Healthcare / financial services / applied AI
Vistra LLC is building multiple AI products for consequential work across healthcare and financial services, all using AI to make complex work clearer, more inspectable, and more useful.
Each product is designed around the rules, evidence, and safeguards of its domain—not a generic assistant searching for a use case.
Useful intelligence should leave a record. The output should be understandable, its limits visible, and the person using it firmly in control.
Healthcare operations
A deterministic, auditable prior authorization platform for health plan teams that need evidence-led decisions and reviewable handoffs.
Financial services intelligence
An AI-powered stock scanning and trading intelligence platform for researching market conditions, setups, and tracked strategy results.
Deterministic, auditable prior authorization
The language model is constrained to intake and extraction with citations. A deterministic rules engine applies relevant public CMS LCD criteria. When the record is incomplete or uncertain, the platform does not guess.
Clinical documents become traceable facts. The source remains attached to every assertion.
Deterministic rules encode relevant public CMS LCD policy criteria and produce the decision state.
Ambiguity routes to human review with the evidence trail intact. It never becomes invented certainty.
Synthetic labeled golden set / offline reference path
The current offline reference-path evaluation produced 24 exact pipeline matches across 24 synthetic labeled cases. An exact match means the stored reference facts, citations, criterion statuses, and deterministic decision matched the expected record.
Failure taxonomy: schema error, citation error, missing fact, unexpected fact, value mismatch, criterion-status mismatch, decision mismatch, and runtime error. These offline results validate the deterministic reference path only; they do not measure live document extraction, clinical quality, or real-world coverage outcomes.
Stock scanning and trading intelligence
AI Newton brings market context, strategy-based signals, and tracked outcomes into one research workspace. It supports analysis and discovery; it does not connect to a brokerage or execute trades.
Summarizes market tone, risk context, participation, and notable setups for the selected trading day.
Surfaces Newton and pattern-based setups with entry, stop-loss, status, and position context.
Translates natural-language screening questions into structured stock discovery with transparent supporting metrics.
Places upcoming and recent earnings information alongside signals so event risk remains visible.
Tracks open and closed strategy positions, returns, and exposure for review—not as a promise of future results.
Public notes / decisions and limits
Implementation notes and public reflections will document the decisions, constraints, and revisions that shape Vistra's products. No LinkedIn post or public build-log destination has been published yet.
POST 01 LinkedIn build log coming soonHealthcare demonstration
The Payors platform uses synthetic data and public policy sources for demonstration and public policy discussion. It does not provide clinical advice, legal advice, or coverage determinations, and it does not replace qualified human review.
Financial information
AI Newton provides research and educational information, not investment advice or a recommendation to buy or sell securities. It is not a broker-dealer, does not execute trades, and does not provide automated trading. Markets involve risk; tracked results do not guarantee future performance.