Pre-Seed · 2026
Agentic shopping, built on real-time
household consumption data.
To the household, Trepo is the operating system for the kitchen — cook what you have, know what you have. To the market, that same product is the first live read on what people actually eat.
One is how we win users. The other is how we win the decade.
Trepo · Confidential
The Pain · 1 of 3
Food has never been more
available. Managing it is still manual.
Delivery in an hour. Ten thousand SKUs in an aisle. Groceries on four apps. Access to food is at an all-time high — and there is still no system for the part that comes after it arrives. You are the system. You stand in the aisle and try to remember what's in your own fridge.
🧠You shop from memory
The list is written from whatever you happen to recall standing in the kitchen. It is wrong before you leave the house.
🔁So you buy it twice
The third jar of cumin isn't a discipline problem. It's a visibility problem. Nobody can see their own pantry from the store.
🗑️And you throw it away
What you forgot you owned goes off behind what you bought instead. The waste is the receipt for the memory failure.
2 · The Pain
The Pain · 2 of 3
Every tool stops at the
checkout line.
There is an enormous industry built around getting food into the home. There is almost nothing built for the six days it sits there.
- →Delivery apps optimise the order. They go dark the moment it's dropped at the door.
- →Recipe apps assume you'll go buy the ingredients. They don't know you already own eleven of the fourteen.
- →Note apps and paper lists don't deplete. Nothing ever comes off the list on its own.
- →The fridge itself is a black box the second the door closes.
The unsolved gap
Purchase is instrumented to the cent. Consumption is not instrumented at all.
Which means the single most decision-relevant fact in the household — what you have and how fast you're going through it — is the one fact nobody records.
3 · The Pain
The Pain · 3 of 3
So the brands selling you food
are flying blind too.
Retail media is the fastest-growing ad channel in the world, and it is targeted almost entirely off purchase history — a lagging, fragmented, per-retailer guess at what a household wants. It can tell you someone bought a thing once. It cannot tell you whether they ate it, liked it, ran out of it, or replaced it.
Fragmented
One retailer's loyalty data sees one basket. The same household shops four other banners that retailer will never see.
Lagging
Panel and scan data arrives aggregated and weeks late, with the household stripped out of it.
Inferred, not observed
Everything downstream of the till is a model. Nobody is measuring the actual event — the food being used.
Both pains are the same missing dataset. The household can't see its own kitchen, and neither can anyone selling into it.
4 · The Pain
The Insight
Consumption is the only honest demand signal.
Purchase data is a lagging guess. Everyone measures the checkout. Nobody measures the pantry.
Why it's true
A purchase tells you a unit left a shelf. Consumption tells you it was wanted, used, and is about to be needed again. Only one of those predicts the next purchase.
Why nobody has it
You cannot buy this data. It only exists if a household voluntarily tells you what it has, every week, for free — which requires giving them something worth doing it for.
Why it lets us win
Agentic shopping is coming fast, and it will not work on fragmented purchase history. Whoever owns the consumption layer owns the agent's inputs.
5 · The Insight
The Solution
Make logging worth doing,
and the signal builds itself.
Trepo is the operating system for the kitchen. It tracks what a household has, tells them what they can cook with it right now, and turns what's running out into the shopping list — on an app, with a countertop device for the households that want zero friction.
- 1The wedge: a free app that earns the logging habit by paying it back immediately — recipes from what's already in the kitchen.
- 2The asset: every log is a timestamped consumption event on a real household. The Kitchen Graph compounds with use.
- 3The vision: once the graph knows the cadence, the shopping stops being a task. The agent replenishes; the human eats.
6 · The Solution
Product
Four reasons to open it.
All of them log data.
The habit is the product. Every job a household comes to Trepo for happens to be a job that writes to the graph — which is why the dataset grows without us ever asking anyone to "track" anything.
Check in
Back from the store, or clearing the fridge. Seconds of input, and the kitchen is current. Writes: what came in.
See the kitchen
The between-trips question — do we still have that? Checked from the aisle, not from memory. Writes: what's standing.
Cook what you have
Save a recipe off TikTok, get told what you're missing. The payback that earns the logging. Writes: intent.
The list, built for you
What's out and what's about to be, by store. The output the whole loop earns. Writes: demand.
7 · Product
Why Now
Agentic shopping is arriving
without its input layer.
- →Food access peaked; management didn't. Availability has never been higher and there is still no system for what to do with it once it's home.
- →The agents are coming, fast. Every major platform is racing to buy on your behalf. That is a solved interface problem and an unsolved data problem.
- →They will run on the wrong data. Fragmented purchase history across banners cannot tell an agent what a household actually eats. It will reorder what you bought, not what you want.
- →Logging finally got cheap. On-device AI made capturing consumption a two-second act instead of a chore — which is the only reason this dataset can exist now and couldn't five years ago.
The consequence
Retail media is the fastest-growing ad channel in the world and it is still being targeted off receipts.
A live consumption layer doesn't improve that market. It reprices it.
8 · Why Now
Traction
We launched. Then this
happened.
287
Our own plan's baseline
5,000
What we forecast for early September
12,924
Where we are tonight, three weeks early
We beat our own forecast by 2.5x, a month ahead of schedule, with no paid acquisition engine behind it. Peak day: 3,033 active households against a pre-launch baseline of roughly 35.
