Product

The review your best analyst would write, on every case.

Evidence gathered, your rules applied, and a memo that shows its work.

Try the live demo Book a demo

A case in Truluce: the agent recommends reject, with the evidence behind each statement

Where it fits

On the gray zone of the waterfall you already run.

Your engine keeps its clean approvals and its high-confidence rejects. Truluce takes what's left.

Insights: send fewer cases to review

The decision goes back to your engine

Your decision engine

Runs the vendor checks: identity, fraud scores, device, phone.

Auto-approveevery result clean

Gray zoneID check, then review

Rejecthigh-confidence signals only

truluce

Gathers every vendor's evidence
Runs your versioned rules
Writes a memo that cites it all

Your analyst decides

With a reason when they disagree.

Your decision engine
Runs the vendor checks: identity, fraud scores, device, phone.
Auto-approveevery result clean
Gray zoneID check, then review
Rejecthigh-confidence signals only
truluceon the gray zone
- Gathers every vendor's evidence
- Runs your versioned rules
- Writes a memo that cites it all
Your analyst decides, and the decision goes back to your engine.
Insights tell you which referrals to stop sending.

For analysts

Open a case. The work is done.

The case at a glance
A memo that cites everything
Sessions side by side
Linked applications
Decide in one place

For Risk leads

A playbook you own.

Versioned rules
Backtests on your history
What each vendor is worth
Agreement by rule and reason
Sign-off on every release

For engineers

Built to plug in.

Taktile and Alloy connectors
REST API and webhooks
Your own vendor keys
Hosted or in your cloud
SSO and an audit log

Fewer referrals

The best review is the one you didn't need.

Truluce sees how your analysts decide every referral reason. When one is approved almost every time, it tells your Risk lead and drafts the change for your decision engine, with a backtest attached.

At Made Card, this analysis took the referral rate from 50% to 15% of applications.

Read the field note

What Made Card's funnel analysis found

IP mismatch
Phones switch between Wi-Fi and cellular
Stop referring
Proxies and VPNs
Privacy relays and work VPNs are everywhere
Refer less
Address checks
Records lag for new homeowners
Refer less
First-party score
A document check can't show intent
Skip the document check

Applications referred at Made Card

50%15%

So far, only false positives removed.

Fraud types

It names the fraud, because each asks a different question.

A document check settles whether it's them. It can't settle whether they intend to repay.

Stolen identity

Is it really them?

Answered by

Both sessions, side by side
Devices and networks shared with other applications
Faces seen before

Synthetic identity

Is the identity real?

Answered by

The SSN's history against the date of birth
An email and phone only days old
Vendor scores, read together

First-party

Do they intend to repay?

Answered by

The applicant's own earlier applications
How the application was filled in
Not a document check: it can't show intent

And when it's none of them, the memo says so, and shows the evidence that cleared the applicant.

Change a rule. Replay your history. Sign it off.

A playbook release in Truluce, waiting for a Risk lead's sign-off

A backtest in Truluce: agreement by referral reason and what each vendor is worth

And what a Risk team asks for next.

Rings, reviewed together

Open cases that share a device or network arrive grouped, with the accounts the ring already opened.

Rules that decay, flagged

When a rule stops catching what it used to, or analysts keep overruling it, your Risk lead hears first.

A new vendor, tried on your history

Upload its retro file. See the fraud it would have caught, and the good applicants it would have flagged.

Evidence the APIs don't return

Partner confirmations and analyst attachments, cited like any other evidence.

What's next

One agent for every manual-review queue.

Onboarding first. Then the same rules, evidence and memo for each layer of a card issuer's fraud stack.

OnboardingLive
Is this a real person, and are they who they say they are?
Point of saleNext
Did they make this first purchase, and what does it say about intent?
Account managementPlanned
Is the person logged in still the person we approved?
PaymentsPlanned
Is this payment really from them?

What Truluce doesn't do.

It doesn't decide for you

Your analyst makes the call.

It doesn't pull credit

Credit data stays out of fraud calls.

It doesn't guess

What it couldn't check, it says.