Data & Analytics

Ad-hoc questions answered at the source, with the query attached

Data teams don't have a modelling problem. They have an interruption problem. Kentron takes the questions that never needed a person, like the pipeline that failed at 3am, the metric that moved, and the one-off pull, and handles them where the data lives. The query it ran comes attached to every answer.

What this replaces

Being paged for a pipeline failure whose fix is always the same three steps.

Book DemoSee the receipt
Ines Duarte6:58 AM

@Receipt orders_fact is stale again. What happened, and can we backfill?

ReceiptApp7:00 AM

An upstream schema change made orders.discount_cents nullable at 02:14, and the nightly load failed on the not-null constraint. Four downstream dashboards are stale and their owners are notified. A backfill statement is drafted and one approval away.

Receiptrcp_0b64

Three jobs Data & Analytics stops doing by hand

Each one is a single run across the systems you already use, with a policy gate before anything changes and a receipt for every call.

  1. 01

    Diagnose the failed pipeline before standup

    You ask

    @Receipt the nightly load failed. What broke, what is stale, and who needs to know?

    • Snowflake
    • Databricks
    • Looker
    • Slack

    What Kentron does

    1. Reads the failure, isolates the failing step, and finds the upstream change behind it
    2. Traces lineage to every dashboard and model now serving stale data
    3. Tells each downstream owner what is stale and until when
    4. Drafts the backfill and holds it behind approval before it touches the warehouse

    What it leaves behind

    The diagnosis quotes the exact schema diff and the failing statement, so the fix is reviewed, not trusted.

  2. 02

    Explain the metric that moved

    You ask

    @Receipt activation dropped 8% week over week. Is it real?

    • Snowflake
    • Google Analytics
    • Metabase
    • Notion

    What Kentron does

    1. Checks instrumentation first, since tracking changes explain more drops than product does
    2. Segments the movement by platform, cohort, geography, and acquisition source
    3. Works out whether the drop is concentrated or broad, and names the segment carrying it
    4. Writes up the finding with the queries attached, so the next analyst starts from your work

    What it leaves behind

    Both the queries and their results are in the receipt, which makes the analysis reproducible instead of remembered.

  3. 03

    Answer the ad-hoc pull without joining the queue

    You ask

    @Receipt give me signups by plan and region for Q1, as a sheet.

    • Snowflake
    • BigQuery
    • Google Sheets
    • Slack

    What Kentron does

    1. Resolves the request against the governed semantic layer, not a guessed table
    2. Writes and runs the query, then sanity-checks the totals against a known-good metric
    3. Delivers a formatted sheet with definitions noted beside each column
    4. Hands it to a human analyst when the request is genuinely ambiguous

    What it leaves behind

    The requester gets the SQL alongside the numbers, which stops the same question coming back next quarter.

For Data, the proof is the query

Kentron stores every query it runs and the result it got in a hash-chained receipt. An answer stops being a number someone remembers and becomes something anyone can rerun.

  • Every answer ships with the SQL that produced it
  • Backfills and other writes wait for a named approval
  • Diagnoses quote the exact schema diff and the failing statement

Receipt

orders_fact nightly load

rcp_0b64

  1. Read: Snowflake

    b13e…7a40

    Read the failed load and its error

  2. Read: Snowflake

    5c8f…e2d1

    Compared the orders schema before and after 02:14

  3. Read: Looker

    e07a…19b6

    Traced lineage to four stale dashboards

  4. Wrote: Slack

    94d2…c3f8

    Notified the four dashboard owners

  5. Held for approval: Snowflake

    2a6b…50ed

    Backfill statement held for approval

5 calls, chain verified

The tools Data & Analytics usually connects

Any tool you use can be connected. These are where this team tends to start.

  • Snowflake
  • Databricks
  • BigQuery
  • Looker
  • Metabase
  • Google Analytics
  • Notion
  • Slack

Use cases for other teams

Put Kentron on the Data & Analytics backlog

Connect one system, then read the receipts it writes before you connect the next.

Book Demo