How it works.
MachineContext is a deterministic pipeline that transforms noisy company signals into verified, cacheable identity objects—and powers semantic search, similarity mapping, and change detection on top.
Architecture
Three stages. Zero guessing.
Resolution
Accept any input: company names, domains, typos, aliases. Normalize and attempt canonical resolution. If multiple candidates exist with similar confidence, return AMBIGUOUS with ranked options.
Extraction
Polite, controlled crawls acquire structured signals from authoritative pages. We prioritize identity signals (Schema.org, Open Graph, legal disclosures) over raw content. Redirect chains are traced. DNS records are verified.
Verification
Confidence is computed from signal multiplicity and consistency. Conflicting signals reduce confidence. Missing signals are explicit gaps, not assumptions. Every output includes provenance and freshness timestamps.
API Surface
Five endpoints. One truth layer.
Every API is powered by the same verified data pipeline. Different interfaces for different problems.
/resolve
Noisy input → canonical entity. Returns domain, confidence, aliases, or AMBIGUOUS with candidates.
/brand/:id
Full verified object for a known entity. Domain, legal name, description, aliases, confidence, freshness.
/search
Semantic discovery by intent. "AI code review tools" returns ranked companies based on self-description.
/similar
Find competitors and alternatives. "Companies like Stripe" returns Adyen, Square, Checkout.com.
/changes
Detect rebrands, domain migrations, and acquisitions. Subscribe to a watchlist, get notified when things change.
Confidence Scoring
Confidence isn't a feeling. It's computed.
Computed from signal multiplicity. More signals = higher confidence. Conflicting signals = reduced. Missing signals = explicit uncertainty.
Guarantees
Security
We ingest only publicly available information. No private data. No authenticated sources. No PII.
Requests are isolated. No session state. Customer data is never shared or co-mingled.
Infrastructure follows SOC 2 Type II controls. Audit logs for all access. Encryption at rest and in transit.
Per-key rate limits prevent abuse. Burst capacity for legitimate spikes. Graceful degradation.
Build on ground truth, not inference.
MachineContext is the infrastructure layer that makes AI decisions auditable and safe.