and result is transparent and reviewable. Approve, redirect, or step in over chat, dashboard, CLI, or MCP — nothing ships behind your back.
Real jobs, not demos
What you can put a team of badgers on
Anything you can describe and check. These are the three people ask for first.
Publishing
Write and publish, on a cadence
A writer agent researches and drafts, an editor agent reviews it against your rules, and the publisher pushes it live to your site on the schedule you set.
- Drafts held for approval, or published straight through
- Images generated and attached to the post
- Runs to a schedule without being asked
Inbox
Work the inbox for you
Agents read what lands, sort it, draft the replies that are obvious, and put anything that needs a human decision in front of you with the context already gathered.
- Triage and categorise as mail arrives
- Draft replies ready for you to send
- Escalate only what genuinely needs you
Messaging
Reach you where you already are
Connect Telegram and your agents can report in, ask a question, and take an instruction from your phone — no dashboard needed to keep things moving.
- Approve or redirect work from a chat
- Get told when something needs a decision
- Kick off a job from your phone
One agent finishes. The next one picks it up.
Work moves down the chain on its own — planned, executed, checked, delivered — and stops at you only when a decision is genuinely needed.
The Leader
Project Leader
Works out what the job needs and who should do it
The Researcher
Specialist
Gathers the material and hands over the findings
The Writer
Specialist
Turns it into the actual thing you asked for
The Publisher
Specialist
Checks it, then ships it — or waits for your nod
Every hand-off is visible in the Plan feed, so you can see where a job is and step in at any point in the chain.
How it works
Describe the job
Tell the onboarding consultant what the project is and what you need done. Pick the AI engines that power it.
Meet your Project Leader
Honey Badgers generates a Project Leader agent that understands the goal and its available tools, then assembles the team it needs.
Review and approve
Work flows through the Plan and Publisher feeds. You approve what goes out; the team keeps the cadence running.
Bring your own AI
Plug in the AI you already pay for
Honey Badgers is the orchestration layer, not the model. You attach the engines — a Codex subscription you already have, a Claude API key, or both — and we run the teams on top of them. No per-token markup from us.
Point a different engine at each project: something cheap for routine grinding, and something stronger where the output actually matters.
What an engine subscription costs
Codex is included in every ChatGPT tier, so the cheapest way to run agents is usually a subscription you may already be paying for. These are OpenAI's prices, not ours.
| Plan | Price | What it suits |
|---|---|---|
| ChatGPT Go | $8per month | The cheapest way in. Light, occasional runs — a single project ticking over. |
| ChatGPT PlusMost start here | $20per month | The usual starting point. Several projects on a daily cadence. |
| ChatGPT Pro | $100–$200per month | Heavy, near-constant use. $100 buys the 5× tier, $200 the 20×. |
| ChatGPT Business | $20per user, per month | Shared workspaces where each teammate brings their own seat. |
OpenAI pricing, checked 29 July 2026 — see their current rates. Prefer to pay per token instead? Connect a Claude or OpenAI API key directly.
Mix and match per project
Engines are set per project and per agent, so a throwaway internal task and your customer-facing work do not have to run on the same model — or the same bill.
The honest caveat
Output quality tracks the engine you choose. A cheap tier on a hard job reads like a cheap tier on a hard job — which is exactly why every result stays reviewable before it goes anywhere.
Pricing
Plans are priced by projects and login seats. Every tier is BYO — you bring your own AI, social, and CMS keys; the subscription covers the platform.