AWS CDS Agentic AI Partner Hackathon · 2026

When El Niño strikes, claims start on WhatsApp.

Floods, huaicos, storms and crashes in the rain, from one person or two. An AI agent on Amazon Bedrock AgentCore takes the claim by voice, photo and pin, and never decides who is at fault.

⚡
Claim filed in minutesVoice note, photo and pin on WhatsApp, at any hour. No app, no call centre.
👥
One person or twoFloods, huaicos, storms and crashes, all on one number.
⚖️
Fair by designBoth versions compared in code. The AI never decides who is at fault.
📄
Paperwork doneReport approved in the chat, PDF sent, case on the insurer’s map.
ClaimlineAI assistant of your insurer
Just me · El Niño flood
El Niño

El Niño keeps coming back. So do the claims.

Every few years warm Pacific water brings torrential rain to Peru’s northern coast. Rivers overflow, ravines send down huaicos of water and mud, and thousands of homes, shops, farms and cars are damaged in the same weeks.

1982–83
≈US$3.3B

in losses, one of the century’s strongest El Niños

1997–98
≈US$3.5B

in losses, about 6.2% of Peru’s GDP

2017 · Niño Costero
0M

people affected; up to US$3.1B in losses

2023 · Cyclone Yaku
0

homes affected, 65,065 people

2026–27 · Now
Very strong

WMO: El Niño is established, peaking late 2026, impacts into 2027

≈ 1 m of water · Castilla, Piura

Sources: Andina and Infomercado (1982–83, 1997–98); IGP and Infomercado citing INDECI (2017); COEN via Gestión (2023); WMO via AFP, September 2026. Loss estimates vary by method. The flooded house is the demo’s example case.

Beyond Peru

El Niño shifts rain around the world: floods in some places, drought in others.

The same warm Pacific water changes weather on several continents. Flood losses come in fast; drought losses come in slowly, as failed crops and dry wells. Both end as claims.

More rain, floodsDrought, heat, fire
South America · Pacific coast

Peru and Ecuador

Torrential coastal rain, river overflows and huaicos, as in 1997–98, 2017 and 2023.

South America · south-east

Southern Brazil, Argentina, Uruguay, Paraguay

Heavy rain and floods, such as Rio Grande do Sul in 2024.

South America · north

Amazon, Colombia, northern Brazil

Drier seasons and drought, with low rivers and crop losses.

Central America · Mexico

The Dry Corridor

Guatemala, Honduras, El Salvador and Nicaragua face drought and lost harvests.

North America

Southern United States

Wetter, stormier winters, from California to the Gulf Coast.

Africa · east

Kenya, Somalia, the Horn of Africa

Exceptionally intense rainy seasons and floods in 2023–24.

Africa · south

Zambia, Malawi, Zimbabwe

Drought in 2024; all three declared a state of disaster.

Asia-Pacific

Indonesia, Philippines, Australia

Dry, hot conditions and higher bushfire risk.

South Asia

India

A weaker monsoon tends to come with El Niño, hitting farm output.

Typical El Niño patterns; each event differs. Sources: WMO; NOAA; 2023–24 El Niño event reports (Zambia, Malawi, Zimbabwe disaster declarations; East Africa rains; Rio Grande do Sul floods). Claimline speaks Spanish and English today.

Natural disasters

Most disaster losses are never insured, and many that are wait for a claim.

Floods, storms and wildfires now cause most insured losses worldwide: many mid-size events and many small claims, often from one region in a few days. That is exactly the load a call centre can’t absorb.

US$0Bnatural-catastrophe losses worldwide in 2025
0%of 2025 insured losses came from floods, storms and wildfires
0%of disaster losses uninsured across 10 Latin American countries
1–2%of homes had flood insurance in the US counties hit hardest by Hurricane Helene

Sources: Swiss Re sigma 1/2026; Swiss Re 10-country study via Monitor Mercantil; AP and NBC analyses of NFIP data; AM Best (2026).

One person or two

Any claim is one of two shapes. One tap picks the path.

