01 · COVER
SERIES A · CONFIDENTIAL
Rebound

Travel like nothing
can go wrong.

The self-healing itinerary. An agentic trip-OS that owns every disruption end-to-end — zero call-center, zero paperwork.

STAGE
Series A
SIZE
€50M
LEAD
Atomico
RUNWAY
24 months
02 · PROBLEM
THE PROBLEM

One flight breaks. Five bookings collapse. Nobody owns the cascade.

When the 2am cancellation hits, the hotel is now a no-show, the car is still billing, the dinner is about to auto-charge, and the app on the traveler's phone says "we'll get back to you in 48 hours."

THE 89-POINT GAP
NPS +72 → −17
Traveler NPS before a disruption vs after. Eighty-nine points of trust, evaporated by a single uncompensated cascade.
03 · MARKET
MARKET

$1.61T global travel. We start where it breaks.

TAM · GLOBAL TRAVEL
$1.61T
All spend on flights, hotels, ground transport, dining, experiences in 2026.
SAM · DISRUPTED TRIPS
$74B
~18% of trips involve a qualifying disruption. Avg €254 refund or comp exposure per event.
SOM · 24 MONTHS
$620M
EU + UK addressable disruption spend at 8.4% penetration target.
BOTTOM-UP MATH
Active travelers (EU+UK)
142M
Avg trips / traveler / year
2.6
% with a disruption
18%
Avg Rebound revenue / cascade
€47
= Addressable revenue
$3.1B
04 · WHY NOW
WHY NOW

Three curves crossed in the last twelve months.

01
Agents are the new browsers.

Claude, Gemini, ChatGPT already book trips on behalf of users. A SERP listing is illegible to an agent. Rebound is the API-native trip-OS agents call — Booking.com's funnel is structurally invisible to them.

02
Inference collapsed 94% in 18 months.

A cascade rebook that cost €4.10 of tokens in 2024 costs €0.22 today. 82% gross margin at €47 per resolved cascade is durable, not a subsidized land-grab.

03
Post-COVID travelers stopped forgiving.

EU261 awareness up 3.4× since 2022. Two disrupted trips is enough to churn 71% of Booking.com's Genius Elite tier. The moment is now, not in 2028.

45% of VC dollars in 2025 went to AI-native startups · first year in history that share crossed half of every vertical's frontier company.
05 · THE PRODUCT
THE PRODUCT

The agent in action — 4m 12s, zero touches.

LIVE CASCADE · LIS → FCO · APR 21
TRUST 0.96 · NO HUMAN IN LOOP
02:14:03
Signal — TP422 cancelled in Amadeus feed
DISRUPTION
02:14:48
Reroute — Iberia IB3214 08:40 booked
+42s
02:15:19
Cascade — hotel +14h · car +4h10m
+31s
02:15:37
Cascade — Roscioli dinner → Apr 22 20:30
+18s
02:16:02
Refund — EU261 filed · €1,200 expected
+25s
07:00:00
Deliver — morning summary · 2 sentences · Ana boards Iberia at 08:40
4m 12s
06 · TRACTION
TRACTION · Q1 2026

Six months in. Numbers that matter.

PAYING TRAVELERS
24,800
+34% MoM · 6 mo
ARR
€4.2M
net retention 142%
CASCADES RESOLVED / AGENT
87%
zero traveler touches
NPS · POST-DISRUPTION
+74
industry median: −17
ENTERPRISE PILOTS LIVE
3
1 airline · 1 Tier-1 consultancy · 1 private-banking concierge
AVG RECOVERED / DISRUPTED TRIP
€1,247
EU261 + hotel + insurance
MEDIAN RESOLUTION
4m 12s
Booking.com SLA: 48h
GROSS MARGIN
82%
inference + supplier API only
07 · MOAT
MOAT

Why Booking.com can't copy this.

Three structural reasons. None of them are features.

01 · ORG
An incumbent cannot fire 26,958 people.

Booking.com's call-center, loyalty team, SEO team, and corporate onboarding org exist because its unit economics require them. Rebound is 42 humans and 1,200 agents. The gap isn't closable by adding AI to a 27,000-person org — the old org is the cost structure the new one is designed to delete.

