Case study · Insurance brokerage
From 20 minutes a policy. To about 4.
An AI admin system that reads every signed policy pack, checks it against the sales record and sends only real mismatches to a person.
- Client
- An Irish insurance broker
- Sector
- Insurance brokerage, Ireland
- Size
- 4 consumer brands
- Engagement
- AI system build
AI Verified: ready to submit
01
The situation.
The broker is part of an Irish group that sells life and mortgage protection under four brands. Every policy sold comes back as a signed document pack.
Before a policy could go to the insurer, the admin team had to prove the paperwork was right:
- The Fact Find, the Application and the Statement of Suitability were compared by eye: names, dates of birth, cover amounts and premiums.
- Each premium and commission lived in one of seven brand sales sheets. Admins searched for the row, then compared it to the paperwork.
Then everything was typed in again: into the Total Recorded Sales sheet, then into each insurer’s own portal. Five providers meant five different processes. A mismatch between documents, or a start date close to a birthday, was easy to miss, so problems surfaced after submission, not before.
02
What we built.
the policy system: an AI policy admin system, running on the broker’ own cloud. It reads every signed policy pack, checks it against the sales record, and hands the admin team a reviewed policy with any issues flagged. Staff step in only where something genuinely does not match.
Signed pack → insurer
6 steps · a person approves
The client signs the policy pack in DocuSign. The system matches the signature to the right sales row and brand.
Every document is read with OCR and classified as a Fact Find, Application, Statement of Suitability or Direct Debit. The key fields are extracted: identity, risk and lifestyle, policy and financial details.
Three AI checkers cross-reference the documents, on identity, risk and financial details. A separate check flags any start date within 14 days of a birthday.
The premium is looked up from the right sales sheet and compared to the paperwork. IBANs are validated.
The policy lands on a nine-stage board, from Unprocessed to Completed, as AI Verified or Not Matching, with every issue listed and one-click contact to the advisor.
The admin approves it and submits it to the insurer portal. The system detects the insurer’s release email.
Admin
03
How it works day to day.
Advisors keep working exactly as before: the system plugs into DocuSign and the existing sales sheets. The admin team works from the board and only handles real mismatches, at about 4 minutes of review per policy.
Self-hosted, client-controlled: the workflow engine and the database run on the broker’ own AWS account. Nothing in the system depends on BraindAI’s cloud, so the client owns both the process and the data.
04
The results.
- 4 minreview time per policy (was 20 to 30 min)
- 3,205policies processed since January
- 74 to 719policies a month, January to June
- 99.9%processing reliability
SourceThe broker’ own the policy system database, January to August 2026 (handover pack, 25 Aug 2026). Busiest week since: 217 policies (5 Oct 2026).
05
How we got there.
Jan 2026
Go-live
Live in daily use with the admin team. 74 policies in the first month, with weekly feedback and fixes from day one.
Feb to Jun 2026
Scale
Monthly volume climbed to 719 policies by June. In May the client asked for its sister brand Hello.ie to be added. It was live by 25 June and has processed 280 policies.
Aug to Oct 2026
Whole team
Version two rolled out to the full admin team. The sales records were reconciled, with 1,278 policies cleaned up in one pass. The busiest week to date reached 217 policies.
06
What happened next.
By September the admin team had moved off the old way of working entirely, and the next ideas were coming from them:
- Whole-team rollout: admin asked to roll the policy system out to every team member and retire the old version.
- A second brand: Hello.ie, added as a department, with 280 policies processed on the same system.
- A website chatbot, built on our insurance offering for brokers: 355 brokerable leads, 731 quote estimates and 12,832 conversations in 6 months.
- AI call analysis: the next project, analysing sales calls for sentiment, quality and coaching across the advisor team.
From 74 policies to 719 a month. From 20+ minutes to 4 per policy.
“Happy for you to roll it out to the rest of the girls on the team here and retire the old version.”
07
Tools and systems.
- DocuSign
- the brand sales sheets (seven)
- the Total Recorded Sales sheet
- the insurer portals (five providers)
- OCR
- AWS (the client’s own account)
- the policy system (the system’s name)
Bring one process your team hates. We will show you what it looks like with AI built in, and compliance in charge.
Start with a conversation.
A free 45-minute strategy call with a founder. No pitch deck. You leave knowing where AI would pay in your business, whether or not you work with us.
On the call
- 01Where AI would pay in your business.
- 02What your first system could be.
- 03What your AI Blueprint would cover.
Before you book
What does it cost?
Every engagement starts with the AI Blueprint, priced up front. The first system is a fixed fee, set from what the Blueprint finds. Nothing is quoted cold.
Where does our data go?
Nowhere new. It runs in secure environments, on your cloud or your own servers. Every action is logged, and nothing trains anyone else's model.
How much of our team's time does it take?
Most at the start, while we learn how the work really happens. After that, a short weekly session with the people using it.
Who owns what you build?
You do. The systems, the data, the prompts and the documentation.
Pick a time that suits you.
Book a strategy callPrefer email? Write to hello@braind.io.