After the injury question, Claimline sends two WhatsApp buttons. The choice is saved in code, so no step is skipped, and the same engine serves every line of insurance.

👥 With someone else · Con otra persona

Two people, one fair report

Each person reports from their own phone, in their own language. Claimline compares the versions in code and asks only about what doesn’t match.

  • 🚗Car crash between two drivers, in the rain or notIn the demo
  • 🏍️Motorbike, mototaxi or bicycle against a car
  • 💧Water leak from the flat above: two neighbours, one damage
  • 🏠Damage to a rented home or car: tenant and owner
  • 📦Goods damaged in delivery: sender and carrier
👤 Just me · Solo yo

One person, a loss of their own

A location pin, when it happened, how high the water got, what was damaged and photos. One approval closes the case and the PDF arrives in the chat.

  • 🌧️Home flooded by El Niño rainIn the demo
  • 🏪Shop or small business floodedIn the demo
  • ⛰️Huaico or landslide damage
  • 🌪️Storm, roof or wall collapse
  • 🔥Fire, theft or break-in
  • 🌾Crops or livestock lost to rain or drought

Built today: crashes between two drivers, and floods at home or in a shop. The other claims use the same two paths; each needs its own questions and report fields.

Fair comparison

A crash in the El Niño rain: two versions, compared in code.

Each fact is labelled AGREED, ONLY_A, ONLY_B or CONFLICT by plain code. Claude only judges whether two free-text values mean the same thing. The agent asks once about each conflict, in neutral words, and never says who is right.

FactDriver A (wrote in Spanish)Driver B (wrote in English)Result
Place“Sánchez Cerro and Grau, Piura”“Sanchez Cerro and Grau avenues”AGREED
Time“at twenty past eight at night”“around eight twenty PM”AGREED
Impact“it hit my rear bumper”“I hit its rear bumper”AGREED
Weather“pouring rain, the road flooded”“it was only drizzling”CONFLICT
Why A braked“the water made me skid”“braked for no reason”CONFLICT

This is the demo’s crash case; Driver A’s words are translated from Spanish here. Both conflicts stay in the joint report, in each driver’s own words, and both drivers approve it.

How it works

From a voice note to an adjuster’s map in minutes.

Report

Text, voice notes (Amazon Transcribe), photos and a location pin, at any hour, in Spanish or English.

Safety first

The first question is whether anyone is hurt. If so, the agent gives the emergency number and hands the case to a person.

Collect each fact

A Strands agent on Bedrock AgentCore asks one question at a time and saves every fact with the person’s own words as evidence.

Compare

With two people, plain code compares both versions and the agent asks only about conflicts.

Approve

Each person reviews the draft in a WhatsApp Flow. Approving means “this is what I said”, not “I accept the other version”.

PDF and board

The report is built in code as a PDF and sent in the chat. The insurer sees the case on its claims board.

The insurer’s side

Every claim on a live map, with the whole conversation.

When the rain hits Piura, adjusters see where the damage is, which cases are approved, which have conflicts, and can open any conversation, fact and PDF.

Castilla · approved
Crash · 2 conflicts
Catacaos · open
0cases in Piura
0approved, PDF sent
0conflicts to review
ApprovedConflictOpen

Illustration of the demo’s claims board. The real board in the demo uses a street map.

Insurance solutions

One WhatsApp number for every line of insurance.

Claimline is the intake of the claim: it hands each insurer a complete, evidence-backed case for its existing process, and deploys into the insurer’s own AWS account.

🚗

Auto and SOAT

A neutral joint report from both drivers, plates typed and checked, conflicts flagged, a PDF for each driver.

🏠

Home and property

Floods, huaicos, storms and fire reported with a pin, the water level, damaged items and photos.

🏪

SMEs and shops

Owners report stock and equipment losses from the shop floor, by voice, without stopping work.

🌾

Agriculture

Peru’s state catastrophic crop insurance (SAC) pays S/800 per hectare and planned to cover over 1.6M hectares in 2023–24. Field reports with a pin and photos fit the “Just me” path.

