Built to pass your security review.
This page shows what runs where, what crosses the boundary, and what never leaves. It answers the questions your architects and your CISO will ask, before they ask them.
The whole platform on one screen.
Solutions on top, ten products beneath them, the semantic plane they all share, and your estate at the base, with the control plane running through everything. Click around: this is how the pieces actually fit.
This is the whole platform on one screen. Click a solution to see which products serve it, a product to see what is inside it and what it trades with the record, or a plane layer to see what it holds.
One graph under everything.
Systems, datasets, columns, business terms, rules, controls, owners, and reports live in one connected record, updated continuously as your systems change. Ask any number where it came from, and the answer traces back through this graph.
Every product writes here, and every product reads it back.
Every product writes to this plane and reads from it. Pick a layer to see what it holds, then trace the loop: contributions flow in as labeled assertions, and the graph hands back obligations, contracts, context, and derived trust.
Business knowledge graph · Meaning as facts. Every binding is a labeled assertion with provenance and a validation state, and conflicting claims coexist with their history. The graph is a rebuildable projection; the authoritative record is the assertion log plus sealed decisions.
one record, ten lenses · nothing copied, everything traceable
machines contribute claims, people make decisions, and only the sealed record is ever the truth
One consequence worth naming: the graph you query is a projection. The record is the append-only assertion log and the sealed decisions behind it, and every projection, graph, search, or score, can be rebuilt from that record, provably identical.
Nine engines do the work.
The ten products are what your teams see. These are the engines underneath them. Here is what each one does and what it hands the rest of the platform.
Knowledge graph
It holds every system, dataset, column, term, rule, control, owner, and report as one connected, continuously updated record.
→ the single record every product reads and writes
Connectors & sync
It ships connectors for warehouses, databases, ETL, BI, files, and streams, push and pull, incremental and event-driven. There are no per-connector fees, so coverage is an engineering decision, not a budget one.
→ fresh metadata from your systems, on schedule or on event
Lineage parser
It reads your actual SQL, ETL, and BI code, reconstructs column-level lineage, and reconciles it against what teams declared.
→ impact and blast-radius answers in seconds
Rule & quality engine
It profiles data, runs rules, and checks freshness, volume, schema, and drift. Everything executes inside your systems, and only pass/fail counts come back.
→ quality results and SLO breaches, caught upstream
Anomaly detection
It watches the numbers between rule runs and flags breaks nobody wrote a rule for, before a report consumes them.
→ early warnings with the evidence attached
Policy & workflow engine
It handles approvals, attestations, maker-checker, separation of duties, and RBAC/ABAC. These are the controls your audit team already expects, enforced in software.
→ governed change with the sign-off trail built in
Evidence ledger
It seals every governed decision onto a hash-chained, append-only ledger. If it can't seal, it refuses the write.
→ exam-ready evidence packs and point-in-time reconstruction
Trust derivation
It recomputes certification and regulatory posture from live evidence every time someone looks. Nobody can hand-set a badge.
→ trust states you can defend in front of an examiner
AI & agent runtime
It runs retrieval and agents with confidence scores and human checkpoints. Every AI action is logged, attributed, and reversible. AI cannot approve.
→ drafts and rankings for humans to decide on
People on top, one platform beneath.
Read it the way the platform runs: the people it serves, the ten products they use, and the shared services every product stands on. Click one to open it, and see the full seven-layer stack on the home page.
Financial-industry ontology packs, led by BCBS 239 with Basel, LCR, IFRS, GAAP, SOX, and CCAR arriving, sit beneath your institution's ontology and an operational knowledge graph reconciled on every governed write.
From observation to proof, in five steps.
Every engine feeds the same loop. This is the path a governed change takes from first observation to evidence an examiner can verify.
Observe. Connectors parse code, profile datasets, and run checks where your data lives. Metadata flows up; rows never leave your systems.
Your rows never leave.
Your rows never leaving is architecture, not a promise in a contract clause. Watch what crosses the boundary, and flip the deployment toggle to see the whole platform move inside your perimeter.
As managed SaaS, only metadata, results, and pass/fail counts cross the boundary, encrypted and held under your keys; your rows never do. Flip the toggle to see the other deployment.
- Metadata flows up
- schemas, lineage, results, decisions. Rows stay in your systems.
- Pushdown execution
- quality rules run inside your databases; only pass/fail counts return
- Samples off by default
- when enabled, masked in the connector before anything moves
- Your keys
- BYOK with a customer-held crypto kill switch
- No phone-home
- signed images, published bill of materials, no telemetry callbacks
- Audit everything
- full access and action trail, exportable to your SIEM
Same product on either side of your perimeter.
Single sign-on and directory sync · security questionnaires returned in five business days · exit terms in the contract: your graph, rules, and evidence export in open formats, with 90 days of transition help.
Where our certifications stand.
SOC 2 Type II and ISO 27001 are in progress, not claimed. Available today under NDA, before any engagement: our latest penetration test summary, a shared responsibility matrix, our subprocessor list, and the connector access and egress model. Security questionnaires are answered within five business days.
How the seal works: every decision is hashed and chained to the one before it. Checkpoints are exportable to anchors you hold, so the chain verifies against a reference outside our control, and verification runs offline against the exported pack with no access to us. No role, ours or yours, can rewrite sealed history; a correction is a new sealed decision that points at the old one. And because the graph and every index are projections of that record, any of them can be burned down and rebuilt from it, provably identical: losing a view never loses the truth.
Start with proof, not a promise.
There are three ways in, and each one earns the next. You approve the exit criteria before we spend any more of your time or budget.
A 60-minute working session
Bring one report that is expensive to prove. We walk the mechanism against it. Nothing is required from your side.
A two-week guided sandbox
Your team works hands-on against a synthetic sample bank. It requires zero access to your systems and zero procurement.
A six-week proof on your systems
We use read-only metadata across five systems and one priority report, scored against criteria both sides sign first. The decision it produces is yours.