Pay for the operating layer, not generated text.
The copy is cheap. The thing that matters is the loop around it: grounded launch state, approvals, real execution, 24h measurement, and memory that improves the next move. Drafting stays free. Bring your own provider keys when you want zero-markup execution, or use ours hosted with transparent pass-through pricing. Spec v0.1 defines the contract.
Per-outcome pricing
Every metered API response includes costSource and estimatedCostUsd in real dollars. BYOK column is what you pay the provider directly (ChiefLab takes no margin). Hosted column is the pass-through cost when you use our server-side key.
Every billable action is approval-gated. The agent stages a publishAction; you approve inline in chat or on the signed reviewUrl fallback; only then does the connector fire. Pre-approval calls return { reason: "requires_approval" } and cost nothing. Hosted execution is invoiced monthly against actual estimatedCostUsd totals — no bundles, no credit packs, no expiry games.
Workspace tiers buy persistence, not credits
The per-outcome pricing above is the floor — it applies the same to every plan. Workspace + Multi-tenant unlock the compounding state: persistent brand brain (approved drafts feed future runs), unlimited history, multi-tenant connectors, and CRM webhook delivery. Launch is the live operator today; every plan gets the full live surface as new operators ship.
Starter
$0
free forever
The daily manual loop. One channel today, measurement tomorrow.
- Today's move + manual publish + URL paste + measurement + next-move card — the full cockpit
- One auto-measured channel (Hacker News OR Reddit) + paste-what-you-see for X / LinkedIn / Email / landing
- 1 visual graphic generated per launch (Gemini if you BYOK, branded SVG fallback otherwise)
- 7-day run history
- Sandbox + key-paste flow — try ChiefLab without a card
Pro
$39
/mo
More channels, more assets, persistent memory. The loop compounds.
- Everything in Starter
- All paste-what-you-see channels with rich metric forms (X / LinkedIn / Email / landing / Product Hunt / short-form video)
- Short-form video scripts + carousels + landing-page rewrites as default assets
- Persistent brain — approved drafts, voice samples, channel performance, commenter voices flow into every next launch
- Unlimited run history; "What ChiefLab learned" card builds with each measurement
- Up to 3 graphics per launch (channel-matched dimensions)
Operator
$149
/mo
Connected execution, team approval, external delegation. The loop runs while you sleep.
- Everything in Pro
- Connected execution — social publishing rail (Zernio: LinkedIn / X / IG / Threads / Bluesky / FB / TikTok) + email sending rail (Resend); approved actions execute automatically with proof captured
- Team approval — multiple reviewers per workspace, role-based gates
- External delegation — request video / creator / editing capability from the provider network; reviewer approves the artifact; ChiefLab attaches it to the loop and measures
- Per-workspace connectors with isolated billing — agency-mode for 10+ tenants
- Webhook delivery of measurement readbacks to your CRM
- Priority support; advanced analytics export
How we think about pricing
Your LLM does the writing — that part is free.
Default outputMode: 'context' returns drafting briefs (system prompt + brand context + per-skill instructions). Your agent's LLM renders the final copy. ChiefLab pays ~$0 in tokens on the default path, so the free tier is generous.
BYOK by default. Zero markup when you bring your own keys.
Set provider keys via chieflab_set_provider_key for image generation, social publishing, and email sending — the bill comes from the underlying vendor, not from us. Every API response carries costSource: "byok" | "hosted" so you can audit who got paid. Hosted is the convenience fallback with a transparent pass-through margin to cover infra. Current adapters per rail listed in /docs/mcp-reference.
No surprise bills.
Every metered action is approval-gated. The agent stages a publishAction, renders the drafts inline in your IDE chat with the approval commands; you reply approve / revise / reject (or use the signed reviewUrl as the away-from-IDE fallback). Only after approval does ChiefLab fire the connector and incur the cost. If you reject, you pay nothing. Pre-approval calls return { reason: 'requires_approval' } cleanly.
Pricing FAQ
Is there a free tier I can use forever?
Yes — pay-as-you-go has no monthly cost. The default path (drafting briefs + brand context + signed reviewUrls + 1 workspace) is free indefinitely. You only pay when an approved action actually fires, and even then BYOK means you pay the provider directly with zero ChiefLab markup.
Can I bring my own provider keys?
Yes — that's the default. Call chieflab_set_provider_key({ provider: "gemini" | "zernio" | "resend", apiKey: "..." }) once and ChiefLab routes through your key. Image generation, social publishing, and email delivery all bill the provider directly. ChiefLab takes zero margin on BYOK calls. Every API response includes costSource: "byok" so you can audit it.
What if I just want to try it without signing up?
POST chieflab.io/api/sandbox/launch with {productUrl} — no auth required. Returns a real launchPack. Rate-limited to 3 calls per hour per IP.
How is ChiefLab different from paying for the model + ChiefLab on top?
Most calls don't go to ChiefLab's model — they go to YOUR agent's model. ChiefLab returns a drafting brief; your LLM (Sonnet, GPT, Gemini, MiniMax) does the writing. We're not a markup on token costs; we're the connector + approval + measurement + memory layer. On the hosted execution path we charge a small transparent margin to cover infra; on BYOK we charge nothing.
Will I get a bill the first time my agent calls a tool?
No. Every billable action is approval-gated. The agent calls chieflab_launch_product → returns drafting briefs (free) + signed reviewUrl + publishActions in 'pending' state. Nothing meters until you approve a publishAction and the connector fires. Even then, BYOK means the bill comes from the underlying provider (current adapters: Google for image gen + analytics, Zernio for social publishing, Resend for email sending), not from us.
Do you have an enterprise plan?
Custom rate limits, on-prem deployment hooks, SSO, custom retention windows, audit-log exports — email hi@chieflab.io. We'll quote within 24h.
Start from the agent you already use.
Give ChiefLab to the agent that built your product. It can recover setup, create a workspace when needed, and start the first launch loop from chat.