Tanushree PanigrahyEnterprise AI · Practice & perspective

The question

What happens when a solution brief is examined by the people who would need to evaluate it, rather than only the audience its positioning is written to persuade?

Why I built it

Enterprise solution work taught me how much domain knowledge is carried in the questions people ask. I built AVIS — Adversarial Vertical Intelligence System — to explore that gap between a broad claim and the evidence a specific evaluator needs.

The approach

AVIS runs locally and uses healthcare buyer personas, including a CISO, a Clinical Informaticist, and a CIO, to interrogate public solution material through multiple turns. The experiment focuses on whether a brief supplies a specific answer or shifts back toward general positioning.

What I looked for

The evaluation examines patterns such as attributing a platform capability to a particular solution, answering a concrete question with configurable-platform language, or substituting a future demonstration for present evidence.

What I took from it

The useful output is a set of questions and evidence gaps to investigate further. It reinforces the value of starting from the evaluator’s context when preparing a solution story.

Limits

This is an exploratory evaluation of public material, not an audit of a vendor or proof that a capability is absent. A missing answer in marketing content cannot establish a product limitation. Model-generated judgments need human review; no benchmark accuracy or production impact is claimed.

← Back to work Talk about this project ↗