← Work

AI Agents · Reputation

Review & Reply Agent

Status

Live

Year

2026

n8nClaude APIEscalation Rules
Source Repo →

Reads a customer review, escalates anything serious to the owner, and drafts the rest for one-tap approval. It never posts on its own — that constraint is the product, not a limitation. Verified across three live executions averaging 3.48 seconds: a five-star queued for approval, a two-star combining food poisoning with a legal threat escalated, a one-star about a child's illness escalated.

The film

Try it

A browser-side simulation of the routing logic — pick an input and watch which rule fires. The logged runs below are the receipts.

Simulation — same rules, sample inputs

Pick a sample. The inputs are fictional; the routing rules are the ones the live agent enforces in code, and the live agent averages 3.48s per run.

Escalation triggers — enforced in code, every review

  • Rating of 3 or below
  • Alleges illness, injury, a safety or hygiene problem, discrimination, or staff misconduct
  • Threatens legal action, a chargeback, or contacting press or a regulator
  • Names a member of staff negatively
  • Reads as fake, a competitor, or a scam
  • The model is unsure, or its output is unparseable — it fails closed

Logged runs

Each replay below animates a logged n8n execution at the speed it actually ran — the decision path, the output, and the latency come straight from the execution log. Nothing staged.

Replay of logged execution #25 — 3.53s actual

Rating

5/5

Sentiment

positive

Topics

food quality, sisig, staff hospitality, parking

n8n execution log0.00s…

Decision · Queue for Approval

Straightforward 5-star praise with no complaints or concerns.

Draft reply

Thanks so much, Marites. We're proud of our sisig—that's exactly the kind of compliment that means everything to us. Glad the team made you feel welcome to linger and enjoy it.