Evid

Ask your data. Get answers backed by evidence.

Governed AI analytics for BigQuery. Charts, explanations, and SQL grounded in your approved tables and business rules.

Postgres, MySQL, and Snowflake coming soon

Book a demoSee how it works
Revenue · Published

In action

Ask anything. See the evidence.

MoM change

−8.6%

  • Revenue dashboard
  • Refund analysis
  • Pricing change

Net revenue fell 8.6% month over month. Refund volume tripled after the new pricing rolled out.

Line chart of weekly net revenue versus the previous period; revenue diverges from week 3 as refunds spike after a pricing change.

Sample data

The problem

Text-to-SQL reaches your warehouse. It still guesses what revenue means.

Text-to-SQL will query your warehouse. It will not know which tables are approved or which revenue definition finance signed off on.

Same question ·

Ungoverned text-to-SQL

Undefended

Tables the model touched

Revenue

—

Waiting for query…

Evid · published context

Defensible

Published context applied

Net revenue

—

Applying published context…

How it works

Governance before conversation. Define the rules once.

01Publish context

Rules, caveats, and instructions. Draft until ready.

Context editorDraft
Net revenue excludes refunds
Caveat · campaign ended mid-month
Publish
02Control access

Approved tables only, on your own data source.

orders_fact
campaigns
tmp_refunds

BigQuery · Production

03Ask naturally

Users ask inside the published dashboard.

Why did revenue fall last month?

Evid found something

Net revenue −12.4%, mostly Enterprise after the campaign ended.

04Inspect the answer

Chart, SQL, and the context that shaped it.

AnswerChartSQLContext
SELECT month, SUM(net_amount)
FROM analytics.orders_fact
WHERE is_refunded = FALSE
GROUP BY 1

Security

Your data never leaves your warehouse.

DATA BOUNDARYSECURE
  • 01Data stays in your warehouseZERO COPY
  • 02Only approved tables are queriedALLOWLIST
  • 03Credentials encrypted at restAES-256
  • 04Every query is inspectableFULL AUDIT
  1. 01

    Your question · Plain English

  2. 02

    Evid · Agent, no data stored

  3. 03

    Published context · Rules · approved tables

  4. 04

    Your warehouse · BigQuery today

CONNECTORSBigQueryLIVEPostgreSQLSOONMySQLSOONSnowflakeSOON

Compare

Raw text-to-SQL vs Evid.

CapabilityRaw-schema text-to-SQLEvid
Query scopeWhole schemaOne published dashboard
Business definitionsLeft to the promptPublished before query time
Table accessWhatever the schema exposesApproved list only
PublishingNoneDraft, then publish
EvidenceQuery and resultAnswer, chart, SQL, context

Deployment

Choose how you run Evid.

Your infrastructure

Assisted

We set Evid up in your environment and hand it over.

  • Runs in your cloud
  • First dashboards published
  • Handover included
Book a demo

Evid Cloud

Available

We host and run Evid for you.

  • Managed hosting
  • Separate data source per dashboard
  • Onboarding with our team
Book a demo

Self-host

Docker soon

Run Evid yourself under the Elastic License 2.0.

  • Your Supabase, warehouse, and model keys
  • Unlimited dashboards
Get notified

FAQ

Questions worth answering upfront.

No. Queries run in your own BigQuery project. Evid stores chats, dashboard context, and encrypted credentials, never copies of your tables.

Your data already knows the answer.

See it on your own data.

Book a demo