Our approach

AI where it helps.
People where it matters.

Speed up the work around the decision. Keep authority with the person making it.

01A clear line of reasoning

From evidence to understanding.

We build around a simple principle: a conclusion is only as useful as the evidence and reasoning behind it.

01

Establish the evidence.

Start with relevant sources. Preserve their context, dates, and limitations so an analyst can assess what belongs in the record.

02

Use AI to assist.

Apply AI to research and organization. Surface relevant information and potential gaps without treating generated text as verified fact.

03

Make the method visible.

Separate calculations from narrative. Make assumptions, formulas, and adjustments explicit so the analysis can be checked.

04

Keep review human.

Give professionals the context to accept, question, or reject the analysis. The final decision remains theirs.

Our commitment

Faster work should not mean
less accountable decisions.

02Grounded in operational reality

The constraints are
part of the problem.

Changing conditions

Context is not static.

Prices change. Inflation matters. Sources age. Comparisons need a clear account of what has changed and why.

Organizational requirements

No universal deployment.

Data access, licensing, security, and approval requirements must be understood for each organization and use case.

Professional review

The record matters.

Useful analysis must be explainable to someone beyond its author: a colleague, a supervisor, or a reviewer.

03Questions, answered

A few important distinctions.

Is Maitryon building another general-purpose chatbot?

No. Our focus is structured decision-support work: connecting source evidence, calculations, assumptions, and review. AI is part of that workflow, not a substitute for the workflow itself.

Does AI make or approve the final decision?

No. Our approach is to support professional judgment. People remain responsible for evaluating evidence and approving conclusions.

Is ODIN available for production use?

ODIN is in development. The workflow shown on this site illustrates our direction, not a released application. Contact us to discuss your use case and the current product stage.

How are security and deployment handled?

Deployment, access, and data-handling requirements need to be scoped with each organization. This public website is not a workspace for sensitive acquisition data. Please share only public, non-sensitive information in an inquiry.

See the direction

Put the principles in context.

Explore ODIN