Skip to content

Six answers, one call

Decision One answer, with its reasons.

Every signal at a key moment becomes one answer, from allow to block, with the reasons behind it. Your app shows the next step; your team sees why.

Beforepayment.initiateHold coached payments

Hold

Call in progressisyes

andScreen being sharedisyes

orPayeeislinked to a mule ring

₹96,000 to R. Sharmajust now

HoldACTIVE_CALLSCREEN_SHARED

“Payment paused. Your money hasn’t left your account. We’ll call you on +91 •••• 3210.”

AppFRM · scam-callbackLog · v14

A rule that holds coached payments, and the decision it made: hold, with the reasons ACTIVE_CALL and SCREEN_SHARED, sent to the customer, the FRM queue and the audit log.
  • One call per moment

    Your app asks once at sign-up, login, a new phone or SIM, a new payee or the payment.

  • One of six answers

    Allow, warn, step up, cool off, hold or block. Your app handles all six.

  • Always with reasons

    Every answer carries the reason codes behind it. No bare scores.

  • Your rules, your call

    Your fraud team sets the policy per journey and market, and every version is kept.

Every signal, one answer.

What each module reads at the moment meets in one place, and comes back to your app as a single decision with its reasons.

This phone, SIM and binding, Behaviour, Call and screen, Payee, each wired to DeepID.

Write the rule in plain words.

A rule reads like a sentence: when these things are true at this moment, give this answer. Your fraud team writes it; no release is needed.

Beforepayment.initiateHold coached payments

Hold

Call in progressisyes

andScreen being sharedisyes

orPayeeislinked to a mule ring

  • Per journey and market

    Write one policy for retail in India and another for business banking. Each journey runs its own.

  • The strictest answer wins

    When several rules match, the strongest action is the one that runs, and every match is recorded.

  • Built from what DeepID reads

    Device, SIM, behaviour, the call and the payee are all fields your rules can use.

Send it where it’s needed.

One decision, three readers. The customer gets the next step, your team gets the case, and audit gets the record. All with the same id.

Beforepayment.initiateHold coached payments

Hold

Call in progressisyes

andScreen being sharedisyes

orPayeeislinked to a mule ring

  • ShowPayment paused. We’ll call you.→the customer
  • Queuescam-callback→your FRM
  • Sendthe same decision→case manager and SIEM
  • Logpolicy retail, version 14→audit
  • The customer sees the next step

    A plain message for each action, which you can keep or replace with your own words and languages.

  • Your team gets the case

    The same decision lands in your FRM or case manager, in the queue your policy names.

  • Audit gets the record

    Every decision is logged with its reasons, the message shown and the policy version.

Prove it before it acts.

Start in shadow: DeepID decides on your live traffic without acting, and you compare its answers with your own outcomes. Change a rule only when the numbers agree.

retail.yamlVersion 13 → 14Shadow

 - id: hold_scam_pattern
   on: [payee.add, payment.initiate]
   when: payee.mule_linked
         or (call.active and screen.shared)
   then: hold
-  queue: fraud-review
+  queue: scam-callback
+  message: payment_paused_callback
 - id: cool_off_new_device
   on: [login.attempt, payee.add, payment.initiate]
   when: device.new
   then: cool_off
-  hours: 24
+  hours: 12
   message: new_device_limit
  • Shadow proof first

    DeepID scores live traffic without acting on it. You compare the results with your own outcomes. Live traffic, no action taken.

  • See every change

    Each new version of the policy shows exactly what changed against the last one.

  • Trace any decision

    The version that made a decision travels with it, so you can always see which rule applied.

Six answers, in your customer’s words.

Every decision is one of these. Here is when each one comes, and what the customer sees.

  • Allow

    Everything holds

    “Payment sent.”

  • Warn

    Something is unusual, but the customer is in control

    “This payee is new to you. Check the name before you send.”

  • Step up

    Identity is uncertain

    “Confirm it’s you with a quick face check.”

  • Cool off

    A new device, SIM or payee

    “You can send up to the starter limit now. You get the full limit in 12 hours.”

  • Hold

    Signs of coercion or a mule payee

    “Payment paused. Your money hasn’t left your account. We’ll call you on +91 •••• 3210.”

  • Block

    Clear attack

    “We couldn’t verify this session. Your money is safe.”

Questions banks ask.

Can we use our own rules?

Yes. Your fraud team tunes the policy per journey and market, and your FRM’s own rules can use the reason codes as inputs. The decision is an input to your systems, never a replacement for them.

Where are decisions recorded?

Every decision is logged with its reasons, the message shown and the policy version, for your analysts and for audit. How long the log is kept is agreed before the pilot.

Can the decision feed our FRM?

Yes. The same decision that reaches the app reaches your FRM, case manager or SIEM as an event with its reason codes.

Is it ever just a score?

No. No score reaches your team without the signals behind it. Each decision carries the reason codes that raised it, so an analyst can see why a payment was held.

Try one-tap login on your own number.

See the network confirm your number with nothing to type, and the next check take over when it can’t. Then plan a shadow proof on one journey, measured on your own traffic.