🌊

Catastrophe surge

One number for everyone, serverless scale when thousands report at once, cases clustered on a map so adjusters go where the damage is.

🤝

Microinsurance and brokers

A low-cost channel for low-premium policies: no app to install and no call centre, in the chat people already use.

SAC figures: Midagri via El Peruano and Agraria.pe. Claimline doesn’t check coverage, estimate damage, score fraud or pay; it feeds those steps.

The claims process, end to end

How a US property and auto claim works, and where AI takes the paperwork.

In the US, a claim runs through the same stages whether it is a flood with one homeowner or a crash between two drivers with two insurers. Claimline automates the first stages today and hands clean, structured data to the rest.

Claimline does it todayClaimline feeds it with structured dataNext, with AI
1 · First notice of loss

Report the loss

WhatsApp, voice, photos and a pin, at any hour. One person or both people.

Today
2 · Statements

Each party’s account

Separate, neutral interviews; every fact saved with the person’s own words.

Today
3 · Documentation

Report and approval

Joint or one-person report, approved in a WhatsApp Flow, PDF built in code.

Today
4 · Triage

Assign the adjuster

The board shows every case on a map with status and conflicts, so the right adjuster goes first where it matters.

Feeds it
7 · Liability

Decide fault, by a person

With two drivers, the adjuster gets both versions and every conflict side by side. Claimline never decides fault.

Feeds it

Paperwork, written for you

PDFFirst notice of loss, timestampedToday
PDFEach party’s statement, in their own wordsToday
PDFJoint report with agreed and disputed factsToday
LOGEvidence log: voice notes, photos, pins, transcriptsToday
PDFProof of loss, pre-filled for signatureNext
PDFEstimate request and subrogation fileNext

Tracked data that answers tomorrow’s questions

Probability: how likely is a flood claim on this street in the next rainy season?
Risk: which intersections produce crashes when it rains?
Readiness: how many adjusters will Piura need in the first week of an El Niño?
Service: how many minutes from the first message to an approved report?

Every case keeps its location, time, weather, facts with evidence, conflicts and each message, in DynamoDB and S3 in the insurer’s own account. The questions show what that data can answer; the models are future work. Chart is an illustration.

Built on research

Each finding became a rule in the code.

FindingWhat Claimline does
Witnesses who talk pick up details they never saw (Gabbert et al., 2003–2006)Each person reports in their own chat and never sees the other’s answers
One loaded verb changes remembered speed (Loftus & Palmer, 1974)Neutral questions; never suggests a speed, a cause or a fault
A chatbot gets more informative answers than a web form (Xiao et al., 2020)WhatsApp, one question at a time, the person’s own words saved
83% of dissatisfied claimants plan to switch or have switched insurer (Accenture, 2014)The approval and the PDF arrive in the same chat, in minutes
Built on AWS

Serverless, pay per use, one command to deploy.

With no traffic it costs almost nothing; in a flood it scales with the claims. The whole stack is AWS CDK, deployed into the insurer’s own account.

AWS End User Messaging Social (WhatsApp)Amazon Bedrock AgentCore Runtime & Memory Claude on Amazon BedrockStrands AgentsAmazon Transcribe AWS LambdaAmazon SNS & SQSAmazon DynamoDBAmazon S3AWS CDK
Architecture: WhatsApp to End User Messaging Social, SNS, router Lambda, AgentCore agent with Claude, DynamoDB and S3, send Lambda back to WhatsApp

Live today

  • Joint crash report deployed on AWS and real WhatsApp, with the PDF delivered to both phones
  • Both paths, the full conversation log and the claims board, in this demo

Next

  • Both paths on AWS with Claude, and the approval Flow published
  • A five-person test, then a pilot with an insurer ahead of the next rainy season

Try it: two phones, one claims board.

Play the El Niño flood or the crash in the rain. The demo runs the real Claimline code; in this public version the agent’s replies are scripted, while the deployed version answers with Claude on Amazon Bedrock.