Feasibility & Proof-of-Concept Report · Pre-Seed 2026

Kettlewise

Inventory autopilot for independent cafés · Kettlewise Inc., Portland, Oregon

1. Bottom Line

Technically: feasible, with two hidden problems the deck doesn't address. POS-connected demand forecasting is a solved category of engineering, and this team has directly relevant experience. Forecasting sales is not the same as drafting a correct order, though. A correct order also requires knowing what's on the shelf and how each drink translates into ingredients. Neither comes from the POS, and both put pressure on the "20 minutes, no spreadsheets" setup claim.

Commercially: early signal, not yet proof. 4 of 6 pilot cafés converting to paid is a real result. But the headline outcome (31% less pastry waste) is owner-reported, the sample is six cafés in one city over eight weeks, and nothing yet shows retention beyond the pilot window.

The largest risk is distribution, not technology. Nothing in the materials explains how Kettlewise gets from 4 paying cafés to 300. At $79/month, the cost of acquiring each café determines whether this business works at all. That cost is currently unknown.

2. Feasibility Scorecard

Dimension	Rating	Basis
Problem exists and is felt	Strong	12 of 14 interviewed owners named ordering their most stressful weekly task. Founder has lived it for 9 years.
Core forecasting tech	Strong	Working product. CTO has built store-level demand forecasting at 200+ store scale.
POS data access	Medium	Square and Toast integrations are live. Access terms are platform-controlled and could change (see §4.1).
On-hand inventory & ingredient mapping	Unaddressed	Not described in the materials. Required for orders to be correct (see §4.2).
Weather & events data	Open question	Source not specified. Contribution to forecast accuracy not measured.
Order delivery to suppliers	Open question	How an approved order reaches the supplier is not specified. "One tap" depends on it.
Customer outcome (waste reduction)	Medium	31% pastry waste reduction, owner-reported. No figures for milk or beans. Not independently measured.
Willingness to pay	Medium	67% pilot-to-paid conversion (4/6). Small sample. Reasons for the two losses are unknown.
Retention	Unknown	No paid-month history beyond pilot conversion.
Go-to-market	Weak	No channel, acquisition cost, or partnership plan described. Waitlist origin and quality unknown.
Market size framing	Medium	Arithmetic is correct, but the figure is a 100%-penetration ceiling. Small for venture scale without the expansion thesis (see §5.3).
Team–problem fit	Strong	Operator CEO plus forecasting CTO is the right pairing. Full-time status and team size are unconfirmed.

3. What the Pilot Actually Proves — and What It Doesn't

Proven (with caveats)

The product works end to end in real cafés, on real POS data, for eight weeks.
Owners will pay $79/month: 4 of 6 did.
Ordering is a recognized pain among Portland owners (12/14).
Current MRR ≈ $316. (MRR = monthly recurring revenue: the predictable subscription income per month.) This is derived arithmetic. The deck states no revenue figure.

Not yet proven

That waste actually fell by 31%. The figure is self-reported by owners who were invested in the pilot.
Any effect on milk, beans, or stockouts. These are half the pitch line and have no data.
Why 2 of 6 cafés didn't pay.
That paying cafés stay past month 3.
That the 20-minute setup claim holds for cafés the founders didn't personally onboard.
That results transfer outside Portland or to owners with no relationship to the founders.

Evidence quality issue: "stressful" is not the same as "will pay"

The 12/14 interview result shows the problem is felt. It doesn't show the problem is expensive enough to pay to solve. The pilot conversion is the stronger evidence, and it comes from six cafés. In investor conversations, lead with conversion and present the interviews as context. Present the 31% as "owner-reported," as the internal materials already do. A sophisticated investor will ask how it was measured, and "owners told us" will not hold up as a primary proof point.

The two non-converters are the most valuable data you have

A 33% loss rate from cafés that already experienced the product is either a pricing problem, a product-accuracy problem, a setup-friction problem, or a "nice but not necessary" problem. Each points to a different fix. Until you know which, you can't tell investors whether 67% conversion will hold at scale or fall.

4. Technical Feasibility

4.1 POS integration (Square, Toast)

Both platforms offer developer access, and Kettlewise is already live on them, so baseline feasibility is established. Two risks remain:

Platform dependency. Your data pipe runs through companies that also sell inventory features. Square offers stock tracking and low-stock alerts. Toast acquired xtraCHEF (invoice processing, costing, and inventory) and has been building it into its platform. Current feature depth should be verified directly. Both companies ship frequently and this assessment may be outdated. If either platform adds forecasting-driven ordering natively, it competes with you for the same customers using data you depend on.
Partner-program terms. Toast in particular has historically gated API access through a partner program. Verify current terms. Any change in access, pricing, or rate limits is a single point of failure until you add more integrations, which the raise funds.