And they use it. Signups are cheap — logged food isn't. Every number below is a household choosing to do work for us.
273,945
Items logged into kitchens
across 4,997 households
108,635
Recipes generated
for 5,468 households
45,921
Recipes saved
by 3,554 households
25,063
Of those saved
straight off TikTok
733,085
AI operations run
on real kitchens
Product metrics pulled live from production on 17 Aug 2026. "Recipes generated" is a live snapshot rather than a lifetime counter. Founder and team accounts are excluded from household counts. The July step-change in signups is real and we are still instrumenting which channel drove it — we would rather say that here than have it found in diligence.
9 · Traction
Market Size
Priced off the basket,
not off the industry.
We don't think the size of the grocery industry tells you anything about us. The question that matters is whether one household is worth enough, and whether there are enough of them.
So: bottom-up, from a single kitchen. A household's annual grocery spend, the share of it that moves online where an agent can transact, and a 3% affiliate take on what we replenish.
Placeholder · needs a sourced figure
Per-household spend, online share and household count below are structural placeholders. I'll pin each to a citable source (BLS food-at-home, Brick Meets Click eGrocery) before this leaves the building — the arithmetic is right, the inputs aren't signed off.
Annual grocery spend, one household$X,XXX
Share transacted onlineXX%
Trepo take on replenished basket3%
Revenue per household, per year$XX
× target householdsX.XM
Serviceable revenue$XXXM
And that's the affiliate line alone — before the CPG data and retail-media revenue that arrives years earlier.
10 · Market
Competition
Everyone knows what you bought.
We know what you have.
Name the real incumbent honestly: it isn't another app, it's the note on your phone and your own memory. Everything funded in this space clusters on one side of a single line.
Purchase data · lagging, fragmentedConsumption data · live, in-home
Retailer
loyalty
Instacart
Panel & scan
(NIQ, Circana)
Samsung Food,
Whisk
Notes apps,
paper lists
Trepo
Why they can't cross
Their data is a by-product of a transaction they own. Nobody logs their pantry into a checkout.
What we had to earn
A daily reason to tell us what's in the house. That habit is the moat — it took a product, not a partnership.
What compounds
Every log makes the next recommendation better, which makes the next log more likely. The graph is the flywheel.
11 · Competition
Business Model
Sell the signal now.
Take the basket later.
Phase 1 · Live
CPG data & retail media
Brands pay for what no panel can give them: household-level consumption on their category, across every banner their buyer shops. Already in market — our first partner one-pager went out on real portfolio data.
Phase 2 · Next
Replenishment & affiliate
Once the graph knows the cadence, the list becomes an order. We take a share of a basket the household was always going to buy — with better timing than anyone else can offer.
Phase 3 · The swing
Both, compounding
Data revenue funds the household growth; household growth makes the data more valuable and the replenishment volume bigger. Neither line caps the other.
The device is not the business. It is the highest-fidelity acquisition and logging surface for households that want zero friction — and it pays for itself at point of sale.
12 · Business Model
The Team
A builder and a
seller.
Matt Taylor
Co-founder & CEO · Product and engineering
Builds the thing. Hardware and software both — the Halo device, the firmware, the OTA pipeline, the ingestion backend and the Kitchen Graph itself are his. Previously Neuro and Tesla.
Placeholder
Need Matt's exact titles, dates and the two or three shipped accomplishments he wants named at Neuro and Tesla. I'm not inventing a bio for a diligence document.
Zach Slaughter
Co-founder · Revenue
Owns revenue, and has stood on both sides of the trade we're building. Retail media and advertising at Buyen; CPG sales and marketing leadership at USCAPE — the buyer of this data and the seller of the products it measures.
Placeholder
Need Zach's exact titles, dates and the accomplishments to name. The framing above is the argument; the credentials need to be his words.
The pairing is the point: this company needs someone who can ship a consumer hardware-and-AI product, and someone who already knows what a CPG brand will pay for the data it produces.
13 · Team
Financials
Revenue, growth,
projections.
Growth is the line that's real today: 287 to 12,924 households, 2.5x past our own forecast, a month early. Revenue starts on the data side, where a partner conversation is already live, and the replenishment line turns on once the graph is dense enough to predict cadence.
$XXk
Contracted data revenue
Placeholder · the model
A 24-month projection needs monthly burn and cash on hand, and I have neither. Give me both and the model gets built as a supporting document — which is where Icehouse says it belongs anyway, alongside the cap table, the hiring plan and a written investor FAQ. The deck slide stays this thin on purpose.
Placeholder · revenue lines
Data revenue booked to date, pipeline value, and the assumed month the replenishment take-rate switches on.
14 · Financials
The Ask
The kitchen is the last unmeasured room in commerce.
We're raising $X.XM on a $XXM cap SAFE to turn a working consumer product into the consumption layer that agentic shopping will have to run on.
Households
Take the graph from thousands of kitchens to the density where cadence becomes predictable per category.
The replenishment bridge
Ship the leg that turns a list into an order — the product step that unlocks the take-rate business.
Commercial proof
Convert live CPG interest into contracted data revenue, so the second raise is priced off revenue, not story.
Confidential. Household examples anonymised throughout; behavioural signal only, never personal identity. Amounts, cap and instrument are placeholders pending founder sign-off.
15 · The Ask
