What Janua Labs does
Janua Labs helps service businesses become easier for AI to find, more consistent from shift to shift, and easier to run. Three connected capabilities: AI visibility, an AI knowledge base, and data and AI implementation. Restaurants and hospitality are the founders' deepest operating experience; Janua also works with multi-location service brands, professional services, and home and local services. Everything rests on measurement rather than opinion.
AI Visibility
A free check at visibility.janualabs.com. The instant public scan runs the Foundation Check on a business's website: a pass or fail test of the requirements AI engines need before they can read it, naming anything not met, with every check shown, plus two separate grades, Classic Search and AI Readability. The free check reads the site; it does not test whether AI assistants name or recommend the business. The full audit, run by the founders, adds live, web-grounded queries to ChatGPT, Claude, Gemini, and Perplexity: real questions a customer would ask, with the answers stored. If an engine cannot be genuinely queried at audit time, it is omitted from that report rather than simulated or guessed.
The fix, inside AI Visibility
The human-led service that closes what the audit finds, quoted up front, followed by optional monthly work on the signals AI engines weigh, with the same audit re-run each month so the business watches its own results move. Details at janualabs.com/optimization.
AI Knowledge Base
Turning a business's policies, procedures, recipes, training guides and equipment manuals into one assistant every employee can ask. It answers only from information the business approved; when there is no approved answer, it says so and sends the question to the person who owns it. Details at janualabs.com/ai-knowledge-base.
Data and AI Implementation
Custom AI, data and workflow systems built to order around how a business actually runs, with human approvals and escalation rules visible, and analytics work scoped around the data a business has and the decisions it needs. Scope, integrations, deliverables and measurement are agreed during discovery. Work is quoted up front. Scoped in a conversation at janualabs.com/book.
Janua AI Front Desk is a worked example of this custom work: an assistant that answers website inquiries around the clock using only information the business approved, hands off to a person the moment it is unsure, and keeps a full record of who approved every answer. It runs on janualabs.com as the company's own receptionist. Live demo at intake.janualabs.com.
How the audit actually works
- Live, grounded queries. The full audit asks ChatGPT, Claude, Gemini, and Perplexity real customer questions with live web search enabled, matching what a person would see asking the same question, rather than scoring against a model's static training memory.
- No simulated results. An engine that cannot be genuinely queried at audit time is left out of that report. Nothing is represented with a stand-in or a guess.
- Full source traceability. Every finding traces to stored evidence, including the actual pages each AI engine read before answering, so a business sees exactly which sources shape its AI visibility, not just a score.
- Open scoring. Reports show every check that ran, the weight it carries, the points it earned, and the method behind the grades, on the page, not in a methodology PDF nobody reads.
- Competitor comparison from real data. Where engines name competing businesses, the comparison is built from the engines' actual answers and verified site data, never a fabricated competitor list.
- Two axes, never combined. Whether the engines can read a site, and whether they name the business, are reported separately. A site can pass every technical check and still be cited zero times; the engines cite documents that answer the question people ask.
Proof of method: a Western New York market, measured
Janua's first measured market, and the proving ground for its method, was personal-injury law in Western New York. Janua Labs maintains a verified roster of that market: 124 firms, each confirmed against its own website or bar registry (verification run August 2026; satellite and virtual offices excluded). The same 10 consumer questions (five phrasings, asked for Buffalo and for Western New York, none naming a firm) are put to all four engines with live search on: 40 real answers per run. Seven runs so far: 17 August, three on 31 August and three on 8 October 2026. In each run the engines named 24 to 37 of the 124 firms; 50 firms were named in at least one of the 280 answers and 74 were never named. One firm appeared in 23 to 34 of the 40 answers in every run. Runs on the same afternoon named 24 to 27 firms in August and 32 to 37 in October, which is the measurement's own noise band and why a single run is never quoted as a trend. The run-by-run table and the method are on the Western New York AI Visibility Study page, aggregate counts only. (Until 8 October this page said "44 of the 124": that was the count of every name the engines produced, including directories and firms outside the roster, not roster firms. Corrected.) The same method is now applied to restaurants and other service businesses.
Independent coverage. 716 Spotlight, the Western New York community site where Janua first tested this method, published a feature on the company and the measurement on 7 October 2026: "Ask AI Who to Call in the 716. A Buffalo Company Is Writing Down the Answers." The article discloses that relationship.
Why this matters: the cost of AI invisibility
Frequently asked questions
Which AI engines does Janua check?
The full audit runs live, web-grounded queries against ChatGPT, Claude, Gemini, and Perplexity. The instant public scan runs the Foundation Check and the two readability grades, and queries no engine; engine queries run in the full audit and the monthly market sweeps.
Does Janua Labs simulate results for engines it can't query?
No. If an engine cannot be genuinely queried at audit time, it is omitted from that report rather than represented with a simulated or guessed answer.
How does Janua Labs show its work?
Every finding traces to stored evidence: the exact questions asked, the answers given, the pages each engine read, and the per-check scoring math, visible in the report itself.
Is the audit one-time or ongoing?
The free check can be run any time, by anyone. Monthly monitoring, the same audit re-run each month so a business can watch its visibility move, is available and quoted up front.
Can Janua Labs guarantee an engine will name our business?
No, and no one honestly can; engine answers shift run to run. What Janua stands behind is the program: checks run, fixes shipped, logs shown.
Does Janua Labs only work with law firms?
No. Western New York personal-injury law was the first market Janua measured, used to prove its method. Janua works with restaurants and hospitality, multi-location service brands, professional services, and home and local services. Restaurants and hospitality are the founders' deepest operating experience.
What does the AI Knowledge Base answer from?
Only content the business has approved, such as its policies, manuals, training guides, and recipes. If a question isn't covered, it is routed to the right person instead of being guessed.