Pre-Seed · 2026 · Brief

Agentic shopping, built on real-time household consumption data.

Trepo is the operating system for the kitchen. Households use it to know what they have and cook with it — and every time they do, it records what a real home actually eats.

Households get their kitchen back. The market gets a signal it has never had.

Trepo · Confidential
Problem & Opportunity

Procurement is measured to the cent. Consumption is not measured at all.

The household

Consumers shop from memory, so they buy it twice — then throw it away

Nobody can see their own pantry from the store. Delivery apps go dark at the door, recipe apps assume you own nothing, and a list never depletes on its own. The fridge is a black box the second it closes.

The brand selling into it

Retail media is the fastest-growing ad channel in the world, and it runs on purchase history

That history is fragmented by design. Every retailer sells its own slice, cut its own way — Kroger's data does not line up with Albertsons', and neither one sees the same household at Costco or Trader Joe's. Brands stitch mismatched store-by-store reports together and still cannot say whether the product got used, got bought again, or sat in a cupboard for a year.

Consumption is the missing dataset. A household cannot see inside its own kitchen — and neither can anyone selling into it.

2 · Problem
Value Proposition

One product. Two customers. The same data.

What the household gets

One system in place of five: the inventory you keep in your head, the list in the notes app, the recipe app that assumes you own nothing, the leftovers you forget, the trip you make twice. Trepo knows what the kitchen has, says what can be cooked with it tonight, and moves what is running out onto the list.

Log it however it happens — a photo of the receipt, the fridge or the shelf, your voice, or a tap on Halo, the device that sits where the food is so logging costs a second and happens while it is still true.

Five jobs, one product. Every one of them writes data.

What the brand gets

Household-level consumption on their category, across every banner their buyer shops — repeat versus trial, the replenishment window, the whole basket around their product, and the brand a household switched to when they left.

Then the ad products only this data makes possible: sampling into homes that just ran out of a competitor, couponing timed to the run-out rather than the calendar, basket tracking that closes the loop back to in-home use, and audience profiles built from what a home eats, not what it clicked.

Which home runs out on Thursday, not "households that probably buy soda."

Nobody is asked to track anything. Households log because logging is how the product works for them — the behaviour we sell is the same behaviour they came for.

3 · Value Proposition
Underlying Magic

Consumption is the only honest demand signal.

A purchase says a unit left a shelf. Consumption says it was used, and is about to be needed again. It cannot be bought at any price — it exists only if a household chooses to tell you, week after week. It is also the input layer agentic shopping does not have: an agent that cannot see the kitchen is guessing.

Twelve-pack in Saturday, gone by Wednesday

Three a day, every week, for months. We know the Thursday before they run out.

Yogurt every nine days, then nothing

They did not run out and they did not stop eating yogurt. They switched brands — and we can see to what.

Three stores in one week

Costco on Sunday, Trader Joe's on Wednesday, the corner shop on Friday. One basket no single retailer can assemble.

Nobody fills in a profile. The household builds itself out of timestamps — what they eat, how fast, and what they will accept instead. That is exactly what an agent needs in order to shop on their behalf, and every log sharpens it.

4 · Underlying Magic
Business Model

One dataset.
Three businesses, in order.

Phase 1 · Live

Ad products on the consumption graph

The products CPG already buys from Kroger, Instacart and NIQ — sold against a signal none of them have. Sampling into homes that just ran out of the competitor. Couponing timed to the run-out, not the calendar. Basket tracking and closed-loop measurement to in-home use. Audience profiles built from real consumption. Already in market — our first partner one-pager went out on real portfolio data.

Phase 2 · Next

Full-scale agentic

Once the graph knows the cadence, the list becomes an order. Trepo is the input layer the agent is missing: what a household eats, when it runs out, and what it will accept instead. We take a share of a basket they were always going to buy — and we get to ask first, because a retailer only finds out you need it when you turn up.

Phase 3 · The swing

Both, compounding

Data revenue funds household growth. Household growth makes the data more valuable and the replenishment volume bigger. Neither line caps the other, and every household added improves both at once.

Who buys this over time: consumers, through the app and Halo · CPG brands and their retail media budgets · retailers and delivery platforms · and the food-automation and home-robotics companies that will need in-home consumption to train on and cannot collect it themselves.

