01
Perplexity
Browser-driven web research with citations.
Days of analyst time, in an afternoon.
CASCADE is our research orchestrator. It pulls from five sources in parallel — public web, paid LLMs, deep-research browsers, our own search infrastructure — and stitches the answers into something citable. The same engine sits behind every Signal Audit.
What CASCADE actually runs
Five sources, one report.
Single-LLM research is a guess with a confident voice. CASCADE triangulates across very different sources, then reconciles the contradictions instead of hiding them.
02
Gemini API
Long-context synthesis across many sources.
03
Gemini DR
Deep-research mode for analyst-grade pulls.
04
DeepSeek
Reasoning model for cross-source contradiction.
05
SearXNG
Self-hosted federated search — no rate limits, no algorithmic spam.
What we have produced with it
Real artifacts, not chat transcripts.
AI vendor teardowns
Forty-vendor scorecards across integration surface, data posture, and total cost of ownership. Used inside audits and published in the research slate.
Competitive intelligence reports
Recurring local-market intel for owner-operators — what competitors are running, what the market is paying, what is changing.
Workflow surveys
What 100 SMB operators actually use AI for — by workflow, by tool, by outcome — with anonymized direct quotes.
Signal Audit research depth
Every audit pulls workflow context, tool comparison, and governance risk through CASCADE before recommending a build.
The same pattern, pointed inward
Q&A over your own documents.
When the team needs repeatable answers from internal SOPs, policies, or call transcripts, we point the same orchestrator at the firm's knowledge base instead of the open web. Same engine, your data, your guardrails.
Internal KBPolicy docsCall transcriptsEngagement notesPermissioned by roleCited answers
Read the research before you hire us.
The published slate is built with the same engine your audit will run on.