ResolveIQ
Incident Intelligence

AI-Assisted Case Resolution

The resolution platform that shows its work.
Evidence in. Decision yours.

ResolveIQ turns fragmented case information into an evidence-backed resolution plan, shows exactly how it reached its recommendation, and keeps a human in control of the final call. Live on AWS, with synthetic data.

No account required · Sample data only · Choose a role to begin

  • Live on AWS
  • Synthetic demonstration data
  • Human approval required
  • Source-grounded
  • No autonomous customer actions
resolveiq · Case RX-4821

Passenger stranded after a canceled international connection

RX-4821 · SEV2 · departure window closing

Deadline at risk

Submitted through Customer Mobile App · 18m ago · Booking ZX42DP

Why this recommendation Support: Strong

Offer hotel accommodation and rebook on the earliest available departure — 6-hour delay, passenger away from home.

Supported by§ Disruption Recovery Policy 4.2§ Incident RI-1048§ Contact history
ApproveModifyEscalate Human decides

How ResolveIQ works

From a messy inbox to a decision you can defend.

Six steps, one continuous thread — and a person makes the last call.

  1. 01

    Case arrives

    A customer message or operational alert enters with its full context.

  2. 02

    Context is organized

    Scattered details become a clear picture of impact and urgency.

  3. 03

    Risk is identified

    Severity and time-sensitivity, assessed in plain language.

  4. 04

    Evidence is retrieved

    Relevant policies and similar past incidents are pulled in.

  5. 05

    Resolution is recommended

    A next action, with the evidence attached to it.

  6. 06

    A human decides

    Approve, edit, escalate, or reject — the person owns the outcome.

One case, start to finish

Passenger stranded after a canceled international connection.

Departure window closing · multiple prior contacts · limited rebooking options · time-sensitive policy obligations. Here is what ResolveIQ changes.

Before ResolveIQ

  • Unstructured customer message
  • Scattered travel details
  • Previous support notes
  • Time pressure, no clock
  • Unclear policy path

After ResolveIQ

  • Situation summary
  • High-severity explanation
  • Applicable policy, cited
  • Similar resolved incident
  • Missing information flagged
  • Recommended action + evidence
  • Human approval controls

Shows its work

Every recommendation carries its evidence.

ResolveIQ never asks you to trust a black box. Each factor behind a recommendation is stated plainly and linked to the source that supports it.

Policy

What the rules require

Disruption recovery obligates accommodation for delays beyond six hours.

§ Disruption Recovery Policy 4.2
Precedent

What worked before

A near-identical stranded-connection case was resolved this way, with a satisfied outcome.

§ Incident RI-1048
Limitation

What's still missing

Partner-airline confirmation is required before the rebooking can be finalized.

Open question

Interactive demo · no account required

Explore ResolveIQ as…

Pick a lens and land exactly where that role starts. Access is brokered server-side with a short-lived session — not just hidden in the UI. Switch roles anytime.

No account required·Sample data only·Switch roles anytime·Administrator is owner-only — see the access-control matrix

Built to be trusted

Every recommendation is traceable, and every decision is human.

Evidence-linked recommendations

Each action names the policy and precedent behind it.

Human approval

No consequential action happens without a person.

Source attribution

Open any citation down to the exact passage.

Similar-incident retrieval

Past resolutions inform the current one.

Audit history

Who decided what, and when, is recorded.

Serverless AWS architecture

Event-driven, scale-to-zero, cost-aware by design.

Behind ResolveIQ

Serverless, and scale-to-zero.

Static Next.js on Amplify Hosting; a Cognito-guarded API Gateway HTTP API; domain-grouped Lambdas; a Step Functions Standard workflow orchestrating ML risk inference, RAG retrieval (Bedrock Knowledge Base on S3 Vectors), grounded generation with a guardrail, and a task-token human-review pause. DynamoDB, EventBridge, and CloudWatch throughout — all provisioned with AWS CDK. No servers, scale-to-zero, per-run cost surfaced in the UI.

All Meridian Air passengers, flights, cases, policies, and operational data shown here are fictional and synthetic.