MOONSHOT OS - INTERNAL
Moonshot Travel Bible - v0.1 - May 2026
A Moonshot Product

Moonshot Travel - a teardown

Mission Control - your 24/7 travel agent that watches your flights, rebooks when rates drop, and turns booking into a conversation.

AuthorAlex Beckman ForJimmy & Ben Read time~45 minutes
CONTENTS
Section 01 - Mission Control

The agent.

Moonshot Travel doesn't open with a search form. It opens with a conversation. Mission Control is the conversational interface to every flight, every fare, every trip.

Mission Control empty
msos.ai - empty state - Mission Control panel
PURPOSE

Mission Control is your 24/7 travel agent. Tell it where you want to go, and it handles the rest - finding flights, tracking prices, rebooking when rates drop.

FEATURES
01.01

"Where shall we go?"

A single text input. No fields, no date pickers, no origin/destination dropdowns. Just type where you want to fly.

WHY IT MATTERS
Every other travel site is a form. Real travel intent starts as a sentence. Mission Control parses that and gets out of your way.
SAMPLE
Find me a flight from SFO to NYC
SAMPLE
What's the best time to book a summer trip?
SAMPLE
Show me cheap flights to Hawaii
INPUT
Plain English - parsed for intent
Empty
Mission Control empty state
Median $188 - Range $188-$300
22 data points
Data-backed claim
01.02

The data-backed claim.

"Based on 22 data points, when Mission Control finds a drop, median savings is $188. Range: $188-$300." Not a marketing number - a real number with a real sample size.

WHY IT MATTERS
Every booking site claims "save money." We name the median, the range, and the n. As the dataset grows, the claim sharpens.
01.03

Sample prompts as scaffolding.

Three tappable prompts sit under the agent's greeting. For users who freeze at a blank input, these are the on-ramp.

Three on-ramp prompts
Sample prompts
ROLLUP
Mission Control becomes the operating layer.
v1.1
Memory
Mission Control remembers preferences. Second trip is easier than the first.
v1.2
Voice + Vision
Talk to Mission Control. Show it a screenshot of a friend's trip.
v2.0
Agent Mode
"Watch SFO-LHR business class. Book when it drops under $4K." It executes.
Section 02 - Results & Filters

Thirty flights, four chips.

When Mission Control finds your flights, the left rail fills with cards and the right rail keeps talking. Four filter chips cut the list without leaving the conversation.

Results
msos.ai - results - SFO to LHR - business and first class
PURPOSE

Results are a conversation, not a spreadsheet. Mission Control summarizes what it found in plain English, flight cards make those words real, chips narrow without retyping.

FEATURES
02.01

The flight card.

Airline, flight numbers, times in airline-display format, route via airports, fare class, flexibility, price right-aligned in USD. American Airlines AA1860/AA0194, 12:21 PM SFO to 10:50 AM LHR, $9,702.

WHY IT MATTERS
Every other travel site buries the price under nine filter facets. The card reads like a magazine, not a database.
Cards
Flight cards - editorial layout
NONSTOP - MORNING - EVENING - UNDER price
Filter chips
02.02

Four chips, not forty filters.

NONSTOP. MORNING. EVENING. UNDER price ceiling. No checkboxes. No sliders. Four taps cover 90% of what real travelers narrow on.

02.03

The summary line.

"THIRTY FLIGHTS - SFO → LHR" sits above the results. Number, route, monospace. A signature that says "we counted, you don't need to."

Summary line monospace
Summary
ROLLUP
Results become a ranking.
v1.1
Per-Route Median
Every card shows "$X below/above median for this route."
v1.2
Moonshot Score
Composite score: price, time, layover, reliability.
v2.0
Personalized Ranking
Results sorted by what Moonshot knows about you.
Section 03 - The Booking Flow

One button.

Selected state. The card expands. A single "BOOK THIS FLIGHT" button. Stripe + Duffel Balance pre-funded.

Booking
msos.ai - selected state - BOOK THIS FLIGHT
PURPOSE

Booking is the moment of truth. One tap, one Stripe sheet. Every additional friction step is a percentage of users we lose at the most expensive moment.

FEATURES
03.01

The selected card.

Tap a flight card and a blue ring lights its border. Card expands with flexibility line and a single full-width button: BOOK THIS FLIGHT.

