Dunback Meadow

Infrastructure for agentic commerce: your AI agent finds, calls, and pays other agents on our live x402 rail. We are entering through scheduled breakfast — a market delivery apps cannot serve.

The v1 platform is in production. Live today: no-code agent creation, the marketplace, the multi-agent workflow engine, the x402 payment facilitator (free and public), and the safety guardrails. Close to 100,000 SKUs are already purchasable by an agent.

Pre-revenue by design: the seed round buys the first transactions, not the first build.

Go-to-market is underway with a hospital for a scheduled-breakfast launch. Founder field research at a Stanford hospital confirmed the gap: no hot breakfast on site before 7am, and delivery drop tables already built.

Entry segment: households with a hard morning deadline and no time to cook. We find them pre-clustered at anchor sites - hospital staff at 6:45am shift change, senior living residents where the adult child holds the account and pays.

US morning foodservice is $80-100B (McDonald's breakfast ~$14B, Dunkin' morning ~$9B). Virtually none of it is delivered on a schedule.

Supply side: restaurants with idle morning capacity, and developers selling agent capabilities.

Then the platform widens - groceries, every retail category, and agent-to-agent Skill calls.

AI agents can now act on our behalf, but commerce has no way to let them transact.

Online commerce still assumes a human browsing a catalogue. An agent that could run the whole errand has nowhere to go: no way to discover what a business sells, call it, or pay for it. Businesses cannot be found by another agent. Developers have capabilities and no surface to sell them on.

Payment rails block it too. Cards charge a fixed fee per transaction, so agent calls priced in fractions of a cent cannot exist.

In our entry market: delivering breakfast costs more than the meal.

Dunback Meadow is the infrastructure for agentic commerce.

People state an intent to their personal AI agent, created on our platform with no code, and the agent executes it by calling other agents. Businesses list Resources Agents to sell through. Developers list Skill Agents and earn per call. A workflow engine orchestrates multi-agent tasks; guardrails handle identity, spending caps and human oversight.

Payments settle peer-to-peer in USDC on our x402 rail, from a self-custody wallet we never control. On-chain settlement makes sub-cent agent-to-agent payments viable. Cards cannot.

Two layers.

Infrastructure: agent payment standards are emerging - Google's AP2, the Agentic Commerce Protocol, Visa and Mastercard agent pilots. They move money between known parties. None is a marketplace where agents discover and call each other, and none settles sub-cent payments. Ours is live today.

Wedge: DoorDash (55-67% US share), Uber Eats, Grubhub. All built for impulse dinner ordering. They let you schedule an order but do not batch scheduled food orders, so breakfast stays unserved.

The real alternative is the status quo: people browsing apps, or skipping breakfast.

Built, not planned. The v1 platform is in production: agent creation, marketplace, workflow engine, x402 facilitator, guardrails.

Architecture, not pricing. Cards assume a human cardholder who can dispute. There is no human in an agent-to-agent call, and an agent cannot get a merchant account. Our rail gives instant finality, hard spending caps, and a payout address any developer receives to in seconds.

A wedge with proof. Scheduling makes demand knowable, so orders batch: $2 a drop, not $8. The restaurant earns $6.97 versus $5.40 on DoorDash, from morning one.

Revenue: 10% merchant commission (vs 25-30% incumbents), live now. Household membership $9.99-14.99/month, turning GMV into recurring revenue. Skill Agent call fees from month 12. Merchant premium tools post-Series A.

Costs: we carry no logistics. Merchants deliver with their own staff, or drivers claiming guaranteed 5:45-7:45am route blocks on our platform. Variable cost per order is agent inference, settlement and refund reserve: about $0.20.

Contribution is positive from order one: $2.95 to us, $6.97 to the merchant, vs $5.40 on DoorDash. Fixed costs are engineering and zone launch.

Model: marketplace take rate plus subscription. 10% merchant commission (vs 25-30% incumbents), household membership $9.99-14.99/month, Skill Agent call fees from month 12. Merchants deliver, so we carry no logistics cost.

Channels: anchor sites - break-room sign-up at hospitals, resident sessions at senior living, adult-child outreach. One institutional conversation replaces a hundred consumer ones. CAC $18-28, then referral.

Metrics: drops per driver-hour, on-time-in-window, orders per household per month, merchant contribution per order, CAC payback, autonomous completion rate.

$3M seed, 26-30 months of runway. Asset-light: merchants deliver, so no logistics burn.

45% engineering. Scheduling and route-manifest engine, agent runtime and no-code builder, merchant integration, wallet and settlement. Five engineers.

20% household acquisition. Anchor-site sign-up, referral, adult-child outreach.

17% anchor partnerships and zone launch. Free Resource Agent build, morning capacity co-design, handoff design.

6% density bridge. Capped, tapering support for early zones.

8% legal and compliance. 4% reserve.

Target: 20+ zones, ~$3M net revenue, Series A ready.

$3M seed at $15M post-money, roughly 20% economic to new investors. Delaware C-corp, HQ Palo Alto.

Dual class: Class A carries 1 vote, Class B 0.1 vote. The founder holds all Class A. Investors and employees receive Class B, so economic participation is shared without fragmenting control of a single-founder company.

Cap table: founder, plus three family SAFEs totaling $43,905 at a $2M cap, converting automatically at this round into under 2% fully diluted. No notes, no prior priced round.

The NFT patent is being assigned from founder to company ahead of close.

Lead investor welcome.

Density. Below ~8 drops per driver-hour the merchant earns less than on DoorDash and quits. We gate every zone on committed demand before launch and track merchant contribution weekly. Kill criterion at month 9.

Demand. Breakfast may be small because demand is small, not because it is unserved. A month-6 retention gate tests this before expansion capital is committed.

Incumbents. Autonomous delivery will eventually erase our cost advantage. We have 24-36 months, and spend them building merchant morning capacity, household habit and standing agent authorizations - what survives.

No

Won the Global 100 "Best Blockchain Services Business of the Year 2026".

Patent:
Method and Platform for creating Non-fungible Tokens with Built-in Terms
US 12,530,684 B2 · Issued Jan 20, 2026