MartinAI
Trusted domain intelligence

Trusted AI for utilities and energy teams

Answers about tariffs, programs, standards and your own data, with the source and a confidence score attached, running where your data is allowed to live.

  • Utilities and regulators
  • Energy program and customer teams
  • Consultancies with large document libraries
  • Organizations that cannot send data to public models

The problem this solves

General-purpose language models answer energy questions fluently and are often wrong on the details that matter: which tariff applies, what a program rule says this year, what a standard actually requires. A confident wrong answer about a rate or a regulation is worse than no answer.

MartinAI's trusted domain layer works from a curated, governed knowledge base rather than the open web. Every answer cites the documents it drew on and carries a confidence score, and the whole system can be deployed inside your environment so documents and data never leave it.

How it works

01

Curate

Your documents (tariffs, program rules, standards, procedures, reports) are ingested into a governed knowledge base with versioning and ownership.

02

Govern

A Knowledge Trust Framework defines which sources are authoritative, how conflicts resolve and when an answer must defer to a human.

03

Answer

Questions are answered from those sources with citations and a confidence score. Low-confidence answers say so instead of guessing.

04

Deploy

Run as a managed service or sovereignly inside your own infrastructure, with audit logs of every question and answer.

Capabilities

Provenance on every answer

Each response links the passages it relied on, so a reviewer can check the source in seconds.

Confidence scoring

Answers carry a confidence score and abstain when the knowledge base does not support a reliable answer.

Curated, versioned knowledge

Sources are approved and versioned; superseded tariffs and rules do not leak into current answers.

Sovereign deployment

Deployable in your environment for data residency, privacy and security requirements.

Works with your utility data

Combine document knowledge with validated bill and meter data to answer questions about specific accounts and facilities.

Audit logs

Every question, answer and cited source is logged for review and for improving the knowledge base.

Who uses it

Frequently asked questions

How is this different from a general chatbot?

It answers only from a curated, governed set of sources, cites them, and reports a confidence score. When the sources do not support an answer it says so. General chatbots draw on the open web and do not tell you where an answer came from.

Can it run inside our own infrastructure?

Yes. Sovereign deployment inside your environment is supported for organizations with data residency, privacy or security requirements, alongside a managed option.

What kinds of questions does it handle?

Questions about tariffs and rate rules, program eligibility and requirements, standards and procedures, internal policies, and, when connected to validated utility data, questions about specific accounts, facilities and periods.

How is accuracy maintained over time?

Sources are versioned and owned; new tariff or rule editions replace superseded ones. Audit logs of questions and answers show where the knowledge base needs additions, and low-confidence answers are routed for human review.

How do we evaluate it?

Book a demo. We load a sample of your documents, run your team's real questions against them, and show the answers with citations and confidence scores so you can judge the quality on your own material.

Related reading
More of the platform