Fintech / Compliance3 people6 months

How a 3-Person Fintech Hit £2M ARR by Replacing an 8-Person Ops Team With AI

Compliance automation, AI-powered KYC, and autonomous reporting that scaled a startup from 40 to 340 customers without adding headcount.

£2M

ARR with 3 people

£2M

ARR reached

from £800k

6 min

KYC processing

from 2 days

91%

support tickets resolved by AI

1,041%

first-year ROI

£434k

annual headcount saving

The problem

They had a working product and paying customers. Their operations were the bottleneck.

Every new Primeledger customer triggered a manual process that consumed 6+ hours of founder time across compliance review, onboarding, and setup. At 40 customers, this was manageable. At 100+, it would have required 4–6 new hires. At 300+, the business model broke entirely.

4 hours of manual compliance review per new customer

£340 in founder time per onboarding
📋

KYC process taking 2 business days — losing customers to faster competitors

Est. 18% drop-off during verification wait
📊

Weekly financial reporting consuming 12 hours of analyst time every Friday

£1,800/month on report generation alone
🎧

Founders handling tier-1 support questions that repeated 80% of the time

6–8 hours/week of CEO time on basic queries
The solution

Four AI systems. One operating layer. Zero new hires.

We designed the automation stack in order of ROI — highest-value constraint first. Compliance was the existential bottleneck. KYC speed was the competitive disadvantage. Reporting was pure waste. Support was founder leverage destruction.

Layer 01

Compliance Intelligence Engine

AI trained on FCA guidelines checks each new customer against 23 compliance criteria simultaneously. Flags exceptions for human review with full reasoning. Sub-3-minute decision on straightforward applications.

Claude 3.5 SonnetCustom FCA rulesetWebhook triggers
Layer 02

Parallel KYC Processor

Document verification, identity checking, and AML screening run in parallel — not sequence. Previously serial steps now resolve concurrently. 6-minute completion vs 2-day industry average.

Document AIAML APIParallel orchestration
Layer 03

Autonomous Reporting Engine

P&L, customer metrics, and FCA regulatory reports generated automatically at 6am each morning. Anomalies flagged with plain-English explanation. Friday analyst work: eliminated.

GPT-4oMake.comAirtableSlack
Layer 04

AI Support Layer

Trained on the full product knowledge base, regulatory FAQs, and 18 months of support tickets. Resolves 91% of queries. Routes complex cases with full context summary — the human never starts from scratch.

Claude 3.5IntercomKnowledge base sync
Execution timeline

From audit to full deployment in 8 weeks.

01
Week 1–2

AI Audit & Architecture

  • Full workflow documentation sprint
  • Data infrastructure audit
  • ROI model built with conservative assumptions
  • System architecture approved
Architecture sign-off
02
Week 3–4

Compliance Engine Build

  • FCA ruleset encoded and tested
  • Compliance AI trained and validated
  • Exception routing logic built
  • Parallel KYC processor deployed
First automated compliance check live
03
Week 5–6

Reporting + Support AI

  • Autonomous reporting engine live
  • Support AI trained on full knowledge base
  • Confidence threshold tuning
  • Human escalation paths validated
Full system live — monitored for 2 weeks
04
Week 7–8

Stabilisation & Handover

  • 2-week parallel run (AI + human)
  • Edge case library built
  • Internal team trained on oversight
  • 30/60/90 day monitoring plan delivered
Full handover — client operating independently
The results

Before vs. after. No cherry-picking.

All metrics measured at 12 months post-deployment vs the 3-month baseline before engagement. No attribution modelling — direct measurement only.

DimensionBeforeAfterChange
Customer onboarding time4 hours (manual)6 minutes (automated)
–97%
KYC processing time2 business days6 minutes
–99%
Support tickets handled by AI0%91%
+91pp
Weekly reporting hours12 hrs/week0 hrs/week
–100%
ARR£800,000£2,000,000
+150%
Team size2 people3 people
+1 hire
Compliance breaches0 (correct)0 (maintained)
No regression
JH

I genuinely didn't believe we could get to £2M without 8+ people. I was wrong. The systems Aroluxa built gave us operating leverage I didn't think was possible at our stage. We're now planning our Series A and the pitch is built around this model.

James HoltCEO & Co-founder, Primeledger
Strategic learnings

Three things we learned that apply to your business.

01

Compliance is a moat, not just a cost

When your compliance process is faster and more accurate than competitors, it becomes a sales advantage. Primeledger now leads with 6-minute KYC in sales calls. It closes deals.

02

Parallel beats sequential — always

Their KYC took 2 days because three verification steps ran one after another. Running them simultaneously cut time by 99%. Most multi-step processes are sequential by accident, not design.

03

The first automation funds the next three

The compliance engine alone saved £340k in year one. That saving funded the reporting engine, support AI, and 6 months of our retainer. The first system should always be your highest-cost constraint.

Ready to replicate Primeledger's results?

Get the same AI system built for your business

In 30 minutes we'll map your biggest operational bottleneck and show you exactly how AI can eliminate it — no fluff, no pitch deck.

Get Your AI AuditSee Automation Plan
No retainers
Results in 90 days
Fixed-scope delivery