Obi for Marketing Engineers
An AI search system to build on: your ICP, your buyers' questions, what AI answers, and the missing piece to ship.
npm i -g @hireobi/cliA new role
The role is new: a full-stack marketer with the skills of a builder. Obi is the AI search part of that stack, in the agent you already script.
The job
You build marketing as systems: agents, workflows and data that run every week. AI search rewards that craft first, because AI answers move with what you publish.
Where it hurts today
- Tracking dashboards show mentions and stop there.
- Prompt lists written by hand, with no ICP behind them.
- No way to tie an AI answer to a deal.
The workflow with Obi
From the agent you already use. Each step is one command.
- Point it at your site
Obi reads your site, reviews and rivals, then proposes your ICP with its sources.
/obi - Review the questions
Your buyers' questions by stage and intent, versioned like code.
obi questions show - Read the scans
What each engine answered, when, and the sources it cited.
obi questions results - Ship the next action
Three actions this week, each with the observation behind it.
obi strategies - Script it
Every command answers in one validated JSON envelope.
obi status --json
Where it plugs in
- GitHub
Obi opens pull requests on the product's repository for the pages and docs it writes.
- Attio
Obi reads the won deals to find the ICP, and ties the leads AI brings to revenue.
- PostHog
Obi shows the visits and sign-ups that come from AI.
- Google Search Console
Obi reads the queries, the indexing, and the impressions in Google's AI features.
What stays in your hands
- Obi recommends and drafts. It never publishes, sends outreach or buys anything on its own.
- Every command that starts paid work takes your explicit yes.
- Your secrets and the rest of your workspace stay on your machine.
Start with your own buyers' questions.
Hire for free