02 · DISTRIBUTION
Agents can't read a SERP.

Booking.com's moat is Google ads and pay-to-rank affiliates. That moat is illegible to Claude, Gemini, and ChatGPT — the discovery layer that already books 18% of European trips. Rebound's API-first, structured, agent-readable surface is a category the incumbent would have to rebuild from zero, against its own cost base.

03 · DATA
Disruption data compounds.

Every cascade Rebound resolves becomes a training signal for the next. 500+ scenarios in our eval suite today; 5,000 in 18 months. Booking.com has more trips, but fewer resolved cascades — because most of its disruption events end at "48 hours, no outcome." Unresolved data doesn't compound.

08 · TEAM
FOUNDERS

Built this before. Built it smaller this time.

Inês Carvalho · CEO
ex VP Product, Skyscanner · ex Uber Rides EU

Owned rebooking at Skyscanner when it went from a price-comparison site to an actual trip company. Knows where the bodies are buried in every major GDS.

SUPERPOWER · reads an airline contract like a novel
Rafał Gorecki · CTO
ex Anthropic (Agent Platform) · ex Stripe Issuing

Shipped one of the first production agent stacks at Anthropic. Before that, built the eval harness for Stripe's fraud models.

SUPERPOWER · ships agents with evals, not vibes
Dalia Halevi · COO
ex Director of Ops, Lufthansa Disruption Desk · ex BCG

Ran the control tower that rebooks Lufthansa's worst-day-of-the-year cascades. Built our Field Ops model from what she saw break.

SUPERPOWER · has been on the other end of the phone call
Kenji Okada · Chief Design Officer
ex Staff Design, Linear · ex Monzo early

The reason the morning summary reads like a friend, not a bot. Owned the voice of Monzo through its 0→1 years.

SUPERPOWER · writes UI copy that calms a panicking traveler
09 · THE ASK
THE ASK

€50M · 24 months · €25M ARR.

WHAT €50M BUYS
  • 40%
    Agent platform + evals. Hire 14 more engineers. Extend the eval suite from 500 to 5,000 disruption scenarios.
  • 25%
    Enterprise Field Ops. Grow from 6 to 14. Land 12 new corporate accounts across EU + UK.
  • 20%
    Supplier coverage. Direct integrations with 40 more airlines and 8 hotel chains. Cut rebook time from 4m12s to < 2 minutes.
  • 15%
    Runway buffer. 6-month cushion past Series B readiness.
18-MONTH MILESTONES
PRODUCT
87% cascades resolved at scale
across 120,000 paying travelers
TRACTION
€25M ARR · 65% 6-mo retention
3 enterprise accounts live
TEAM
60 people · 1,200 agents
London office · RPE > €250k
LEAD
Atomico · Niall Wass
PARTICIPATING
Index · Sequoia Scout · existing seed
NOTABLE ANGEL
Elon Musk (personal · agentic travel thesis)
10 · CLOSING
TRIP IS TRIP

In 2026, Rebound is the OS for disrupted travel on Earth.

The same agent architecture rebooking a Lisbon–Rome cascade in four minutes will rebook a suborbital leg in 2034, and an orbital transfer beyond.

Trip is trip. Rebound scales.

2026
EU disrupted travel
2030
Global disrupted travel
2034+
Sub-orbital & beyond
11 · APPENDIX · FACTS
APPENDIX · FACTS

Numbers this deck rests on.

GLOBAL TRAVEL MARKET
$1.61T
UNWTO + Euromonitor, 2026
TRUST GAP · POST-DISRUPTION
89 pts
NPS +72 → −17, internal + EU Skift
AI-NATIVE VC SHARE · 2025
45%
Pitchbook, Q4 2025
BOOKING.COM MARKET CAP
$182B
Apr 18, 2026 close
BOOKING.COM HEADCOUNT
~27,000
FY2025 10-K, BKNG
REBOUND HEADCOUNT
42
+ 1,200 agent instances
AVG EU261 RECOVERY
€1,247
Rebound Q1 2026 data
MEDIAN CASCADE RESOLUTION
4m 12s
signal → new itinerary → claim