StackIndex™
StackIndex™ is the intelligence engine behind every score, signal and classification on Instawhat.ai. It scores AI tools across 4 evidence-backed layers, and classifies every AI news story with a credibility label — so you can trust what you read.
How AI tools are scored
Every tool is scored across 4 intelligence layers. The weights reflect what actually matters in a real business context — not marketing benchmarks.
How AI news is classified
Every news article is classified by the Bureau's Flash agent before it appears on the site. Flash fetches the full article, finds corroborating sources, and assigns one of five labels.
9 agents. Insta directs. 8 specialists execute.
Each specialist agent has a defined scope and writes only to its own output contract. Insta directs the operation and synthesises all agent output into the final bureau notes.
Director of the Bureau. Synthesises all agent output into final bureau notes and directs the intelligence operation.
Continuously scans the AI ecosystem for new tools, monitors product updates, and seeds the evidence database.
Reads real user reviews, evaluates workflow utility, and assesses how tools perform in day-to-day business use.
Digs into APIs, GitHub activity, security certifications and infrastructure maturity.
Monitors AI news sources around the clock, detects hype, and assigns credibility labels to every story.
Applies the 4-layer StackScore™ formula and computes a final, confidence-weighted score for every tool.
Adds historical perspective — tracking score trajectories and flagging whether a tool is rising or falling.
Benchmarks each tool against its direct competitors, surfacing where it leads and where it lags.
The integrity guardian. Audits every score run for drift, affiliate bias and violations before anything goes live.
Built to resist manipulation
Scores are only useful if they can't be gamed. Every structural safeguard below is enforced in the pipeline, not just in policy.
Affiliate status cannot influence a StackScore™. Pulse enforces this on every single run.
Flash flags promotional language inflation. Low signal-to-hype triggers a confidence penalty.
Scores below 0.60 confidence require human review before they go live.
A change of more than 15 points in one cycle triggers a Pulse audit before publishing.
Every score is backed by traceable sources. Quality is weighted, not averaged.
"The AI economy doesn't need more noise.
It needs better signal."
Instawhat.ai · StackIndex™ · Bureau of AI Intelligence