# What is AgentRel?

**AgentRel (Agent Relations)** is the work of making your product easy for AI agents to discover, recommend, and use. DevRel, for the agents that now pick tools on a developer's behalf.

## Agents pick the stack now.

**Then:** A developer reads the docs, a few comparison posts and a forum thread, then picks a tool.

**Now:** The developer asks an agent. The agent reads those pages itself, names one or two tools, and writes the integration.

If you're not in what the agent reads, you're not in its answer.

## Every recommendation passes three tests. AgentRel is the work of passing all three.

### Found. Does the agent come across you at all?

Before most answers, assistants search the web and read the top results. If your product isn't on those pages, the agent never considers it.

**What to do:**

- Write content that answers the questions your buyers ask, like comparisons and “best tool for X” guides.
- Distribute it on Reddit and LinkedIn. Assistants read and cite those discussions.
- Publish your docs as plain HTML or markdown, with an llms.txt, so agents can read them.

### Chosen. Does it pick you over the rest?

Once an agent has several options, it leans on what other sites say about each one. Comparison pages, reviews and forum threads decide which tool it names first.

**What to do:**

- Get mentioned on the comparison and list pages agents already cite for your category.
- Publish original benchmarks, so agents have a concrete reason to prefer you.
- Check what each assistant says about you, and correct wrong facts on the pages it learned them from.

### Used. Does the code it writes work?

A coding agent follows your docs to the letter. A broken quickstart or an outdated code sample becomes a broken integration in the user's project.

**What to do:**

- Test every code sample in your docs against your live API.
- Make sign-up and API keys work without a human clicking through a dashboard.
- Offer an MCP server or CLI so agents can call your product directly.
- Give a coding agent a real task with your docs, and fix wherever it gets stuck.

## Where AgentRel fits.

| | Who it's for | What it wins | Ends at |
|---|---|---|---|
| **SEO** | Search engines | Rankings | A click |
| **AEO / GEO** | AI answer engines | Mentions in answers | An answer |
| **DevRel** | Developers | Awareness and adoption | A developer who tries you |
| **AgentRel** | AI agents | Recommendations and working integrations | Working code in someone's product |

AEO stops at the answer. AgentRel continues until the code works.

## The AgentRel company.

Manicule is built around AgentRel and backed by Y Combinator. We work with devtools companies like Greptile, Supermemory, Rootly and Skyvern.

- **Let our team do the work.** We own strategy, docs, content, benchmarks and distribution, and tie it to AI visibility and revenue. [Explore services](/services)
- **Track it yourself.** See how ChatGPT, Claude and Gemini recommend you, and read the answers behind every score. [Explore the software](/agent-visibility)

Supermemory, three months with our team:

- **1M+** SEO impressions
- **+30%** Answer success after the docs rebuild
- **#1 + #2** “mem0 vs supermemory” on Google
- **200K+** Views of our Reddit benchmarks

[Read the Supermemory story](/case-studies/devrel)

## Questions about AgentRel.

### Is AgentRel the same as DevRel?

No. DevRel serves human developers. AgentRel serves the assistants and coding agents that choose and use tools for them. Most devtools companies need both.

### Is AgentRel the same as GEO or AEO?

GEO and AEO are part of it. They cover getting mentioned in AI answers. AgentRel also covers whether the agent can get your product working once it recommends you.

### Who owns AgentRel at a devtools company?

Usually the head of DevRel or developer marketing, working with the docs team. At seed and Series A, it's often the founder.

### Which agents does AgentRel cover?

Assistants like ChatGPT, Claude and Gemini that recommend tools, and coding agents like Claude Code and Cursor that install them and write the code.

### How do you measure AgentRel?

Ask each assistant your buyers' questions many times and count how often it recommends you. Then give coding agents real tasks with your docs and see if they finish.

### How long until we see results?

Docs fixes can change whether a coding agent finishes a task within days. Recommendation share moves over weeks, as new pages get indexed and cited.

## Make your product the one AI agents choose.

Work with our team, or start with the software. [Book a call](https://savvycal.com/manicule/intro) · [Start tracking](https://app.manicule.com/sign-in)