Internal · Call prepJasmin Flori-Hess · Raindrop
🎯 Goal: understand · listen · one next step. No pitch, no deck, no price.
00 How you introduce yourself (under 30 seconds)
1 · Who
"I'm Asim, I run Codeaza, an engineering company. About 8 years in engineering, the last 2 mostly in insurtech."
2 · What we do
"We take messy documents (policies, claims, compliance stuff) and turn them into structured data people can trust. A lot of that is measuring whether the AI got it right, and fixing it when it didn't."
3 · Hand it back
"That's why your post on interpretation caught my eye. But I'd much rather hear about Raindrop. How did the POC go?"
⚡ Short version: "I'm Asim, I run Codeaza. We're engineers who help insurtech teams turn documents into data they can trust. Mostly I want to hear about you."
🕐 When: right after the congrats, or when she asks "tell me about you". Then stop and let her talk.
01 Her right now
📈 Demand is ahead of capacity
POC, carriers, 3 markets, a raise. Small team.
POC, carriers, 3 markets, a raise. Small team.
🛡 Trust is her product
Carriers will ask her to prove the accuracy.
Carriers will ask her to prove the accuracy.
02 Numbers to know
98%
mapping accuracy she claims
15h
to read 8 policies → the idea
6,000+
clauses mapped (2025)
1 / 50
InsurTech NY · only Aussie
2–10
people on the team
03 Her journey (hover for detail)
🏥
2017
AO Foundation
🏦
2021
UBS Zurich
🐊
2022
Outback → idea
💧
Apr 2023
Founded
🌱
Nov 2024
Seed round
🚀
Jun 2025
Launch
🗽
Sep 2026
Guidewire · NY · Zurich
⚡
Oct 2026
AI cohort · POC
🏆
Nov 2026
Award nominee
Milestones per year: she's accelerating
What it means for the call
- She's moving fast and probably stretched
- She needs to execute: onboard carriers, ship, prove
- Good time for outside help
04 How Raindrop works
📄
Policy PDFswordings, PDS, endorsements→
🧩
Standardisesplit into "Droplets": cover, exclusions, conditions, definitions→
🧠
Insurance Ontologythe core IP: dependencies, triggers, relationships→
🧑💼 Brokers: Neptune AI answers with cited sources
🏢 Insurers / MGAs: benchmark, design and compare wordings
🤖 Tech + AI agents: white-label API
💬 Her pitch: "A System of Meaning, not another tool that reads PDFs." The insurance intelligence layer for people and AI agents.
05 Team · what she cares about
🔍
Proof & provenance"Whose interpretation is your AI acting on?" (you commented)
🚫
No buzzwordsHates "AI wrappers". Values hard, unglamorous work.
🧠
Domain > modelThe ontology is the moat, not the LLM.
🙋
Human in the loopAlways check the AI's output.
06 The 30-minute call
0–3Open
3–10The POC
10–20Go deep (her pains)
20–25Reflect
25–30Next step
👋 Warm upCongrats + set the frame
"Keen to understand what you're building."
🔥 Her nowHow did the POC go? Who for? What next?
"What made it extreme pressure?"
🔎 Pick 3–4 threadsAccuracy · onboarding · drift · team · US/CH · agents (cards below)
"Say more about that…" "What does that cost you?"
🪞 Play backHer top 2 pains, in her words. Then one short story.
"Is that fair?"
🤝 One stepSmall, dated, specific.
"Would it help if I…?"
Who talks
Jasmin 70%
You 30%
07 Question cards (pick 3–4, follow her energy)
🎯 Accuracy
"How do you measure that 98%? Per clause, per policy?"
📥 Onboarding
"New carrier wording that doesn't fit the ontology: who catches it? How long does it take?"
🔄 Drift
"When the model, prompt or ontology changes, how do you know nothing quietly broke?"
👥 Team
"How's engineering set up? What's the bottleneck?"
🌍 Expansion
"US + Switzerland at once: what has to be true technically?"
🤖 Agents
"Are customers asking to plug AI agents into the ontology yet?"
📖 Your one story, only in "Reflect": "An insurance document platform we work with: 12-step extraction pipeline, ~70% → 90%+ correct, with per-layer quality reports so regressions show up before release." Never name the client.
08 Listening map
👂 If she says…💭 It means…🎒 Pocket idea (don't pitch)
"Carrier wants proof on their wordings"
→
Needs evidence for risk teams, not a demo
→
📊 Evaluation + evidence layer
"Onboarding a carrier takes weeks"
→
Manual mapping, won't scale to the US
→
⚙️ Onboarding pipeline + review queue
"We're stretched / hiring is slow"
→
Capacity gap, roadmap slipping
→
👥 Engineering pod under Jamie
"They want it in their claims system / Guidewire"
→
Integration is blocking a deal
→
🔌 Integration build
"Investors ask about scale / defensibility"
→
Needs technical credibility for the raise
→
🏗 Architecture + accuracy story
"People want agents on our data"
→
New channel, no bandwidth
→
🤖 Agent API / MCP server
"We're fine, just busy"
→
No acute pain yet
→
⏳ Don't push. Reconnect at a real moment
09 Closing the call: which next step?
🏁 Win condition: a dated second call, ideally with Jamie (CIO). That's where work gets scoped and closed.
10 Do · Don't
✅ Congrats first✅ Ask, then ask "why"✅ Use her words back✅ Plain talk✅ Finish on time✅ Light touch: Zurich, kids, AI face-swap story
❌ Pitch / deck / price❌ Buzzwords❌ Doubt the 98% (be curious instead)❌ Mention Auckland❌ Run over 30 min❌ Name our client
11 After the call
💬
Thank-you DMsame day · one thing she said + next step date→
⚡
Do the promised thingwithin 48h→
🗂
Send Claude notes / Firefliesmeeting log · deal · follow-up task→
❤️
Keep engagingher posts, between meetingsSources: Jasmin's LinkedIn and posts (Aug–Nov 2026), Raindrop company page, raindropme.com, Insurance News (Jun 2025), Insurance Business, CRM #479. Internal only · 2 Nov 2026.