The device is not a revenue line. It is the highest-fidelity logging surface in the home, it deepens the dataset per household, and it pays for itself at point of sale.
5 · Business Model
Go-to-Market

No paid acquisition. 13,000+ users anyway.

13,371
Users signed up
281,726
Items logged into kitchens
112,207
Recipes generated
32,042
Recipes saved straight off social
LAUNCH · LATE JUNE 11 JULY TODAY 13,371 signups

All of it organic — word of mouth and the feed, with no acquisition engine behind it. Signups are cheap; logged food is not. On the sell side we go founder-led, direct to CPG brand and category teams with a one-pager built on their own portfolio.

6 · Go-to-Market
Competition

Everyone is building the agent. Nobody is building its input.

Purchase history · laggingConsumption · live, in-home
Retailer
loyalty
Instacart,
DoorDash
Panel & scan
(NIQ, Circana)
Samsung Food,
Whisk
Smart fridges,
notes apps
Trepo

Agentic shopping

Instacart, DoorDash, Amazon and the assistants can place the order. None of them can see the kitchen, so the agent is guessing at its own input — it knows what you bought from them, not what you have.

The data incumbents

NIQ, Circana and retailer clean rooms sample purchases: weeks late, one banner at a time. Their data is a by-product of a transaction they own, and nobody logs a pantry into a checkout.

The consumer product

Recipe apps without inventory, or a $3,000 fridge with a camera in it. Ours is a free app plus a cheap device that sits where the food is — which is why the logging actually happens.

7 · Competition
Team

We have been on both sides of this trade.

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

Exact titles, dates and the two or three shipped accomplishments Matt wants named. 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 are building. Retail media and advertising at Viant; CPG sales and marketing leadership at USCAPE — the buyer of this data, and the seller of the products it measures.

Placeholder

Exact titles, dates and accomplishments. The framing above is the argument; the credentials need to be Zach's words.

8 · Team
Financials

Unit economics,
and what is still open.

Revenue starts on the data side, where a partner conversation is already live. The replenishment line turns on once the graph is dense enough to predict cadence per category. At $261 a household a year across both lines, 5M households is roughly $1.3B — 3.7% of US homes.

Replenishment affiliate per household, per year$7,532 US grocery spend per household × 19% online × 3% take — US Census, Digital Commerce 360$43
Retail media per household, per yearInstacart earns 2.9% of basket value in advertising (Maplebear 10-K) applied to the whole household basket$218
Combined revenue per household, per year$261
The 2.9% comp is Instacart's realised rate on its own fulfilled orders; selling against the whole multi-retailer basket is the thesis, not a proven rate.
13,371
Users signed up
$261
Modelled revenue per household, per year
$XXk
Contracted data revenue
$XXk
Monthly burn
Placeholder · the model

A 24-month projection needs monthly burn and cash on hand. Give me both and it gets built as a supporting document alongside the cap table and hiring plan. Also needed: data revenue booked to date, pipeline value, and the assumed month the replenishment take-rate switches on.

Needs your call · target households

5M is the recommendation. Matt and Zach to confirm before this is shown.

9 · Financials
Current Status & Use of Funds

Live product. Real kitchens. Now the buyers.

Product

The app is live and in daily use. Halo is going out to beta households now, and the web build that opens up Android is in progress.

Data

13,371 users and 281,726 items logged — dense enough to profile a household, not yet dense enough to predict cadence in every category. That gap is what this round closes.

Demand

The first CPG partner conversations are live, opened with a portfolio one-pager built entirely on our own household consumption data.

We are raising $X.XM on a $XXM cap SAFE to spend on three things: household density so cadence becomes predictable per category, the ad products that convert live CPG interest into contracted revenue, and the replenishment bridge that turns a list into an order.

matt@trepo.ai hellotrepo.com
Confidential. Signups and product metrics pulled live from production, 19 Aug 2026. "Recipes generated" is a live snapshot rather than a lifetime counter. Founder and team accounts are excluded. The July step-change is real and we are still instrumenting which channel drove it — we would rather say that here than have it found in diligence. Household examples are illustrative of the signal and anonymised throughout; behavioural data only, never personal identity. Amount, cap and instrument are placeholders pending founder sign-off.
10 · Status & Ask