ORM Monitoring

Watch brand mentions, judge which are real reputational issues, draft responses - never post automatically.

Vertical: ATL · Status: ⬜ Planned Back to ATL · workflows index

---

Trigger

Continuous, per mention batch from the vendor feed. Same shape as feed enrichment.

Flow

mention feed (Meltwater | LocoBuzz)
        |
enrich + classify                  [ORM Agent - proposed]
        |
severity + category
  (service complaint | factual error | market commentary | spam)
        |
draft response OR mark do-not-respond
        |
human review                       <- always
        |
[human] publish reply

Why sentiment scoring is not enough

"IIFL share price falls" and "IIFL mis-sold me a policy" score alike on a sentiment API and need completely different handling - one is market commentary to monitor, the other is a service complaint with an SLA.

That judgement is exactly what the ORM Agent is for, and its verdict shape is deliberately the Compliance Agent's: findings, severity, evidence, cited rule.

Agents involved

ORM Agent (proposed). Architecturally a sibling of the Compliance Agent.

Blocked on

The Meltwater or LocoBuzz feed. That integration is the whole cost - the agent shape is already proven elsewhere in the product.

Also needs an escalation matrix and a legal sign-off path for public replies.

Roadmap source

Deliverable ⑥ (page 5); ORM -> Meltwater | LocoBuzz (page 3); ORM -> own (page 4) — see roadmap_ref_extracted.md.