01 · COVER APRIL 2026 · CONFIDENTIAL
Quell

Your self-healing
trip agent.

When something breaks, Quell rebooks, refunds, and lets you sleep.

Stage Seed
·
Size $10M
·
Lead a16z
·
Closing Q3 2026
02 · PROBLEM

Travel books in 20 seconds.
Travel breaks in the other 99% of the trip.

The booking funnel is solved. Everything after the tap — the cancelled flight, the hotel that lied in its photos, the refund that takes six weeks — is still run by a traveler sitting on hold at 11pm, writing a form they shouldn't have to write.

The trust gap
89 pts

Gap between "trust a brand to help me book" and "trust a brand to help me when it breaks." Booking platforms earn the first number. No one owns the second. That's the wedge.

03 · MARKET

Built on the biggest consumer market there is.

TAM · global travel
$1.61T
annual spend · 2025
SAM · disrupted trips
$184B
~11.4% of trips need intervention
SOM · 5 years
$9.2B
at 5% penetration of disrupted trips
Bottom-up math
1.4 billion international trips per year · 2025
× 11.4% disruption rate (flight, hotel, refund event) = 160M disrupted trips
× €64 revenue per completed-trip event (trip fee + swap + refund cut) = €10.2B addressable
We only need 5% of that to be a $500M ARR company. The math is boring; that's the point.
04 · WHY NOW

Three curves just crossed.

01
Agent orchestration got cheap.

A closed loop of six agents holding a traveler's memory, negotiating with hotels, and filing refunds would have cost millions to run in 2023. Today it costs single-digit euros per completed trip. That's the threshold this company was waiting for.

02
Generative search ate the SERP.

When travelers start their trip in Claude, ChatGPT, or Perplexity — not in a search bar — the ranked-listings business model is a dead fish walking. Incumbents can't fight it; their entire GTM is optimized for a funnel that's no longer entered at the top.

03
Traveler behavior broke first.

Post-pandemic trust in travel suppliers collapsed. 68% of travelers under 40 say they expect an "agentic" resolution — the app just handles it — not a support queue. The generation that will never call a call-center is now the paying demographic.

05 · THE PRODUCT

One 2am text. No hold music.

Don't read the feature list. Look at what Quell did overnight for a real traveler.

Lufthansa 441 · Tuesday, 2:14 AM
02:14Lufthansa cancels. Disruption agent catches it before the airline texts.
02:14Swap agent books Austrian 811 · same arrival window.
02:15Comms agent moves the 8pm dinner to 9pm.
02:15Comms agent texts the partner.
02:16Refund agent files the €380 claim against Lufthansa.
09:42Traveler wakes up. It's handled. Refund lands 72 hours later without an email.
Zero traveler actions. Total cost to Quell: €0.84 in inference. Revenue: €18.
Median time · signal → resolution
12s
Traveler actions required
0–1
Agents involved
6
06 · TRACTION

Private beta · 7 months in.

Paying users
12,400
+41% MoM for 6 straight months
Trips healed
48,200
median €612 refund recovered
ARR run-rate
€4.2M
81% gross margin
NPS · post-disruption
74
measured after the agent resolves
The one metric we run the company on
87% of disruptions resolve with one or no human touches.
87%
07 · MOAT

Why Booking.com can't copy this.

Not feature-for-feature. Structural.

01
A 27,000-person org can't remove its search ranking team.

SEO, affiliate, ads, ranking ops — that's the majority of the company's revenue surface. The incumbent can add an agent on top, but it can't remove the funnel the agent would replace. We built without the funnel from day one. The absence is the moat.

02
Commission economics punish honest recommendations.

The incumbent earns ~15% commission on every booking. Telling a traveler "don't stay at this hotel, the photos lied" costs them money. We charge only when the trip works, so our recommendations are free to be honest. You can't retrofit that pricing model onto a $120B business without tanking the stock.

03
Traveler memory compounds. The incumbent's data doesn't.

Booking platforms know what you clicked. They don't know why last year's anniversary hotel was a disaster. Every Quell trip writes to a private memory graph — failure modes, preferences, why-this-mattered notes — that gets sharper every trip. After 18 months, the switching cost isn't "move your preferences"; it's "rebuild three years of trip-level self-knowledge." That's a moat that builds itself.

08 · TEAM

Four founders. Twelve years building the thing the last startup was missing.

ET
Elena Tombazis
Co-founder & CEO
Prior: Product Lead, Trips at Airbnb. Shipped the original recovery-fund flow.
Superpower: knows which 3 of 10 disruption flows actually matter to a traveler.
MK
Marek Kowal
Co-founder & CTO
Prior: Founding engineer at Mistral. Built the agent-orchestration layer on top of tool-use.
Superpower: agents that don't hallucinate tools because he wrote the runtime.
PC
Priya Chen
Co-founder & Head of Ops
Prior: Director of Disputes at Stripe. Automated the chargeback workflow for 12 card networks.
Superpower: refund agents that actually get the money back.
JN
Jens Nilsson
Co-founder & Head of Product
Prior: Early PM at Linear. Shipped the issue-velocity dashboard every AI-native team now copies.
Superpower: ships production code on Monday, writes the evals on Tuesday.
09 · THE ASK

$10M seed · led by a16z.

18 MONTHS · MILESTONE 01
The 87% holds in production.

Disruption-resolution rate stays at 87% with one or no human touches, across 10× the current trip volume. That number is the product.

18 MONTHS · MILESTONE 02
50,000 paying travelers.

€32M ARR run-rate at our current revenue-per-completed-trip. Gross margin holds above 78%. NPS above 70 through the cohort.

18 MONTHS · MILESTONE 03
10 countries · 10 languages · zero CS.

End-to-end refund workflows live in 10 languages, run entirely by the refund agent. Zero customer-service headcount across all 10 countries. The AI-native thesis, paid in cash.

What the $10M buys
45%
Engineering hires (18 → 30) · depth in agent orchestration and evals
25%
Inference and agent infrastructure · 10× current trip volume capacity
20%
Market expansion · 7 new countries on the existing stack
10%
Reserves · 24 months of runway floor
10 · APPENDIX · FACTS

The facts behind the pitch.

Global travel market, 2025
$1.61T
UNWTO + Phocuswright aggregated · excludes business-travel adjacencies.
Trust gap · book vs. recover
89 pts
Delta between traveler trust in "help me book" and "help me when it breaks." The wedge.
Share of VC dollars, 2025
45%
Of all global VC dollars deployed went to AI-native startups. The category isn't speculative anymore — it's the category.
Incumbent market cap · for reference
~$180B
Booking Holdings public market cap, April 2026. The prize is large because the incumbent is large. The structural asymmetries (prior slide) are why we think a slice of it is reachable.
Sources: UNWTO World Tourism Barometer · Phocuswright 2025 Annual · Expedia Group Trust Index · PitchBook Global VC Monitor.
Model assumptions and unit economics full detail available under NDA.
Data refreshed: 14 April 2026.