Active in production since October 2025

Detect every vulnerable customer before it goes wrong

AI monitors every customer interaction in real-time for signs of financial difficulty, mental health distress, confusion, and coercion. Flags instantly. Escalates automatically. Full audit trail for every FCA Consumer Duty review. Built for regulated contact centres.

Real-Time Detection Automatic Escalation Live in 2-3 Weeks
100%
Interactions monitored (not just escalated ones)
<2s
Detection time during live call
0
Vulnerable customers who fall through the gaps
2-3 weeks
To full deployment

Human agents miss vulnerable customers. The FCA won't accept that.

Consumer Duty requires you to demonstrate consistent identification and appropriate treatment of every vulnerable customer, not the obvious ones.

Vulnerability is inconsistently detected

Whether an agent spots a vulnerable customer depends on their training, their mood, and how busy the queue is. The FCA expects consistent identification regardless of agent, channel, or time of day.

The FCA Consumer Duty mandates proactive action

The Consumer Duty requires firms to proactively identify customers in vulnerable circumstances and take action to deliver good outcomes. Waiting for customers to self-declare is not sufficient.

No audit trail means no defence

When the FCA investigates a complaint involving a vulnerable customer, you need evidence that you had systems in place to identify and respond appropriately, at scale, not just in the cases you remember.

Vulnerable customers in collections carry the highest risk

Pursuing debt collection with a customer experiencing mental health difficulties or financial crisis is a regulatory and reputational liability. The cost of a single enforcement action dwarfs the cost of prevention.

How Consumer Vulnerability Agent works

Monitors every interaction. Flags the moment indicators appear. Escalates before harm occurs.

01

Monitor every interaction

Integrates with Amazon Connect for voice and your existing SMS, email, and WhatsApp channels. Every live conversation is transcribed and analysed in real-time throughout its duration, not just at the end.

02

Detect vulnerability signals

AI identifies language patterns associated with financial difficulty, mental health distress, cognitive confusion, bereavement, and signs of third-party coercion. Weighted scoring across multiple signal types, not simple keyword matching.

03

Escalate automatically

When vulnerability is detected, the agent receives an in-call prompt to adjust their approach. High-confidence signals trigger immediate supervisor alert. The interaction can be paused, transferred, or flagged for specialist follow-up, configurable to your protocols.

04

Log and evidence everything

Every detection, every escalation, every outcome, logged automatically against the customer record. Exportable vulnerability reports for regulatory review. Your Consumer Duty evidence is built continuously across every interaction.

What the agent detects in real-time

Covering all four FCA vulnerability drivers: health, life events, resilience, and capability.

Vulnerability Screen: Live Call #8,341 FLAGGED: Supervisor Alerted
Financial Difficulty: HIGH
Unable to make any payment, mentions job loss
Mental Health: MEDIUM
Distress signals, mentions feeling overwhelmed
Cognitive Confusion: LOW
Comprehension within normal range
Coercion or third party: LOW
No signs of third-party pressure
Bereavement: LOW
No bereavement indicators detected
Low Resilience: MEDIUM
Multiple concurrent stressors indicated

How we compare

Human detection vs keyword matching vs Consumer Vulnerability Agent.

Human Detection Keyword Matching Consumer Vulnerability Agent
Coverage Depends on agent attention 100% but crude, misses context 100%, context-aware, weighted signals
Accuracy High false-negative rate, many missed High false-positive rate, noisy alerts Contextual scoring, confidence-weighted
Detection Speed During call if noticed, often after Real-time but unreliable Real-time, under 2 seconds
Consistency Varies by agent, queue pressure, time of day Consistent but not intelligent Same standard across every interaction
Audit Trail Agent notes only, incomplete Keyword logs, no context Full log: signals, confidence, outcome
FCA Consumer Duty Evidence Anecdotal, agent-dependent Logs with no decision context Exportable, regulator-ready reports

Scoped, then delivered

We scope against your volumes, your systems and your compliance obligations, then quote a fixed price for delivery. Day rates for ongoing work. No per seat licence and no minimum term on a tool you have not seen working.

01

Discovery call

Thirty minutes on your volumes, your platform and what you need evidenced. We will tell you if this is not a fit.

02

Scope and baseline

Named intents or categories, agreed metric definitions, and your current numbers captured before anything is built. See how we measure.

03

Fixed price delivery

Four to six weeks to production, priced up front against the agreed scope.

04

Run or hand over

We operate it, or we hand it to your team with runbooks and dashboards. Your call.

Built on AWS

Production-grade infrastructure. Your data stays in your AWS region.

AI Engine

Amazon Bedrock

Context-aware vulnerability detection using Claude. Weighted scoring across multiple signal types with full reasoning chains.

Contact Centre

Amazon Connect

Real-time voice transcription and in-call agent prompts. Integrates with your existing Connect instance.

Processing

AWS Lambda + SNS

Serverless event-driven pipeline. Scales to any interaction volume without provisioning.

Specialisation

AWS Agentic AI Partner

Built by an AWS Advanced Partner with deep expertise in agentic AI and contact centre solutions.

Common questions

Which vulnerability categories does it cover?

All four FCA Consumer Duty vulnerability drivers: health (physical and mental), life events (bereavement, relationship breakdown, job loss), resilience (low financial resilience, over-indebtedness), and capability (low financial literacy, language barriers). We configure the weighting to match your customer profile and regulatory obligations.

How is this different from keyword matching?

Keyword matching generates false positives and misses context. Our AI understands the meaning and sentiment behind what a customer is saying, not just whether a word appears. A customer who says "I can't cope" in a financial context scores differently than one discussing a sporting event. Contextual understanding reduces false alerts and catches signals that keywords miss entirely.

Does it work during live calls or only afterwards?

Real-time, during the live call. The agent receives an in-call prompt the moment vulnerability signals breach the configured threshold, while they can still adjust their approach, offer forbearance, or transfer to a specialist. Post-call analysis is also available for training and audit purposes.

Does it work with our existing contact centre?

Yes. We integrate with Amazon Connect for voice, and connect to your existing SMS, email, and WhatsApp channels via API. Your agents' workflow changes only by receiving in-call prompts when vulnerability is detected. No new interface to learn.

How quickly can we deploy?

2-3 weeks from contract signature for standard deployments. This covers vulnerability model configuration, channel integration, escalation protocol setup, and dashboard access. You review and sign off before go-live.

Does this satisfy the FCA Consumer Duty vulnerability requirements?

The Consumer Duty requires firms to have processes to identify and respond to vulnerable customers. This agent provides systematic, documented identification across 100% of interactions, consistent application of response protocols, and a full audit trail for regulatory review. It is designed specifically to support Consumer Duty compliance evidence.

Stop relying on agents to spot what they might miss

Detect every vulnerable customer. Escalate in real-time. Build your Consumer Duty evidence automatically.

Book a Discovery Call

Or email us at hello@rel8.cx