FINANCE

On-demand margin-erosion attribution deck for the board

Chat-triggered agent that investigates gross-margin erosion across BigQuery cohorts, decomposes the drivers iteratively, and assembles a narrated board-ready section in Notion.

CategoryFinance
Enginepaperclip
Difficultyadvanced
Triggerchat
Steps5
Setup~25 min

How it runs

The automated pipeline, trigger to output.

  • TriggerChat request with quarter and segment scope
  • ActionRun iterative BigQuery price/mix/cost queriesGoogle BigQueryBigQuery
  • LogicDecide if variance is explained or drill deeper
  • ActionCompose margin bridge and narrativeOpenAI
  • OutputWrite attribution section to Notion board pageNotionNotion

What it does

When a finance lead asks for a margin write-up, an agent queries BigQuery, decomposes the erosion into price, mix, and cost effects, and keeps drilling until the variance is explained. It then writes a structured, board-ready attribution section, including a clear bridge from last quarter's margin to this quarter's, into a Notion page.

When to use it

Use it ahead of board or QBR prep when you need more than a single query, an actual investigation that follows the numbers wherever the erosion leads and produces defensible narrative, not just a table.

How it works

  1. 1A chat request kicks off the agent with the quarter and segment in scope.
  2. 2The agent runs iterative BigQuery queries to isolate price, mix, and cost contributions.
  3. 3A reasoning step decides whether the variance is fully explained or needs another drill-down.
  4. 4OpenAI composes the margin bridge and narrative.
  5. 5The finished attribution section is written to a Notion board-prep page.

Set it up

What you configure once, before turning it on.

  1. 1
    Connect BigQueryDatasets, queries, schemas.
  2. 2
    Connect OpenAIModels, embeddings, files.
  3. 3
    Connect NotionPages, databases, comments.
  4. 4
    Set each agent's modelWe leave models unset so you pick the tier — fast + cheap, or top-quality.
  5. 5
    Tune it to your dataEdit the prompts, filters, and field mappings so it matches how your team works.
  6. 6
    Test, then turn it onRun once against a sample, confirm the output, then enable the trigger.

Run this workflow in your colony.

14-day trial. No DevOps. No Sales call. Provisioned in under a minute.