WHY IT MATTERS
Every airline website monetizes at this step with junk fees and upsells. We don't. The clean interaction is the differentiator.
Selected
Selected - blue border, single CTA
Stripe Checkout sheet
Stripe Checkout
03.02

Stripe Checkout.

User pays Stripe directly. Apple Pay and Google Pay one-tap. Stripe handles SCA, 3DS, fraud screening.

03.03

Duffel Balance pre-funded.

Webhook hits backend, backend calls Duffel API to create the order from pre-funded Duffel Balance, PNR returns. Stripe -> us -> Duffel.

USER PAYS
Stripe Checkout
WEBHOOK
Stripe -> backend
DUFFEL ORDER
Backend -> Duffel - debit Balance
PNR
Returns to user - ticket issued
4-step ticketing pipeline
Booking pipeline
ROLLUP
Booking becomes unattended.
v1.1
Saved Travelers
Passport and KTN saved. Second booking is two taps.
v1.2
Auto-Book Rules
"If SFO-LHR business drops under $4K, book it."
v2.0
Corporate Mode
Team accounts. Policy guardrails. Approval flows.
Section 04 - Three Products

The rebook thesis.

Most travel sites end at booking. We start there. Three products turn one ticket into a system that watches fares, captures drops, and books savings automatically.

Lock the Drop - Stay Loyal - Best Deal
Three rebook angles, same dataset, three personas
PURPOSE

The thesis: airfare is a market that moves after you buy. Most travelers eat the loss because nobody watches their fare after booking. Moonshot does.

THREE PRODUCTS
04.01

Lock the Drop.

Same flight, same seat, same airline. If the price drops after you book, we rebook you onto the identical flight at the lower fare and refund the difference.

SAME FLIGHT
Same airline, same routing, same time
REBOOKED
Automatic - no traveler action
REFUND
Difference back to card
WATCH WINDOW
Booking until 72hr before departure
Lock the Drop
same flight, lower price
Lock the Drop
Stay Loyal
same airline, any flight
Stay Loyal
04.02

Stay Loyal.

Same airline, any flight within travel window. Status miles still earn. Loyalty stays intact.

WHY IT MATTERS
For frequent flyers, status with one airline is more valuable than $200 saved on a one-off ticket.
04.03

Best Deal.

Any airline, refundable base. We watch the entire market for your route. Maximum savings, minimum loyalty constraint.

WHY IT MATTERS
For travelers who care about price first. This is the product where the median $188 savings lands.
Best Deal
any airline, refundable
Best Deal
TRAJECTORY
From three products to one engine.
v1.1
June 1 Verdict
After 30 days of all three live, decide based on dataset.
v1.2
Hotel Lock the Drop
Same model, hotel rooms. Massive TAM.
v2.0
Total Trip Watch
Flights + hotels + cars, one watched bundle.
Section 05 - Trip Management

After the book.

My Trips is where the watch begins. Booked ticket, watched fare, savings ledger.

My Trips dashboard
Post-book home
PURPOSE

My Trips is the proof of work. The user sees their booking, the fare we're watching, alerts as the price moves, and money saved.

FEATURES
05.01

The trip card.

Each booked trip gets a card: route, dates, airline, PNR, current fare watch status, "savings to date."

Trip card
Trip card
Savings ledger
per-trip and lifetime
Savings ledger
05.02

The savings ledger.

Per-trip dollar savings, lifetime running total. The ledger is the screenshot. Every user with a positive ledger is a referral source.

05.03

The rebook event.

When Mission Control catches a drop and rebooks, the user gets a notification ("we just saved you $247"), trip card updates.

"saved you $247"
Rebook event
ROLLUP
Trips become a portfolio.
v1.1
Trip Timeline
Per-trip price history graph.
v1.2
Annual Report
Year-end recap. Spotify-Wrapped energy for travel.
v2.0
Travel CRM
All trips, travelers, loyalty programs in one dashboard.
Section 06 - The Data Moat

Price observation is the company.

Every Mission Control search is a price observation. Every rebook is a confirmed pricing event. The densest airfare time-series outside of GDS providers.

Dataset is the moat
Why this is special
PURPOSE

The dataset-as-moat thesis is locked. Every architectural choice optimizes for data integrity. Three products generate three signal types per user.

SIGNAL SURFACES
06.01

Per-route time series.

