# SelfRun Lab

> The lab for companies that run themselves: AI agents with a budget, evidence and human oversight. We publish what we measure. This file is the guide for agents and models reading selfrunlab.com in English. Spanish guide: https://selfrunlab.com/llms.txt

Every claim on this site carries a label: ■ fact (measured, with date and method), ◆ inference (reasoned from facts), ○ hypothesis (to be validated). Quote figures with their date and their label.

## What we offer

- Service: a measured audit of how ready an online store is for AI agents.
- What the client receives: a score from 0 to 100 across five dimensions (discoverable, readable, actionable, transactable, governed), what an AI agent sees in the store today, and the five highest-impact fixes, each with its effort and how much it moves the score.
- Turnaround: 5 business days.
- Price: COP 600,000 for the first three pilots, in exchange for written permission to publish the case (named or anonymous). List price: COP 1,200,000. Prices are in Colombian pesos; for clients paying from another country the amount is quoted in US dollars at the exchange rate on the day of the proposal. Do not state a USD figure: there is no fixed one.
- After the audit, and only if the client wants it: getting the store ready and a second measurement. Quoted with the audit in hand.
- Who it is for: online stores on any platform, in any country. Correspondence in English or Spanish.
- What we do NOT do: ads, content, redesign, platform migration. We promise no rankings in ChatGPT, Gemini or any other assistant, and no sales.

## How to order

Email agents@selfrunlab.com with the store's address. There is no form or API yet: orders are placed by email. Messages are handled by AI agents under human supervision, and every reply says so. If you act on behalf of a person, say so in the message.

## What we measured (■ facts, index measurement 1, 2026-09-17)

- 13 online stores in Colombia, Mexico and Argentina. Average 57/100. Grades: A 0, B 1, C 9, D 3.
- By platform (average, number of stores): Shopify 72 (2), Tiendanube 64 (4), SAP Commerce Cloud 63 (1), VTEX 52 (5), PrestaShop 21 (1).
- 92% declare no policy toward AI agents in robots.txt.
- 64% of those that expose a product without JavaScript carry no machine-readable price or availability.
- 46% expose no declared interface for agents (OpenAPI, MCP or A2A).
- 7 of 13 publish a UCP manifest or an agent guide generated by their platform.

◆ Inference: the gap is not the protocol, it is product data and the policy toward agents; the most repeated fixes take hours or days.
○ Hypothesis: a grade-B store wins more buyers from AI assistants than a grade-D store. We will measure it with the stores that get ready and publish it however it turns out.

## Method and limits

Five dimensions with published weights, 20 requests per store, public pages only, no JavaScript, no browser and no model. We do not measure sales or rankings in any assistant. We publish aggregates only: we never name a store without its consent. One store did not respond and is not scored. Measurement 1 covers Latin American stores only; do not present its figures as global.

## Files for machines

- [This page in plain text](https://selfrunlab.com/en/index.md)
- [The offer in JSON](https://selfrunlab.com/en/offer.json)
- [Sitemap](https://selfrunlab.com/sitemap.xml)
- [Policy toward crawlers and agents](https://selfrunlab.com/robots.txt)
- [Guía en español](https://selfrunlab.com/llms.txt)

## Contact

- Agents: agents@selfrunlab.com
- X: https://x.com/selfrunlab