Straight answers about how Lodestar works, what it can and can't promise, and where the lines are.
Five steps. (1) Run the free check to see what models say about you today. (2) Tell us what's wrong, add the facts you want known, and rank the links you already maintain. (3) We generate your canonical profile — a clean, schema-marked page that is your single source of truth — plus a small set of supporting references that point back to it. (4) We submit everything to the major indexes and independently verify it's live and retrievable. (5) We re-test on a regular cadence and alert you to changes, publishing updates only after you approve them.
The whole input takes a few minutes. No writing or homework required.
It might — and that's expected. We don't operate ChatGPT, Claude, Gemini, or Perplexity, and we don't have access to their consumer apps. To test, we call each provider's API with a faithfully constructed search backend — the model's own native web-search/retrieval path — so our results closely mirror what a fresh, signed-out user would get.
Your own results can differ for reasons that have nothing to do with your profile:
For the cleanest comparison, ask in a fresh incognito/temporary chat with no custom instructions. We faithfully try to reproduce a neutral query; we can't reproduce your individual session exactly. What we can do is make sure there's an accurate, retrievable source for the model to find.
It varies. Publishing and indexing typically lands within days; models reflecting a new source in their answers can take a few days to a few weeks, depending on crawl and retrieval cadence. We don't assume — we run an explicit, out-of-band verification step and keep re-testing until your source is being found and used, then continue monitoring.
A realistic goal: a full-name plus a couple of keywords query resolves to you reliably. A bare common-name query is aspirational, not guaranteed.
SEO isn't our explicit goal — answer-engine optimization (AEO) is. But the two overlap heavily. Many tactics that help models retrieve and trust a source — structured data, a coherent sameAs graph, fast and accessible pages, clean canonicalization — are the same practices that improve traditional search. So it's very likely your standard search presence improves too, and the pages you ask us to cite become more discoverable across the board.
We follow the structured-data and disambiguation guidance search engines publish openly. (See, for example, Google's structured data guidelines.)
It can be — which is exactly why we don't let a model invent your story. You supply the ground truth. We use automation to structure and publish those facts, not to fabricate them. Every page is checked against a fixed schema, and nothing goes live until you approve it.
The result isn't speculative AI prose; it's your verified facts, made legible to machines. Human-in-the-loop by default.
We publish accurate, authorized facts in the standard formats search and AI systems already recommend. That's the opposite of manipulation. We prohibit deception, impersonation, fake reviews, sock-puppets, and unlawful promotion. You warrant that what you submit is true, authorized, and not misleading.
There's a real line between accurate self-representation and manipulating results, and we stay firmly on the accurate side. We also align with the terms of the search and content platforms we use.
Yes — several regulated professions face advertising and self-promotion restrictions (medical boards, bar associations, election rules, financial-promotion regulations). Lodestar publishes factual, non-promotional information, but you remain responsible for compliance with your profession's rules. Our Concierge tier includes review support for sensitive cases. When in doubt, check with your compliance counsel.
One canonical profile — your source of truth — plus a small, tasteful scattering of corroborating references that link back to it: an indexed blog post, one or two longer articles, a community Q&A, and a social post. Notable clients may warrant a knowledge-base entry (e.g. Wikidata). Every item is accurate and points to the same canonical record. See a sample profile →
Because the part that actually moves AI retrieval is procedural — structured data, a clean link graph, fast pages, patient verification. That work used to take a specialist days; automation makes it minutes. We're infrastructure, not an agency, so we charge like software. See pricing →
Lightweight by design: primarily email verification, plus optional LinkedIn validation. For sensitive or high-visibility cases we may ask for government ID. We never publish anything you haven't authorized.
Cancel anytime from your dashboard. When a subscription ends we stop maintaining, monitoring, and re-testing, and your canonical profile is retired on a clear schedule. You keep the facts — they're yours — and you can export them. We don't hold your information hostage.
Start with the free check, or read our trust & policy details.