FINANCE
AI Policy Reviewer that Reads Expense Justifications and Tiers Approvals
On submission, an AI agent reads each expense's free-text justification against your policy document, decides whether it is compliant, borderline, or a violation.
How it runs
The automated pipeline, trigger to output.
- TriggerExpense report submitted (webhook)HTTP webhook
- ActionLoad policy doc and approver hierarchySnowflake
- ActionAI reviews justifications, assigns tierOpenAI
- LogicRoute by worst-case tier: clear / borderline / violation
- ActionSend tiered sign-off request in SlackSlack
- OutputWrite decision and AI rationale to SnowflakeSnowflake
What it does
Most policy checks only compare numbers. This one reads the words. An AI reviewer evaluates each line's written justification against your actual expense policy — catching vague reasons, mislabeled categories, and personal charges dressed up as business — then assigns a compliance tier and routes accordingly so managers only see the cases that need judgment.
When to use it
Use this when your policy hinges on intent and context (client entertainment, travel exceptions, equipment) rather than hard dollar caps, and reviewers waste time reading justifications that are obviously fine. It scales nuanced review without hiring more approvers.
How it works
- 1An expense-report-submitted webhook delivers line items with their free-text justifications.
- 2The policy document and submitter's manager are loaded from Snowflake.
- 3An OpenAI agent reviews each justification against the policy and returns a tier with reasoning per line.
- 4A logic step routes by worst-case tier: compliant, borderline, or violation.
- 5Compliant reports are auto-cleared; borderline ones go to the direct manager in Slack, violations escalate to the finance approver.
- 6The decision and the AI's rationale are written back to Snowflake for audit.
Set it up
What you configure once, before turning it on.
- 1Connect HTTP webhookTrigger any URL on agent actions.
- 2Connect SnowflakeWarehouses, queries, shares.
- 3Connect OpenAIModels, embeddings, files.
- 4Connect SlackChannels, DMs, threads, mentions.
- 5Set each agent's modelWe leave models unset so you pick the tier — fast + cheap, or top-quality.
- 6Tune it to your dataEdit the prompts, filters, and field mappings so it matches how your team works.
- 7Test, then turn it onRun once against a sample, confirm the output, then enable the trigger.
More Finance workflows
Month-End Uncoded Spend Chaser via Snowflake to Slack
On a month-end schedule, queries Snowflake for unaccrued or uncoded expenses, groups them by department owner.
Receipt Upload OCR Policy Check with Manager Escalation
When an employee drops a receipt into a Drive folder, it extracts the line items, checks them against expense policy.
Weekly Proration Anomaly Audit to Notion
Each week it aggregates all flagged proration discrepancies from Snowflake, scores them against anomaly thresholds, and publishes a finance-ready audit page in Notion with totals.
Accrual Chase Board in Monday with Per-Owner Tasks
On a schedule it reads open uncoded expenses from Snowflake and creates or updates a Monday item per department owner.
Detect Mid-Cycle Plan Change Mischarges and Queue Credit Memos
Listens for Stripe subscription plan changes, recomputes the correct prorated amount.
Draft and Send Proration Over-Billing Correction Emails
For each confirmed over-billing credit, an agent drafts a clear, customer-specific apology email explaining the proration error and the credit applied.
Run it inside a business
This workflow drops into a full company template. Import the org, and this is one of the playbooks its agents run.
Run this workflow in your colony.
14-day trial. No DevOps. No Sales call. Provisioned in under a minute.