← ALL CASE STUDIES Digital Advertising Group · AdTech & Fintech

Run Meta ads by conversation

A self-serve ad platform where an AI compliance officer reviews every creative — and clients manage real budgets by messaging a WhatsApp agent.

2 weeksAI layer, from first commit to production
10K+creatives scanned daily across 5,000+ accounts
$20M+in monthly ad spend managed
100+chat transactions every day

The challenge

The group's clients fund and run Meta ads at serious scale — thousands of ad accounts, millions in monthly spend. Setting up each Meta business account took about an hour of manual work. Policy violations surfaced only after Meta flagged them, putting accounts at risk. And every budget change meant logging into a dashboard.

The solution

We built the platform end to end, then put an AI operations layer on top of it — shipped to production in two weeks.

A multimodal AI compliance officer reviews every actively spending creative daily: it transcribes voiceovers, reads video frames, and returns a policy verdict with a 0–100 risk score — before Meta ever flags the ad. The pipeline runs under a hard per-run cost cap, with versioned prompts and swappable models.

Clients manage their entire operation by messaging a WhatsApp agent: checking performance, pausing campaigns, topping up budgets with real money movements — over 100 transactions a day — with conversational campaign launch rolling out now. Underneath, account provisioning that took an hour now completes in minutes, and a nightly reconciliation engine audits charges against Meta's settled numbers, automatically correcting hundreds of discrepancies every day.

How it stays safe

01 / LEAST PRIVILEGE

The agent never holds the keys

The WhatsApp agent can't sign its own tokens. Every money movement runs on short-lived, narrowly-scoped credentials that are useless outside their exact purpose.

02 / SELF-HEALING FINANCE

Reconciliation that audits itself

Every charge is re-checked against Meta's settled spend data days later. Over- and under-charges are corrected automatically, with a full audit trail on every adjustment.

03 / COST-GOVERNED AI

The scanner has a budget

The compliance pipeline runs under a hard dollar cap per run, refunds budget on failures, and logs every verdict — AI that is itself measured and governed.

The results

Today the platform manages $20M+ in monthly ad spend across 5,000+ provisioned ad accounts, scans 10,000+ creatives every day, processes 100+ chat transactions daily, and catches hundreds of billing discrepancies before anyone notices them. The AI layer that does all of this went live two weeks after the first line of code.

ALL FIGURES FROM LIVE PRODUCTION SYSTEMS · CLIENT ANONYMIZED BY AGREEMENT

Want results like these?

A 30-minute working session on your goals and the smartest first step.

Book a Consult