Everything the crew can do, behind one approval gate.

1.0 — Research & signals

Find the thread before it goes cold.

Ved reads your product and your commits, then audits search and AI answers. Shravan listens where your buyers talk. Every signal is scored by intent.

• Shravan listens on Reddit, Hacker News, Product Hunt, and GitHub
• Ved turns merged commits into story angles
• Agni searches the Ad Library across six ad networks

Ved on Pro · Shravan on Studio, Agni on GTM Desk

2.0 — Brand voice

Your voice, not generic AI.

Start from a product URL. Your saved brand voice, product context, and examples guide every draft, and you can tune them whenever a draft misses.

• Site intake reads your product
• Saved examples steer tone and structure
• Reject a draft, update the voice, draft again

Tone

Plain, specific, a little dry

Audience

Technical founders, first 100 customers

Avoid

Hype words, exclamation marks, emoji

Examples

Launch postChangelog threadReddit reply

Kavi reads this before every draft.

3.0 — Content engine

One idea. A native draft per channel.

Kavi drafts per platform from Ved's research and your commits: posts, threads, replies, and blog articles, with variants to pick from.

• X threads checked against 280 characters per post
• Credit cost on every draft before it runs
• Edit, reject, or approve in one place

4.0 — Creative studio

One brief. A still, a carousel, and a clip.

Images, LinkedIn carousels rendered from your brand tokens, and video priced per second before it renders.

• Carousel text rendered, so every word is spelled right
• Video tiers: budget, pro, and premium
• Attach the asset to any queued post

A 20-second pro-tier clip starts at 280 credits before other costs. Product demos are a flat 150 credits.

5.0 — Publishing & approval queue

Nothing ships without your yes.

Approve mode is the default for every agent, and approving is a human step in the dashboard. Chanakya publishes what you approve, adapted per platform, at the hours your audience reads.

• Publishes to LinkedIn, X, Reddit, and email
• YouTube Shorts in verification; Instagram and TikTok in beta
• UTM tags on every link
• Approval-first by default. Later you can let a trusted agent auto-publish, a switch you flip in the dashboard.

6.0 — Engagement

Join the conversations worth joining.

Shravan surfaces mentions and high-intent threads. Mitra drafts replies in your voice. Reddit replies publish after you approve them.

• Replies scored by intent, not volume
• Hacker News is listen-only
• Studio and GTM Desk

7.0 — Analytics & attribution

Know what each post actually drove.

Shravan pulls Google Analytics and Search Console, follows the UTM tags Chanakya added, and rolls it into a weekly report.

• Channel attribution from your own analytics
• Mentions and sentiment in one roll-up
• Honest empty states until data arrives

8.0 — GEO & AI visibility

Get cited when buyers ask an AI.

Ved audits your pages for search and answer engines, tracks where AI answers cite you, and ranks fixes by buyer intent.

• GEO checklist with drafted fixes
• AI-citation tracking
• Scheduled GEO scans don't consume credits

• Organization schema on the homepagePass
• FAQ schema missing on /pricingFix drafted
• llms.txt publishedPass
• AI-answer citations for your categoryTracked weekly

9.0 — CLI, MCP & API

Run the crew from where you already work.

The aicmo CLI signs in with a device code and scoped tokens. The MCP server brings the same tools to Claude, Cursor, ChatGPT, or Codex. Draft, list, edit, and schedule from there; approving a draft stays a human step in the dashboard.

• Scoped tokens such as queue:write and publish
• CI-friendly JSON output
• API under /api/v1 for your own scripts

ClaudeCodexCursorChatGPT

10.0 — Credits & mission control

See the price. Then decide.

Every action has a credit cost shown before it runs. When credits run out, actions pause. Nothing fails silently or overspends. The dashboard cockpit and a morning brief show what's queued and what needs a yes.

• Monthly credits reset each cycle, no rollover
• Top-up packs stack on top and don't expire
• Exact costs in-app and via the credit_costs MCP tool