The case for Forklet® as a pre-seed investment — the recipe intelligence layer that sits between every cook, every kitchen, and every meal decision on the planet.
Recipes live in silos — blogs, Instagram, TikTok, YouTube, cookbooks, apps — with no connective tissue between discovery, planning, shopping, and execution. Every cook rebuilds the same workflow from scratch, every time.
A cook sees a recipe on Instagram. She screenshots it. She opens a notes app. She types the ingredients. She opens a grocery app. She re-enters the ingredients. She cooks — maybe — if she hasn't lost the screenshot by Thursday. This is the state of the art in 2026. No intelligence layer connects these steps. Forklet® does.
Allrecipes, Food Network, and NYT Cooking each own a slice of discovery. Instacart owns the cart. No one owns the intelligence layer that sits between them. That gap is Forklet®'s entire surface area. The platform that normalizes ingredients, orchestrates kitchen sessions, and learns a cook's preferences becomes the operating system of the home kitchen.
Forklet® is the MCP Foundation for food — a protocol layer that ingests recipes from anywhere, structures them into a personal culinary graph, and orchestrates the full cook-to-table workflow.
Share any URL, paste any text, scan any barcode — Forklet® imports it, normalizes the ingredient units, tags dietary attributes, and adds it to a structured personal recipe graph. The Queue Swipe Deck turns the import experience into a frictionless triage flow: swipe right to import, left to discard, undo anything. The intelligence layer does the heavy lifting before the cook ever opens the fridge.
The KitchenSession is Forklet®'s core invention: a live, stateful cooking context that tracks prep steps, fires push notifications at each stage, and logs outcomes to the Bidirectional PrepTimeline. Two provisional patents cover the Timeline Sync and Session Orchestration mechanisms — the defensible moat around the core workflow loop.
Forklet®'s barcode scan pipeline runs a Meta Llama vision model on a Cloudflare Worker
at vision.forklet.app/scan.
User photos never leave the device and are never stored — a privacy
architecture that differentiates Forklet® from every AI food app in the market.
The rarest kind of technical founder: one who has lived inside the problem professionally, then built the solution end-to-end in code.
Steve spent years as a professional baker managing sourdough schedules, pizza dough timelines, and large-format production runs. He experienced the workflow gap first-hand — the missing layer between recipe discovery and kitchen execution. Founder-market fit doesn't get tighter than this. He is the target user, the domain expert, and the engineer simultaneously.
a16z Speedrun backs exceptional founders building from scratch, with capital up to $1M. Forklet® maps cleanly to each criterion.
No candidacy doc is credible without an honest look at the counterarguments.
Steve is shipping two products (Forklet® + Aqwtr™) simultaneously. The speed of product development is real — but so is the concentration risk. The Speedrun capital is precisely what unlocks the first hire to distribute execution load and remove the single-point-of-failure.
Early App Store traction is promising, but long-term retention in the recipe/cooking category is historically difficult — most apps lose users after the novelty phase. Forklet®'s answer is the KitchenSession loop: the app earns its place in the cook's active workflow, not just discovery. That's the retention mechanism that most recipe apps are missing.
iOS-first means App Store guideline risk (two rejections already resolved) and 30% payment rake on in-app purchases. The roadmap includes web and Android surfaces — the MCP Protocol Layer architecture is platform-agnostic by design, which is the long-term hedge against any single distribution channel.
Forklet® is a live product, with patents pending, a registered trademark, a technical founder with direct domain expertise, and a clearly unclaimed intelligence layer in a $1.5T market. a16z Speedrun is not an exposure play — it writes real checks. This application should be submitted, not debated.