CHATBOTS

Slack pre-submit expense what-if bot

Employees describe a planned purchase in Slack and get an instant verdict on whether it will be reimbursed, the applicable cap.

CategoryChatbots
Enginepaperclip
Difficultyintermediate
Triggerchat
Steps5
Setup~15 min

How it runs

The automated pipeline, trigger to output.

  • TriggerEmployee messages the bot in SlackSlack
  • ActionLoad policy rules and caps from AirtableAirtableAirtable
  • ActionClassify spend and compute allowed amount with the modelOpenAI
  • LogicBranch: covered / capped / out-of-policy
  • OutputReply in-thread with verdict and required docsSlack

What it does

Gives employees a conversational way to sanity-check a purchase against company expense policy before they spend a dollar. They message the bot in plain language ("$280 dinner for 4 with a client in NYC"), and it replies with a clear reimbursable / partially reimbursable / not covered verdict, the per-head or category cap that applies, and the exact documentation they'll need at submission time.

When to use it

Deploy when employees keep asking finance "will this get approved?" in DMs, or when out-of-policy submissions are clogging the approval queue. It shifts the policy question to the moment of decision instead of the moment of reimbursement.

How it works

  1. 1An employee mentions the bot or DMs it in Slack with a free-text purchase description.
  2. 2The agent fetches the current expense-policy rules and category caps from an Airtable base.
  3. 3It reasons over the description against those rules to classify the spend and compute the allowed amount.
  4. 4A logic branch splits clean, capped, and out-of-policy outcomes into tailored guidance.
  5. 5The bot replies in-thread with the verdict, the cap, required receipts/approver, and a one-line policy citation.

Set it up

What you configure once, before turning it on.

  1. 1
    Connect SlackChannels, DMs, threads, mentions.
  2. 2
    Connect AirtableBases, tables, views, automations.
  3. 3
    Connect OpenAIModels, embeddings, files.
  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.