Run business WhatsApp customer messages with an agent

Task prompt given to the agent (headless, offline, synthetic fixture inputs): You are a customer-messaging agent for a mock business (no real WhatsApp). Process each message in inbox.jsonl using only facts from kb.md and availability in slots.json. Write replies.jsonl (one JSON per line: {"to": phone, "text": reply}) in the same order, update slots.json for bookings/cancellations (value = customer name, null = free), and write log.csv with header phone,name,intent,action for every message. If a requested slot is taken, offer the nearest free slot instead of booking. Inspired by https://agentusecases.ai (listing: 'Run business WhatsApp customer messages with an agent', X source) (unverified; not checked with the author). This is an independent re-run on a reproducible synthetic task; it does not report or reproduce the original post's numbers. Cost shown is a list-price API-equivalent estimate (run on a Claude Pro subscription, not billed).

Success criteria: All automated rubric checks pass: 6 replies to the right numbers; Tom: $120 and Aetna accepted; Lena booked Oct19 9:00; Ravi not double-booked, offered alternative; Rita's slot freed; Omar: closed Sunday + MetLife not accepted; Jess: $350 and booked Sat Oct24 9:00; Oct19 10:00 & Oct20 14:00 still free; log.csv header + 6 rows.

Best model for this task so far: DeepSeek V4.1 Flash (Venice) (success, $0.0029, 12s). Best AI model for Sales, Marketing & Support →

Run this on another model

All runs (7)

86% pass rate across 7 eligible runs: 6 pass, 1 partial, 0 fail. Partial and failed runs are listed below with everything else.ⓘ