All work
B2B servicesPlovdiv, BG3 weeks build · 2 weeks tuned in production

€40k+ recovered in 90 days. No manual follow-ups written.

AI-powered overdue invoice recovery that writes follow-ups in the founder's voice and adapts to debtor behavior.

€40k+
recovered in 90 days
0
follow-ups manually written
31%
reduction in 60+ day aged receivables
1 click
to approve a draft from Slack
Challenge

The state we found.

A services company with €120k in active overdue receivables had stopped chasing anything under €5k. The founder wrote the heavy ones himself at 9pm. Junior staff were uncomfortable doing collections.

The result: 60+ day aged receivables grew quarter over quarter, and €40k of recoverable money was sitting in limbo.

Solution

What we built.

We built an AI agent that pulls each unpaid invoice with its full email thread history, drafts the next follow-up in the founder's voice, and adapts tone based on prior responses. Drafts go to a daily review queue — Slack approval sends them.

Architecture
// 6 steps · production
  1. 01
    Nightly sync of unpaid invoices from the accounting system into Airtable
  2. 02
    For each debtor, fetch the full email thread history via IMAP
  3. 03
    OpenAI generates a follow-up draft using the founder's prior writing samples as voice context
  4. 04
    Tone classifier on the previous reply (defensive, ignoring, partial-pay) selects the right escalation register
  5. 05
    Daily Slack digest with one-click approve / reject / edit
  6. 06
    Auto-pause on debtors who paid or replied in the last 48 hours
Stack
OpenAIn8nAirtableIMAP / SMTPSlack
Next case study

92% of support tickets resolved without a human.

Read next
Final word

Book the call. Bring the bottleneck.

30 minutes. No deck. We either see the automation in your process — or we tell you it is not worth building.