4.2 The two gaps between "forecast" and "order"

This is the most important technical finding in this report. A POS records what was sold: "1 large oat latte." A supplier order requires what to buy: "2 cases of oat milk," given what's already in the walk-in. Bridging that requires two things the materials don't describe:

Ingredient mapping (recipes). Each menu item has to be translated into ingredient quantities: how much oat milk is in a large latte, how many grams of beans per shot. Pastries are mostly one-to-one, which may be why the pilot's only measured result is pastries. Milk and beans require recipe data per drink and size, plus waste allowances for steaming and dialing in.
Current stock on hand. To order the right amount, you need to know how much is left. The POS doesn't know. Options are manual counts (friction, and against the "no spreadsheets" promise), inferring stock from sales minus deliveries (drifts with spillage, staff drinks, and comps), or a hybrid with periodic quick counts.

Why this matters for your claims

If setup really takes 20 minutes, the product is probably either estimating recipes from defaults or forecasting pastries more precisely than milk and beans. Both are reasonable starting points. But the pitch line promises "milk, pastries, and beans," and the evidence currently covers pastries only. Document exactly how the live product handles recipes and stock today, then align the claim to it.

4.3 Forecasting on small menus

A single café's per-item volume is low and noisy. Selling 9 almond croissants on Tuesday and 3 on Wednesday is close to random variation. This is harder than grocery forecasting, not easier: the CTO's 200-store experience is relevant, but it came with far more volume per item. Specific challenges:

Cold start: a new café has little usable history at signup.
Intermittent demand: slow items sell zero on many days, which breaks simple models.
Asymmetric cost: overordering a pastry wastes a few dollars. Running out of oat milk at 10am loses regulars. The model should optimize for the cost of each error, not just average accuracy.

Opportunity: pooling patterns across cafés (for example, "Tuesday-after-rain" effects learned from 50 cafés applied to café 51) is the most credible technical moat available to Kettlewise, and it only grows with the customer base. If this isn't the architecture yet, it should be the roadmap.

4.4 Weather and local-events data

Weather data is commercially available from multiple API providers at low cost and is low risk. Local events data is patchier: coverage, accuracy, and licensing vary by city, and small-neighborhood events (a school fair two blocks away) are often missing from any feed. Two open questions:

Where does each input come from, and what does it cost per café?
Do these inputs measurably improve forecasts at single-café scale? Test this directly by running forecasts with and without each signal against actual sales. If events data doesn't move accuracy, drop it from the pitch rather than defend it.

4.5 Getting the order to the supplier

"One tap" only holds if the approved order actually reaches the supplier. Independent café suppliers are a mixed group: a regional dairy distributor, a local roaster, and a neighborhood bakery may take orders by text, email, phone, sales rep, or their own web portal. Formal electronic ordering standards like EDI (Electronic Data Interchange: structured machine-to-machine ordering used by large distributors) are uncommon at this scale.

Delivery method	Build difficulty	Reliability	Notes
Auto-generated email/SMS to supplier	Low	Medium	Works for nearly every supplier. No confirmation loop unless built.
Supplier portal integration	High, per supplier	High	Doesn't scale across thousands of small suppliers.
Draft only; owner sends	Trivial	Depends on owner	Honest, but weakens "one tap."

The product also has to respect supplier constraints: order cutoff times, delivery days, case sizes, and minimum order values. A forecast that says "buy 1.4 cases" isn't an order.

5. Commercial Feasibility

5.1 Does $79/month pay for itself?

The value proposition depends entirely on how much a café wastes today. No baseline data exists, so the figures below are illustrative assumptions only, meant to show the shape of the math:

Café A — discards $15/day of pastry at cost 31% reduction → ~$4.65/day → ~$140/month saved → clears $79 Café B — discards $5/day of pastry at cost 31% reduction → ~$1.55/day → ~$47/month saved → does NOT clear $79

Avoided stockouts (lost sales, lost regulars) and owner time saved add value that isn't captured here. For a low-waste café, though, the case rests on time and stress relief, which is a softer sale. Action: measure real baseline waste in dollars at your paying cafés. That number becomes your ROI story and your best sales tool.

