Skip to content

The Venturi dashboard

app.venturi.systems is where your attributed AI estate becomes something people act on. Every invocation Venturi sees is resolved into the attribution graph, scored for confidence, and materialized into fast, filterable surfaces. The dashboard is how your finance, engineering, security, and leadership teams read those numbers, drill into the evidence behind them, resolve disputes, and turn attribution into chargeback, savings, and adoption decisions.

This section is the guide to the product surfaces. It explains how the app is organized, how to read what each screen shows you, and how to do the things you came to do — defend a chargeback, close a coverage gap, accept an optimization, or hand finance an export they can reproduce without us.

What the dashboard is — and isn't

  • It is a read-and-decide surface over the materialized attribution index. Tables load fast, every number names its method and freshness, and every result links to the evidence that produced it.
  • It never blocks your AI traffic. The dashboard reads a pre-computed index; it is completely off the gateway hot path. Nothing you do in the app can affect production inference.
  • It changes attribution truth only through an explicit, audited override. Comments, assignments, and disputes are metadata about an attribution — they never silently rewrite it.

How the app is organized

A persistent left navigation groups every destination by the job you're trying to do. You see only the destinations your role entitles you to — unentitled items are absent, not greyed out.

Group What you do here Key pages
Monitor See spend, adoption, and usage across the estate Overview, Teams, Services, Models, Invocation feed
Resolve Close coverage gaps, triage unknowns, settle disputes Coverage, Unknown spend, Attribution detail, Disputes
Optimize Act on savings, adoption, and budget governance Optimization, Adoption, Governance, Alerts
Operate Export, bill, and check trust and system health Exports, Billing, Trust center, Support, Status
Admin Manage users, roles, security, credentials, audit Users, Roles, Security, Credentials, Audit log

A global command palette (keyboard-invoked) jumps you straight to any entitled page, team, or attribution record — "open attribution evt_…", "go to team Payments", "start export". A global search bar resolves any of the five graph node types, plus models, teams, connectors, policies, disputes, and saved views, returning typed results with deep links and never crossing your tenant boundary.

Start here

  • Tour the surfaces

    Learn the screen layout, the time-basis and freshness model, saved views, and how to read any table on the platform.

    Dashboard overview

  • Defend a number

    Open the evidence drawer behind any attribution, file a dispute, and propose an audited override.

    Attribution detail

  • Close the gaps

    Read your coverage score, work the Unknown Spend Analyzer, and turn honest unknowns into resolved ownership.

    Coverage & unknowns

  • Save and enable

    Review optimization recommendations and read cohort adoption intelligence.

    Optimization & adoption

  • Govern and export

    Set budgets and policies, build a FOCUS chargeback export, and see your billing basis.

    Governance & exports

Principles the whole app honors

  • Every result carries its interpretation metadata. Wherever you see an attribution — a tile, a table row, a drill-down leaf — it carries its stage origin, output state, coper confidence and band, evidence basis, model version where inference was involved, degradation state, freshness, and export eligibility. See confidence and evidence.
  • Unknown is shown, never hidden. Spend Venturi can see but cannot responsibly assign renders as an explicit unknown — an em-dash, never a fabricated $0.00.
  • Method and freshness are always labeled. A figure is marked estimated or reconciled, and stamped with the time it was current as of. Two tiles that name "total AI spend" for the same scope always reconcile to the same figure.
  • Access fails closed. Authentication, role checks, exports, billing, and residency routing all fail closed. Only your AI gateway traffic is fail-open.
  • No prompt or completion content is ever shown — because none is ever stored.

Who uses which surface

If you are… You mostly live in…
FinOps / Finance Overview, Governance, Exports, Billing
Platform / VP Engineering Teams, Services, Models, Coverage, Optimization
An engineer resolving a number Attribution detail, Disputes, Unknown spend
CTO / CHRO / People leadership Adoption, Coverage, scheduled reports
Security / Audit Trust center, Audit log, Exports
Tenant Admin Admin (users, roles, security, credentials, audit)

You can complete your role's core task without an Venturi engineer interpreting it for you — that is the bar every surface is built to.

Where to go next