PRODUCT

Code checking AI. Not another model grading the answer.

Emet shows you what your agent did, and whether what it said is true — it asks your data the same question and re-derives every number in code, with the math.

QUERY AUDIT / Q2 ROAS

AI ANSWER3.46x
YOUR DATA2.43x
Q2 · Apr 1 – Jun 30CORRECTED
THE LOOP

The answer goes out. The check begins.

  1. Your system

    Your agent answers your user

  2. The finished turn is exported, after the answer is sent

    AFTER THE ANSWER IS SENT · TURN → EMET
  3. Emet / read-only verification

    Emet re-runs the query on your data (read-only)

  4. Every number re-derived in code

  5. Every sentence checked against the result

  6. Accuracy Report + Slack alert

Never in the request path

Read-only access you control

No model grades its own output

ANATOMY OF A CATCH

One answer. Five checks.

01 / 05

What your agent did.

Emet reads the finished turn your agent exports: every tool call, the arguments it was called with, and the raw result that came back.

EMET / TRACE AUDIT01 / 05
QUESTION"What was our Q2 return on ad spend on Meta?"
AI ANSWER"Q2 ROAS was 3.46x — our strongest quarter yet. Increase spend."
SPAN TREE / FINISHED TURN
turn / question → answer
model.0 / query planned
tool.call_1 get_meta_insights
model.1 / final output
TOOL SPAN · JSON
{ "span_id": "tool.call_1", "kind": "tool", "tool_name": "get_meta_insights", "payload": { "arguments": { "metric": "roas", "date_start": "2026-06-16", "date_end": "2026-06-22", "attribution": "7d_click_1d_view", "time_zone": "America/New_York", "currency": "USD" }, "result": { "roas": 3.46 } } }

The AI said strongest quarter yet. Your data said down 50%.

See a full Accuracy Report →
WHAT EMET CATCHES

The answer can look right. The source tells another story.

Query audit

Wrong date range

ROAS for one week reported as the quarter

Wrong attribution, time zone or currency

Last-click on a data-driven account; a date range off by one in the account's time zone; spend summed across currencies

Re-derivation

Wrong number

"Campaigns were ~26.8% of email volume." Actual: 98.7%.

Bad math

A period-over-period change computed against the wrong prior period

Claim check

Overgeneralized

"Email was over 96% of revenue in August and YTD." True for August only.

Invented cause

"The $13,867 gap is due to a 5-day attribution window." Nothing in the data says why.

Examples from a one-week client case study at a multi-brand retailer.

Interpretive claims — opinions and recommendations — are deliberately not scored.

EMET VS. LLM EVALUATION

A different root, not a better judge.

Eval tools check faithfulness to context. Emet checks truth against the source.

Checks the answer against

LLM evaluation tools: The context the model saw, or a test set you maintain

Emet: Your live source data, re-queried with its own read-only access

How

LLM evaluation tools: An LLM judge, or a scorer you write yourself

Emet: Deterministic re-derivation built in, per data source; an independent model checks the wording

Checks the query

LLM evaluation tools: Only if you write that check

Emet: Built in: dates, windows, time zones, currency, attribution

Output

LLM evaluation tools: A score per response or per test

Emet: A verdict and a reason per claim, with the math

Keep your tracing. Langfuse, Arize Phoenix, LangSmith and Logfire keep working, and Emet runs alongside them.

See integrations →
ONBOARDING

Mapped to your definitions.

Source-of-truth validation

At onboarding we confirm every data source is connected and Emet can query each one, so it knows exactly what it's checking against.

Your metrics

ROAS, pipeline, margin — as your business defines them, not a generic default.

Your calendar

Fiscal quarters, attribution windows, time zones and currencies, mapped once.

DEFINITIONS / ACME PAID MEDIA
roas = conversion_value (7-day click, 1-day view) ÷ spend fiscal_q2 = Apr 1 – Jun 30 account_tz = America/New_York currency = USD · daily fx

Scroll → to read the full definitions

WHERE VERDICTS GO

The right people get the receipt.

#emet-alerts
EmetAPP9:00 AM
CORRECTED Meta Ads ROAS — Q2 Paid Media analysis
AI said → Emet re-derived: 3.46x 2.43xFirst seen: Just now
claim"Q2 ROAS was 3.46x"re-derivedroas_q2 = 90,110.63 ÷ 37,084.09 = 2.43xwhythe AI queried Jun 16–22 instead of all of Q2 (Apr 1 – Jun 30)

Example Slack alert · actions shown for reference

Agent: paid-media-copilotSource: meta_ads_apiAlert: Emet CorrectionsReport: EVR-2026-Q2-001

Slack — one alert per issue, not per report

Emet Observability Dashboard — every report, plus accuracy, usage and cost by team and model

Export — CSV

See the Slack integration →

SECURITY / BY DESIGN

Security →

Read-only, scoped by you, revocable at once

Never in the request path — exports run after your user has the answer

Stores the trace, the claims, the verdicts and the query results behind them — never a copy of your tables.

FAQ

Questions worth asking.

Stop babysitting AI.
Start trusting it.

Emet delivers deterministic verification for enterprise AI data analysis — in seconds, with a receipt.