Growth, on a Loop.
GenieOS is the marketing OS your agents run, from your codebase or whatever LLM you live in. Set the objective, connect your database, and let the loop turn results into the next campaign.
Free To Sign Up · No Card · Live In Minutes
Your Database, Connected · Your KPIs, Watched · Checkpoints, Yours
AFTER THE DEPLOY
You shipped the app.
Now ship the Growth.
You can build anything in a weekend now. Users are the part that doesn't deploy.
$ npm run deploy✓ deployed in 38s $ users --new --today0THE AGENTIC LOOP
Objective set. Iterations through. Objective Met.
Set the objective over MCP and the loop runs: create, launch, measure against your own tables, learn, go again. It stops when the KPI says done. Checkpoints wherever you want them.
users.activated_at missed target by 8%.
- Next iteration drafted
Awaiting your checkpoint.
- Objectiveyour KPI
- Create
- Launch
- Measureyour data
- Learn
- Completeobjective met
ONE API
Every channel, live in minutes. From one API.
Email, social and landing pages. Your agents don't click through an interface, they query. MCP for your agents, REST for your codebase, webhooks for everything the loop learns.
// claude_desktop_config.json, or any MCP client{ "mcpServers": { "genieos": { "command": "npx", "args": ["@genieos/mcp"], "env": { "GENIEOS_API_KEY": "gos_live_…" } } }}// that's it: your agent now does marketing// npm install genieosimport { GenieOS } from 'genieos'; const mg = new GenieOS(); // marketing designed "welcome", you fire it by IDawait mg.templates.send({ to: '[email protected]', template: 'welcome', variables: { first_name: 'Ada' },});# pip install genieosfrom genieos import GenieOS mg = GenieOS() # same tools, same brand brain, sync or asyncmg.templates.send( to="[email protected]", template="welcome", variables={"first_name": "Ada"},)# a branded send, one requestcurl -X POST https://api.genieos.pro/v1/messaging/transactional \ -H "Authorization: Bearer $GENIEOS_API_KEY" \ -H "Idempotency-Key: welcome_ada_001" \ -d '{ "templateKey": "welcome", "to": "[email protected]", "variables": { "first_name": "Ada" } }'Swipe across the example to explore the editor.
Block
Text
Content
NEW RELEASEColor
InheritFont family
Weight
Style
Decoration
Font size
Line height
Letter spacing
Padding
Top
Right
Bottom
Left
Background colour
TransparentBorder radius
Border width
Border style
Border colour
InheritLayout
Refine in
Seconds.
Your agents patch the subject line over MCP, or a human opens the email editor and drags the blocks. Deterministic brand checks either way.
Stop pasting context into every Prompt.
One URL, read once. Brand and strategy become structured objects your agents query, with deterministic checks on the way out. Email and social ship from the same memory.
> send_transactional({ segment: "joined_this_month AND source_count = 0" })
// brand and voice read from memory
> publish_social_post({ angle: "directory launch", networks: ["linkedin", "x"] })
- Sg
- GenieOS
Social
NEW RELEASE
Your health data finally has a home.
Explore the Directory//“This is going to save me 1 day per week … can I start using it now?”
$ npx @genieos/mcp init
→ api key gos_live_k3H…9fQ
✓brand read from yoursite.com (9s)
→ 24 tools registered
✦ wish away.
Questions, Answered.
Which agents can drive GenieOS?
Anything that speaks MCP: Claude, Cursor, ChatGPT and friends. Or plain code through the REST API and typed SDKs in Node and Python.
What can GenieOS read from my database, and what leaves it?
You choose the exact tables and columns. GenieOS reads them to segment and to score campaign KPIs; your data stays yours and the scope is yours to shrink at any time.
Can the loop run without me?
By default, no. Checkpoints before anything ships or spends. Loosen the leash per loop when you're ready.
How is this different from Klaviyo or Customer.io?
Klaviyo's AI drafts. Customer.io analyses, in read only mode by design. In every one of them a human carries the learning between campaigns. In GenieOS the results write the next campaign, and your agents can read every result over the same API.