/home / work / ionclear
FintechAI IntegrationMVP Engineering2025

Compliance that runs at the speed of money.

Ionclear needed KYC, transaction monitoring, and SAR reporting for a new neobank — in ten weeks, before their partner-bank sponsorship deadline. We built it from scratch.

ionclear · monitoring
247 alerts/h99.98% uptimep50 42ms
Flagged flows · last 24h
2,417
TXN-4019● high
TXN-4021○ normal
TXN-4025○ normal
TXN-4028○ normal
10 wks
from kickoff to live
₦1.8B
Series A, 8 weeks post-launch
99.98%
uptime in first 6 months
0
compliance findings at audit
Engagement
Client
Ionclear (Series A)
Timeline
Jan – Mar 2025
Team
3 Litcode eng · 1 founder
Stack
Next.js · Rust · Postgres · Temporal
Contract
Fixed-price · ₦5M
The brief

“We have a bank partner, a signed MOU, and 74 days until we lose it.”

Ionclear had secured sponsorship from a top-20 US bank — contingent on a fully functioning compliance stack by the time their first cohort of merchant customers came online. They had product designers. They had a founder with a compliance background. They had no engineers.

They needed: KYC onboarding, real-time transaction monitoring, a case management UI, automated SAR filing, and a full audit trail — all built to pass a bank-partner readiness review. From scratch. In under 90 days.

Approach

Build the hardest thing first.

We scoped the build around a single question: “what's the one thing the bank auditor will fail us on?” That became the critical path.

01
Monitoring engine first

We wrote the transaction-monitoring rules engine in Rust on day three — before the login screen existed. Every other system was built around its contract.

02
AI as an assist, not authority

LLMs triage alert queues and draft SAR narratives. A human always approves. We documented every prompt, every eval, every fallback — auditors love paper trails.

03
Audit log is the database

Append-only event log is the source of truth; everything else is a materialized view. Replay from any point, explain any decision, no reconstruction needed.

Cases
HIGH23
MED104
LOW812
CLOSED12k
Case #ION-4019
Structuring pattern detected across 7 deposits over 11 days.
● High risk · needs review
Case management

Where the analyst actually lives.

The compliance team spends 80% of their day here. We co-designed the UI with Ionclear's Head of Risk — three times — until the triage queue fit the mental model. Case close time dropped from 22 minutes (their old Excel process) to 3:40.

AI-drafted SAR

Two clicks, one sign-off.

Analysts used to write Suspicious Activity Reports by hand — 90 minutes each. Our drafter reads the full case history, composes a compliant narrative, and hands the analyst a polished draft. They edit, sign, submit. Average: 4 minutes.

# SAR narrative · draft v1
On 2025-04-14 through 2025-04-25, subject
account ION-4019 received 7 cash deposits
of ₦4.8M–₦4.95M, each below the ₦5M
CTR reporting threshold. Pattern consistent
with structuring per 31 CFR §1010.314.
— drafted by ionclear-sar-v2 · 01:42
AcceptEdit
“Litcode are the reason we have a product. Full stop.”
Mira Wexler
Founder & CEO · Ionclear
Outcome

What happened after we left.

Mar 2025
Bank-partner readiness review passed — zero findings.
Apr 2025
Live with first merchant cohort. 240K transactions in week one.
May 2025
₦1.8B Series A led by Ribbit, with our architecture in the deck.
Jul 2025
Monitoring engine open-sourced as a library by Ionclear's team.
Oct 2025
Retained Litcode for 2-day/month advisory. Still ongoing.
Next case study · 02 of 04
Fovea Health
Vision screening at scale · 2.1M patients screened