deepidv
All AI Prompts
FinTechTask Prompt

Compliance Prompt for Perpetual KYC (pKYC) Risk Trigger Design

This **Luna** task prompt takes your active customer risk records and transaction monitoring streams and builds an automated **Perpetual KYC (pKYC)** trigger model. Luna, the deepidv compliance overseer, returns an event catalog covering every signal that should recalculate a risk score, a trigger model with materiality thresholds and routing decisions, a deduplication spec that collapses overlapping point-solution alerts into a single case, and a transition plan that retires calendar-based reviews wherever regulation allows. Built for financial crime and compliance leads at fintechs who want risk recalculated the moment a material event lands, without their investigators drowning in duplicate alerts from fragmented point solutions.

Compliance Prompt for Perpetual KYC (pKYC) Risk Trigger Design

How to use this prompt

  1. 1

    Open Luna in the deepidv dashboard and paste the full prompt, or run it in Claude, ChatGPT, or Gemini if you are drafting the trigger model outside the platform.

  2. 2

    Replace the INPUT section with your customer risk record schema, transaction monitoring streams, the point solutions currently raising alerts, and your investigation team's capacity figures.

  3. 3

    Run the prompt and read the event catalog first: every trigger event with its source stream, plus the instrumentation gaps your current streams cannot yet detect.

  4. 4

    Hand the trigger model and deduplication spec to your financial crime engineering team, and route the transition plan to your MLRO for sign-off on each retired periodic review.

  5. 5

    Re-run the prompt after each new monitoring stream, product launch, or regulatory update so the trigger model keeps pace with the events your customers actually generate.

The prompt

Luna, analyze our active customer risk records and transaction monitoring streams to establish an automated Perpetual KYC (pKYC) trigger model. Configure dynamic event rules that automatically recalculate customer risk scores and route material risk changes for investigation without generating duplicate point-solution alerts.

ROLE:
You are Luna, the deepidv compliance overseer. You convert static periodic-review KYC programs into event-driven perpetual KYC models that recalculate risk the moment a material signal lands.

CONTEXT:
Fixed one, three, and five year refresh cycles leave risk changes invisible for months and flood analysts with low-value periodic reviews. A pKYC model replaces the calendar with events: each material change in a customer's profile or behavior recalculates the risk score immediately, and only material changes route to an investigator. The failure mode to avoid is alert duplication, where overlapping point solutions raise the same event three times under three names.

INPUT, the user will paste:
- Customer risk record schema: the attributes scored today, the scoring bands, and the current periodic review cycle per band
- Transaction monitoring streams and the events each stream emits (velocity changes, new counterparties, geography shifts, sanctions or PEP list movements)
- Point solutions currently raising alerts, with the event types each one covers
- Investigation team capacity and the current alert-to-case conversion rate
- Materiality definitions the compliance team already uses, if any

TASKS:
1. Analyze the supplied risk records and monitoring streams to identify every event type that should recalculate a customer risk score.
2. Build the pKYC trigger model: for each event type, the recalculation rule, the score movement that counts as material, and the routing decision.
3. Design the deduplication layer that recognizes when multiple streams or point solutions describe the same underlying event and collapses them into a single case.
4. Define the phase-out path for calendar-based reviews, keeping periodic review only where a regulation explicitly requires it.

OUTPUT FORMAT, return the following structured response:

1. EVENT CATALOG
- Every trigger event with its source stream, the risk attributes it touches, and its expected frequency
- Events the current streams cannot yet detect, listed as instrumentation gaps

2. pKYC TRIGGER MODEL
- Recalculation rule per event type with the score movement threshold that defines a material change
- Routing decision per outcome: auto-clear, monitor, or investigate

3. DEDUPLICATION SPEC
- The matching keys that identify duplicate alerts across streams and point solutions
- The collapse rule that merges duplicates into one case with a full signal trail

4. TRANSITION PLAN
- Which customer segments move to pKYC first and the criteria for expanding coverage
- The periodic reviews retained for regulatory reasons and the ones the trigger model retires
- Projected investigator load under the new model measured against current capacity

Where the supplied records or streams are insufficient to define a trigger or a materiality threshold, flag the question for the compliance team instead of guessing.

Test it in Claude or another LLM

This prompt is built for the Luna agent inside deepidv, where Luna reads a firm's live risk records and monitoring streams and writes the trigger rules back into the compliance workspace. You can dry-run the same workflow in any general LLM first with synthetic record and stream data to see the trigger model take shape before connecting real systems.

  1. 1

    Paste the full prompt into Claude, ChatGPT, or Gemini, but replace the opening 'Luna,' with a role instruction such as 'Act as a financial crime architect designing a perpetual KYC trigger model from risk records and monitoring streams.' Keep the four OUTPUT FORMAT sections exactly as written.

  2. 2

    Under the INPUT section, paste the synthetic sample block below so the model has a risk schema, event streams, and capacity figures to design against.

  3. 3

    Add one framing line: 'This is synthetic test data. Where a trigger or materiality threshold cannot be derived from the input, flag it as an open question for the compliance team instead of guessing.'

  4. 4

    Check the output shape: an event catalog with instrumentation gaps, a trigger model with materiality thresholds and routing decisions, a deduplication spec with matching keys, and a phased transition plan with projected investigator load. If the model invents a stream the input does not contain, tighten the role line and re-run.

  5. 5

    Once the output shape is right, run it live in the deepidv dashboard where Luna executes it against your real risk records and monitoring streams.

Synthetic sample data to paste alongside the prompt

Fake test data, safe to share with any LLM. Swap in your own once the output looks right.

CUSTOMER RISK RECORDS (synthetic, fake):
- Scored attributes: geography, product mix, expected volume, PEP flag; bands low/medium/high
- Review cycles (fake): low 5y, medium 3y, high 1y
MONITORING STREAMS (fake): card transactions (velocity and geography events); wires (new counterparty and threshold events); screening (sanctions/PEP list movement events)
POINT SOLUTIONS (fake): legacy TM tool raising velocity alerts; separate screening tool raising list-hit alerts; both fire on the same wire event
TEAM CAPACITY (fake): 6 investigators, roughly 40 cases/week; alert-to-case conversion 12%
MATERIALITY (fake): none defined; all alerts currently treated equally

FAQ

What is Perpetual KYC (pKYC)?

Perpetual KYC replaces fixed periodic review cycles with event-driven monitoring: a customer's risk score is recalculated the moment a material change lands in their profile or transaction behavior, instead of waiting for a one, three, or five year refresh. Only material score movements route to an investigator, which cuts low-value periodic reviews while catching risk changes months earlier.

How does the trigger model avoid duplicate alerts?

The deduplication spec defines matching keys that recognize when multiple monitoring streams or point solutions describe the same underlying event, then collapses them into a single case that carries the full signal trail. Analysts see one investigation per event instead of the same wire raising three alerts under three different names.

Does pKYC eliminate periodic reviews entirely?

No. The transition plan retains periodic reviews where a regulation explicitly requires them and retires the rest in phases, starting with the customer segments where event coverage is strongest. Luna lists each retained review with the regulatory reason behind it so the decision is defensible in an audit.

Can I use this prompt outside the deepidv dashboard?

Yes. The structure works in Claude, ChatGPT, or Gemini as a design framework and will return the event catalog, trigger model, deduplication spec, and transition plan. Live risk record analysis, stream connections, and case routing only work when it runs inside the deepidv dashboard.

Run it with live verification data

These prompts work in any LLM. Inside the deepidv dashboard, Luna, Arbiter, and Arc run them against your real sessions, screening lists, and audit trails.

Book a Demo