Obi, the free AI Search Employee driving revenue from your terminal

Obi is a free AI Search Employee for tech companies: an agent you hire from your terminal that learns your ICP, finds the missing pieces that get ChatGPT, Gemini and Claude to recommend you, and does the work end to end, so AI search brings you buyers and revenue.

#What is an AI Search Employee?

An AI Search Employee is an AI agent that does the work of AI search for a company: it learns who the company sells to, reads what AI assistants answer when those buyers ask what to buy, and ships what is missing so the assistants recommend the company. It works like a hire with a job, where a dashboard only reports a number.

AI search is the channel where buyers ask ChatGPT, Gemini, Claude or Perplexity which product to choose. The work of being chosen there is also called GEO (generative engine optimization) or AEO (answer engine optimization). Obi is the AI Search Employee we built to do that work for tech companies.

#Why AI search is the best GTM channel for startups

Buyers now shop inside AI. A B2B buyer describes the team, the stack, the budget and the problem to an assistant, and the assistant answers with three or four names. The shortlist moved into that answer.

An LLM reads the buyer, the whole conversation, and writes one answer for that buyer. So the brand that wins is the one that fits the buyer's situation, more than the one with the most links. When no page answers a buyer's precise situation, the first company to publish one gets a real chance to be named. That is why a young company can win fast, often in weeks, with no years of SEO behind it.

It also brings the right buyers. The person asking is already describing their need, at the moment they choose. Our manifesto says why we hold this thesis.

#Obi works from your terminal

Obi runs inside the coding agent you already use: Claude Code, Codex, Cursor, or any terminal. Type obi after the install and it greets you in the terminal, then you call it from your agent's chat.

Running in your agent changes three things:

• No new dashboard. The work happens in the chat you already have open, next to your code and your docs.
• Your agent's power. Obi researches and reasons with your agent's own search, browser and subagents, so it reads your site, the reviews and your rivals in parallel.
• Your plan pays for the reasoning. Obi runs on the agent you already pay for, which keeps the price of the work low.

It also works in six more coding agents it was tested in: Windsurf, GitHub Copilot, Gemini CLI, Cline, Amp and OpenCode. The agents page lists each one.

#How Obi drives revenue, step by step

Obi follows the same workflow every week, and each step is one command you can read.

1. It learns your business. Obi reads your site, your blog, the reviews and the "alternatives" pages, then your rivals' home, product and pricing pages. It proposes your offer, your buyer, your market and your competitors, each claim cited, for you to confirm.
2. It writes your buyers' questions. Each question is written the way your buyer types it into an AI: first person, one ask, a real situation. The set covers the whole buying path, from the category to comparisons and trust, in any country and language you choose. You review the set before anything is scanned.
3. It scans the AI answers. Obi's servers ask the engines your questions and keep what each one answered, when, and the sources it cited. A scan is the only thing Obi calls a measurement; your own agent's answer stays a suggestion.
4. It checks your website. Obi reads your robots.txt for each AI search crawler, finds pages AI cannot index, facts hidden behind JavaScript, and gaps in your structured data and llms.txt. Each finding comes with its fix.
5. It ships the missing pieces. After each scan, Obi ranks three actions for the week by urgency, buyer importance, evidence and effort, then drafts the work: pages, comparisons, FAQs, fix briefs, pitches and replies.
6. It measures again. When you ship a change, Obi records what changed and when, and reads it before its next decision, so you see what moved.

Nothing goes live without you. Obi recommends and drafts; it never publishes, sends outreach or buys anything on its own. Every feature is on the features page.

#What makes Obi different from AEO tracking tools

Most AI visibility tools are dashboards: they track a list of generic questions and report how often your brand appears. Obi starts from your ICP and ends each week on work you can ship.

The difference is the method. Tracking generic questions without context is spray and pray: a vanity metric that moves no pipeline. Obi asks the questions your buyers really ask, and gives AI the piece it is missing.

#Start in one command

Install the CLI with npm i -g @hireobi/cli, register the Skill with obi skill install, then open a new chat and type /obi in Claude Code or Cursor, or $obi in Codex. Obi asks which website to work on and follows the language of your conversation.

#How much Obi costs

Obi is free to start: you can hire him for $0 and see your first answers. When you want more questions and weekly scans, you add packs, and every pack costs the same. You buy them in your agent, after your confirmation. The plans and prices are on the pricing page.

For a team that wants the work done for it, our Forward Deployed Agent-led Engineers build the whole AI search architecture with you.