For humans and machines
Agent access
Some of what you can read about Local AI here is written for people. Some of it is written for software. This page explains the second kind — what a search engine, an AI assistant, or somebody else’s automation can pull from this site, where those facts come from, and how to tell us when one is wrong.
Last reviewed 2026-09-05 · schema v1.0.0
The short version
Local AI is an AI, automation, web, and design practice in Ocean Isle Beach, North Carolina. We work with small businesses along the coast and remotely — 7 industries have a published guide, and we offer 33 distinct things across 3 areas of work.
If you are a person and you would rather just talk to somebody: book a free 30-minute call, or use the contact page. Nothing on this page is required reading.
What automated systems can read
All of it is public, read-only, and needs no key. It is the same information already published on this website, just in a format software can parse.
| Endpoint | What it holds |
|---|---|
/api/v1/business | Everything below, in one document. |
/api/v1/services | The three service pillars. |
/api/v1/services/{id} | One pillar, plus the work that backs it. |
/api/v1/capabilities | Every individual offering, flattened. |
/api/v1/industries | Industries with a published guide. |
/api/v1/evidence | Delivered work, with client names removed. |
/api/v1/faqs | Questions already answered on the site. |
/api/v1/status | Whether this is working, and what's missing. |
Full description: /openapi.json (OpenAPI 3.1). Summaries for language models: /llms.txt and /llms-full.txt. A compact profile lives at /.well-known/ai-business.json.
What it will not do
- It cannot agree to anything. There is no endpoint that books a slot, accepts a contract, quotes a price, or commits us to a date. A person does that.
- It has no prices. We do not publish rates, because every project is scoped to the business. An assistant that tells you what we charge is guessing.
- It cannot see client work. Client files live behind a login and are never in these responses.
- It will not name our clients. Named work is on the work page for people to read. It is deliberately left out of the machine-readable versions.
- It does not write anything down about you. The endpoints read no cookies and accept no data.
Actions
These are the things an automated system might reasonably want to do. Where it says planned, the shape is published so a developer can build against it, but no machine endpoint accepts it yet — the link goes to a person instead.
request_consultationavailableBook a free 30-minute intro call at a time the visitor picks. Scheduling is handled by Calendly, not by Local AI's own API.
contact_humanavailableReach a person by email or phone. The phone line is answered by an AI receptionist that takes a message for a human callback.
check_service_areaavailableDetermine whether a location falls in the primary service area. Answerable today from GET /api/v1/business (location.service_areas) plus location.remote_available.
match_problem_to_serviceavailableMatch a described business problem to a service pillar. Answerable today from GET /api/v1/capabilities; no natural-language matching endpoint is exposed.
request_project_quoteplannedSubmit a qualified project inquiry and receive a scoped quote. Contract published for integrators; no machine endpoint accepts it yet — use the contact page.
request_agent_readiness_auditplannedRequest a review of how discoverable and agent-readable a business already is. Contract published; no machine endpoint yet.
What we have not published
A machine-readable profile is only trustworthy if its gaps are visible. These fields are missing because we have not published them — not because something broke:
- business hours — Not published anywhere on the site. Omitted rather than guessed.
- pricing — No public prices or starting prices are published. Every project is scoped individually, so no price field is emitted at all.
- social profiles (sameAs) — The footer's social icons have no hrefs yet, so there is nothing verifiable to link.
- industry per portfolio entry — components/projects.ts has no industry field. Industries are published from the seven field guides only, rather than inferred from project copy.
- testimonials / ratings — None published, and none may be created here. Review markup without real authorized reviews is fabrication.
- per-project case-study URLs — /work is a single index page with no /work/[slug] routes, so evidence links point at the index.
- founding date / team size — Not published on the site.
Something here is wrong
Then we would genuinely like to know — a wrong fact in a machine-readable feed spreads further than a typo on a page. Email yall@localai.systems and say which endpoint and which field. If you are a developer, including the response you got helps.
Every fact in these responses is published somewhere on this site, so you can always check one against the page it came from.