5.2 Go-to-market: the critical gap

The 18-month target is 300 paying cafés. Working backward:

Target: 300 paying cafés in 18 months Current: 4 Net adds needed: ~296 → ~16+ net new cafés per month, every month, before accounting for any churn Waitlist: 120 cafés (origin, location, qualification unknown) If 30% convert: ~36 cafés (illustrative assumption) Gap: ~260 cafés from a channel not yet described

At $79/month, acquisition cost is decisive. CAC (customer acquisition cost: total sales and marketing spend divided by customers won) has to be recovered from subscription revenue. Illustratively, if winning a café costs $400 in founder time, travel, and onboarding, payback takes about 5 months of revenue before any other costs. If it costs $1,000, payback stretches past a year. Small cafés also close at meaningful rates, so long payback periods carry real loss risk.

Plausible channels to test, none of which appear in the materials:

POS app marketplaces (Square and Toast each run app directories): low CAC if discoverable, but visibility is not guaranteed.
Supplier and roaster referrals: local roasters already sell to dozens of cafés and benefit from cleaner orders. This overlaps with your later supplier revenue idea and may be your strongest distribution lever.
Founder-led local sales, city by city: high trust, high CAC, doesn't scale on its own.
Café-owner communities and industry events: credibility channel that fits the "we ran cafés" story.

5.3 Market sizing

The arithmetic is correct: 38,000 × $79 × 12 ≈ $36M. Three problems with how it's framed:

It's a ceiling, not a market. It assumes every independent coffee shop in the US buys. The deck also cites no source for the 38,000 figure. Add one.
The reachable share is smaller. Only cafés on Square or Toast can use the product today, and that subset is unquantified.
$36M is small for venture investors. TAM (total addressable market: revenue if you captured every possible customer) at this size will lead many pre-seed investors to pass unless the expansion thesis (bakeries, juice bars, small restaurants) is credible and sized. That expansion is currently a single sentence. Size it, and show at least anecdotal evidence that those segments share the pain.

5.4 The milestone math for your next raise

300 cafés × $79 = ~$23,700 MRR → ~$284K ARR (ARR = annual recurring revenue, MRR × 12) $750K ÷ 18 months = ~$41,700/month all-in budget

Two founders plus two hires, two new integrations, a supplier marketplace pilot, and an acquisition budget will be tight on roughly $41.7K per month unless founders take well-below-market salaries. The deck shows no allocation or runway. Separately, ~$284K ARR may or may not be enough to raise a seed round, depending on growth rate and retention at that point. Build the use-of-funds table, and decide whether the supplier marketplace pilot belongs in this round or competes with go-to-market for scarce money.

5.5 Supplier referral fees: a trust risk

If suppliers pay Kettlewise, café owners may reasonably ask whether recommended orders favor paying suppliers. Your product depends on owners trusting the draft enough to approve it in one tap. Design any supplier revenue to be transparent, or structure it around efficiency (cleaner orders, fewer errors) rather than supplier selection.

6. Competitive Reality Check

The materials describe competitor categories but name no products. One categorical claim, that restaurant inventory platforms are "built for chains and multi-site groups," is too strong as written. Several serve independent operators. Named alternatives an investor may raise are below. Capabilities and pricing are based on general knowledge and may be outdated. Verify each before presenting.

Alternative	Type	What it does (to verify)	Where Kettlewise may differ
Clipboard, memory, texts to suppliers	Status quo	Free, familiar, zero setup.	The real competitor. Owners must see a dollar or time gain big enough to change habit.
Square inventory features	Indirect / platform	Stock counts and low-stock alerts inside the POS.	No predictive ordering, as far as known. Platform could add it.
xtraCHEF by Toast	Indirect / platform	Invoice capture, costing, inventory within Toast.	Back-office focus. Toast owns the distribution advantage.
MarketMan	Direct-adjacent	Restaurant inventory and purchasing, including ordering. Serves independents as well as groups.	Broader and heavier. Kettlewise must win on simplicity and café-specific forecasting.
BlueCart	Indirect	Ordering marketplace connecting food service buyers and suppliers.	Ordering without demand forecasting, as far as known.
Choco	Indirect	Supplier ordering app for restaurants, with AI-assisted order handling. Active in the US.	Ordering channel rather than demand forecasting. Could overlap with your supplier plans.
Restaurant365, WISK, Craftable	Indirect	Inventory and back-office tools of varying depth and price points.	Generally broader scope. Check which offer forecasting.

