Medical-device RMA intelligence

The agent layer above your RMA queue.

AegisRMA reads every return the moment a unit lands — matching the serial against warranty and recall records, scoring the order for fraud patterns, and writing the disposition: refund, replacement, repair, or recall. Routine touchpoints stay in your QMS. The decision moves up here.

  • Sits on top of Trackmedium, QT9, RMAPortal, Salesforce, ReverseLogix
  • Constrained against Part 806, ISO 13485, 21 CFR Part 820
  • Same logic at 4 a.m. Saturday and 2 p.m. Tuesday

Capabilities

Three guardrails built into the routing.

The decision is constrained against the regulatory frame your returns live in. Not bolted on after the fact.

Compliance guardrails
Part 806, ISO 13485, and 21 CFR Part 820 are part of the decision.
Regulatory constraints are hard-coded into the routing logic — not added as a post-hoc reconciliation on the way to the audit trail.
  • Each disposition is logged with the inputs that produced it.
  • A constraint that fails the read is louder than a constraint that passes.
  • Risk class maps to disposition class — a higher-risk action cannot land in a lower-risk lane.
Recall watch
A flagged serial hard-routes. Refund and replace are not reachable.
Part 806 recall files are ingested as a first-class input. The moment a serial matches, the recall lane owns the decision.
  • Recall watch is a continuous ingestion process, not a periodic sweep.
  • Refund and replacement paths are unreachable from a matched recall read.
  • The recall-flagged unit is logged against the matching campaign file.
Human-in-the-loop approval
A flagged disposition does not close the loop without an approver.
Every flagged disposition — recall hit, fraud hold, policy conflict — is written to the approval queue and waits on a human reviewer of record.
  • Standard dispositions route through automatically.
  • Flagged dispositions appear in the pending-review queue for sign-off.
  • The audit trail records the approver, the time, and the rule that produced the hold.

Core flow

One read per unit. Five steps to a disposition.

Each return note moves through the same pipeline. The order matters — a recall match at step 3 arrives before a refund eligibility check at step 5.

  1. RMA intake
    The return note arrives.

    A unit lands in the queue. Serial, customer, channel, fault class are extracted.

  2. AI disposition
    The agent reads the return.

    Warranty window, fault history, and ordering patterns are scored against the current rules.

  3. Recall diff
    Part 806 files are diffed against the serial.

    A match hard-routes the unit to the recall lane; refund and replace lanes are unreachable from that path.

  4. Compliance check
    Regulatory constraints constrain the choice.

    ISO 13485, 21 CFR Part 820, and the channel-specific frame gate the lane a disposition can land in.

  5. Approval queue
    Flagged dispositions wait on a reviewer.

    Standard dispositions close the loop. Flagged dispositions — recall hits, fraud holds, policy conflicts — surface in the approval queue.

Cost of non-compliance

What misses the regulatory frame costs you.

Returns enter the queue faster than they leave it. Each unit the agent fails to catch on a Part 806 read, a warranty diff, or a fault-class score becomes a refund, an audit observation, or a recall — and the exposure lands in the same three places, every quarter.

Financial · HME distributors
Refund fraud and out-of-warranty leakage.
Returns arrive faster than the warranty and fault-history checks can score them. Refunds land before the diff runs, and the loss is gone before the next reconciliation cycle.
  • Refund-claim fraud: phantom faults, inflated order histories, repeat address patterns.
  • Out-of-warranty units admitted under goodwill — every goodwill refund over the audited baseline.
  • Channel-attribution miscoding that quietly inflates the refund reserve.

Anchor: refund & fraud loss as a % of returns revenue.

Audit · Endoscope OEMs
FDA 483 observations and ISO 13485 §8 findings.
A missed recall diff or an unscreened return becomes an inspection observation the next cycle. The auditor reads the audit trail first; the cost is the citation, not the inspector.
  • 21 CFR Part 820 §820.198 — complaint files that don't trace back to a disposition decision.
  • ISO 13485:2016 §8.2.3 — monitoring and measurement that isn't tied to the offending event.
  • Warning-letter and consent-decree exposure that originates from a return the queue never screened.

Anchor: 483 line items that trace back to RMA intake.

Recall · OEM & distributor
Recall coverage gaps on Part 806 serials.
A Part 806 file is only as good as the channels it reaches. Units that return through a different distributor, a refurbished channel, or a goodwill path can land before the diff runs — and the recall lane never opens.
  • Part 806 serials that return under a different channel — distributor or DME — than the originating campaign file.
  • Refurbished / third-party resold units that bypass the originating OEM's recall lane.
  • Remediation cost per missed unit — the field-correction math that closes after the recall window.

Anchor: coverage gap per campaign, per quarter.

The exposure compounds. A refund that gets past a recall diff is a 483 finding on the next inspection cycle. An audit observation that traces back to an unscreened return is the line a recall campaign never reached.

Internal reference

Competitor positioning and segment context sit with the operator team.

The comparison context — how this product lines up against Trackmedium, QT9, RMAPortal, Salesforce, and ReverseLogix — is held in the internal-reference area alongside the LinkedIn outreach template and the segment-channel briefs. Those pages are reachable from the site footer; they are not linked from the public surface.

Try it

Open a return, or scope the pricing.
Both end at the same agent — one through a single unit, one through the subscription details.