Every search returns 30+ flights with fare, fare class, restrictions, timestamp. Multiplied across thousands of searches per day.

Per-route time series
Density
Confirmed pricing events
Behavioral data
06.02

Confirmed pricing events.

Most price data is "what airlines said." Booking data is "what people actually paid." We have both - paired together with decision.

06.03

Rebook events = ground truth.

Every successful rebook is the dataset confirming its own prediction. Ground truth label on the "this is a high price" signal.

Self-validating model
Ground truth
Moonshot Intelligence - B2B
B2B
06.04

The B2B layer.

Consumer activity generates the corpus. Underneath, Moonshot Intelligence - paid feed for corporate travel, hedge funds, insurers, airlines.

CORPORATE
Median fare feed by route
HEDGE FUNDS
Airline yield + demand signal
INSURANCE
Trip-cancel underwriting
AIRLINES
Competitive benchmarking
ROLLUP
The dataset becomes the moat.
v1.1
Per-Route Stats
Public median and percentile bands.
v1.2
First B2B Pilot
Travel insurer or corporate travel manager.
v2.0
Moonshot Intelligence
Larger revenue line within 24 months.
Section 07 - NDC Roadmap

The supply story.

Three paths to fares: Duffel today, Verteil aggregator next, airline-direct NDC after. Each unlocks more inventory and bigger margin per ticket.

Duffel today - Verteil tomorrow - Airline-direct after
Supply roadmap
PURPOSE

Supply is the cost ceiling and the margin floor. Today thin spread on Duffel. Verteil unlocks 85+ airlines. Airline-direct earns full distribution margin.

THREE PATHS
07.01

Duffel - the wedge.

Live now. Duffel aggregates content via NDC and GDS, clean REST API, transparent per-PNR fee. Best-in-class developer experience.

COVERAGE
Major carriers across NDC + GDS
PRICING
Transparent per-PNR fee
DEV EXP
Great docs, sandbox
LIMITS
Thin margin, limited fare types
Duffel - live now
Today
Verteil ~$1-5K/mo
85+ airlines
Next
07.02

Verteil - the expansion.

Aggregator at next price tier (~$1-5K/mo). 85+ airline direct connections. Better margin per ticket.

07.03

Airline-direct NDC.

The endgame. Direct NDC to specific carriers. Requires IATA TIDS or Accreditation, ARC for US. $25-75K per carrier. Full margin retained.

Airline-direct - full margin
Endgame
TRAJECTORY
Supply deepens, margin lifts.
v1.1
Duffel Optimized
Cache NDC source field per offer.
v1.2
Verteil Trial
A/B Verteil vs Duffel for 30 days.
v2.0
First Direct NDC
Highest-volume carrier first. Margin doubles.
Appendix - Tech Stack

How it's built.

Two-repo architecture: Vercel for web frontend, Replit Autoscale for backend, Neon for Postgres, Clerk for auth, Duffel for supply, Stripe for payments.

STACK
A1.01

The frontend.

Next.js on Vercel at msos.ai. Repo github.com/alexbeckman83/moonshot-web. Server components for SEO, client for Mission Control. Clerk at clerk.msos.ai.

Next.js on Vercel
Frontend
Replit Autoscale
Backend
A1.02

The backend.

Node + Express on Replit Autoscale. Repo github.com/alexbeckman83/moonshot-travel. Neon: ep-noisy-firefly-aeckrzmj.

A1.03

Supply, payments, webhooks.

Duffel API for inventory. Stripe Checkout for payment. Stripe webhook -> backend -> Duffel order. Flag DUFFEL_USE_PAYMENTS=false gates old path.

Duffel + Stripe
Webhook pipeline
~$50-150/mo
Cost structure
A1.04

The monthly burn.

Total ~$50-150/mo at pre-launch / early-launch scale.

VERCEL
$0 hobby / $20 pro
REPLIT
~$10-30/mo idle
NEON
~$0-25/mo
CLERK
Free under 10K MAU
DUFFEL
Per-PNR fee
STRIPE
2.9% + 30c per txn
TRAJECTORY
Built for demand, not theater.
v1.1
Observability
Sentry on backend. PostHog on frontend. Cloudflare DDoS.
v1.2
Replit -> Railway
Migrate backend. Better deploy ergonomics.
v2.0
Multi-Region
Neon read replicas in EU.