Defensibility

"Forecasting tuned to small menus" and "one-tap ordering" are features a well-funded competitor could replicate within 18 months. More durable advantages available to Kettlewise are (1) pooled cross-café demand data that improves forecasts for every new customer, (2) supplier and roaster distribution relationships, if built, and (3) operator credibility that makes owners trust your drafts. Build toward the first two deliberately. They don't exist yet.

7. Team Feasibility

Strong fit. A CEO with 9 years in café operations, including running three locations, and a CTO who built demand forecasting for 200+ stores is close to the ideal founding pair for this problem. The coffee-chain advisor adds operator credibility.

Gaps to close before investor conversations: confirm both founders are full-time (investors will ask on the first call), name the team, and identify who owns sales. Neither founder profile describes selling software to small businesses, and that is the skill the next 18 months depend on most.

8. Risk Register

Risk	Likelihood	Impact	Mitigation
CAC too high for $79/month to work	High	Severe	Run measured channel tests before scaling spend. Prioritize roaster/supplier referral and POS marketplace channels.
Waste reduction smaller than reported once measured	Medium	High	Independent measurement study (§9). Reframe claims to measured figures.
Milk/bean orders inaccurate due to recipe/stock gaps	Medium	High	Document current handling. Add lightweight periodic counts. Track order-adjustment rates.
Square or Toast builds native predictive ordering	Medium	High	Add more integrations. Build cross-café data advantage. Own the supplier relationship layer.
Early churn after novelty fades	Medium	High	Track monthly retention. Interview churned or non-converted cafés.
Café closures erode customer base	Medium	Medium	Factor into retention model. Avoid long CAC payback.
Market too small for venture investors	Medium	High	Size and evidence the bakery, juice bar, and small-restaurant expansion.
Supplier referral fees undermine owner trust	Low (later)	Medium	Transparent fee design, decided before launch.

9. Proof-of-Concept Plan: Next 90 Days

Each test is designed to turn a claim into evidence. Success thresholds are suggested starting points. Set your own before running each test, not after.

#	Test	Method	Success looks like
1	Exit interviews with the 2 non-converters	Structured 30-minute calls: what they expected, what failed, what price would have worked.	A clear, specific reason you can act on.
2	Measured waste study	At new cafés, log discards (count and dollar cost) for 2 weeks before go-live and 4 weeks after. Include milk.	Measured reduction you can state without "owner-reported."
3	Forecast accuracy backtest	Compare forecasts against a naive baseline ("same day last week") using an error measure such as MAPE (mean absolute percentage error). Run with and without weather and events.	Consistent improvement over naive. Each external signal earns its place or is dropped.
4	Order acceptance rate	Track how often owners approve drafts unchanged versus edited, by item category.	High unchanged-approval rate. This is the real test of "one tap."
5	Waitlist qualification	Contact all 120: confirm role, location, POS system, and supplier count. Offer paid onboarding.	A known conversion rate from waitlist to paid.
6	Channel tests with cost tracking	Run 3 channels for 60 days (e.g., roaster referral, POS marketplace listing, founder-led local sales). Log every hour and dollar spent.	At least one channel with a CAC you can recover within an acceptable payback window.
7	Setup time audit	Time the next 5 onboardings, done by someone other than the founders if possible.	Median time that supports or replaces the 20-minute claim.
8	Supplier delivery map	For every paying café, document how each supplier accepts orders.	A clear spec for how approved orders are sent.
9	Retention tracking	Monthly paid-status tracking for all customers.	4 of 4 still paying at month 3, or a known reason why not.

10. Claims to Revise Before Investor Conversations

Current claim	Recommended framing
"31% less pastry waste"	"Pilot cafés reported 31% less pastry waste (owner-reported; measured study underway)."
"Order tomorrow's milk, pastries, and beans in one tap"	Keep as the vision. Be ready to show which categories are accurate today.
"$36M+ annual opportunity"	Cite the source. Present it as a US café ceiling, with a sized expansion market alongside it.
"Restaurant platforms are built for chains"	Name specific products and state precisely where each falls short for a single-location café.
"Setup takes 20 minutes"	Replace with a measured median from real onboardings.
"120 cafés on the waitlist"	State the source, location mix, and qualified count.
Unstated: go-to-market	Add a channel plan with tested or estimated CAC. This is the first question most investors will ask.