INVOICE PROCESSING

Investigate unrecognized charges and propose a vendor identity

An agent-driven workflow that takes a Stripe charge no vendor registry could match, researches the merchant descriptor on the web, proposes the likely vendor and category.

CategoryInvoice Processing
Enginepaperclip
Difficultyadvanced
Triggerwebhook
Steps5
Setup~25 min

How it runs

The automated pipeline, trigger to output.

  • TriggerUnmatched charge received via webhookHTTP webhook
  • ActionSearch web for merchant behind descriptorExa
  • LogicReason and propose vendor identity
  • ActionWrite pending vendor row to SnowflakeSnowflakeSnowflake
  • OutputPost proposal to Slack for approvalSlack

What it does

When a charge cannot be matched automatically, this workflow does the detective work a finance analyst would: it interprets the cryptic billing descriptor, searches for what company it belongs to, and proposes a structured vendor record so the registry self-improves over time.

When to use it

Use it when your auto-match rate stalls because descriptors like "DRI*SVC8842" are unreadable. Instead of manually googling each one, the agent drafts an identity and a human just confirms.

How it works

  1. 1An unmatched charge from the reconciliation pipeline triggers the workflow via webhook.
  2. 2The agent normalizes the raw descriptor and searches the web for the underlying merchant.
  3. 3It reasons over the results to propose a vendor name, category, and likely subscription type.
  4. 4A draft vendor registry row is written to Snowflake in a pending state.
  5. 5The proposal, with confidence and sources, is posted to Slack for an analyst to approve or correct.
  6. 6On approval, the pending row is promoted so future charges from the merchant auto-clear.

Set it up

What you configure once, before turning it on.

  1. 1
    Connect HTTP webhookTrigger any URL on agent actions.
  2. 2
    Connect ExaNeural search across the web.
  3. 3
    Connect SnowflakeWarehouses, queries, shares.
  4. 4
    Connect SlackChannels, DMs, threads, mentions.
  5. 5
    Set each agent's modelWe leave models unset so you pick the tier — fast + cheap, or top-quality.
  6. 6
    Tune it to your dataEdit the prompts, filters, and field mappings so it matches how your team works.
  7. 7
    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.