← ArticlesEssay · 02
Essay · 02May 20264 min read

Why design food proteins at all.

Protein design left the lab faster than food noticed. We are pointing the stack at meat, dairy, and everyday nutrition.

Null Bites·May 2026

Protein design is already changing drug discovery. Food is still mostly using molecules that plants and animals happened to give us. Null Bites exists because that gap is too large to ignore.

Where the AlphaFold dividend went · 2020 to 2025
Therapeutics
← Food
Drug discovery, antibody design, target validationNutrition · ingredient design

Rough estimate of where post-AlphaFold protein-design investment has landed. Almost the entire cycle has gone to therapeutics.

Pharma absorbed most of the AlphaFold dividend. The largest leap in molecular design since recombinant DNA went straight into drug discovery and mostly skipped the dinner plate. We are taking the same stack and changing the target: bite, melt, gel, digestibility, cost.

The loop, before and after
2020Before

365 days

per protein candidate

  • Throughput1 candidate / cycle
  • ValidationBench only
  • Cost$50K+ per attempt
  • OwnerWet-lab specialist
2026After

1 afternoon

per design batch

  • Throughput1,000s of candidates
  • ValidationSimulator → bench
  • CostPennies per candidate
  • OwnerAutonomous loop

The hard part moved from making one candidate to choosing which candidate deserves bench work. We treat each food protein as something we can write, score, reject, and rewrite. The company that learns fastest wins.

The bottleneck moved from the lab bench to the design loop. That is the layer we are building around.
What we redesign

A lot of what people feel in food comes from protein behavior: texture, melt, set, chew, foam. Those are no longer fixed properties.

Myosin

Meat texture

The fiber that makes a steak bite like a steak.

Ovalbumin

Set & structure

The fold that turns liquid egg into a custard.

Casein

Stretch & melt

The micelle that makes cheese pull on a pizza.

Lactoferrin

Iron uptake

The carrier that delivers iron the body actually uses.

How we ship
  1. 01

    Write the molecule

    Sequence-level control.

  2. 02

    Reject quickly

    Cheap design passes before bench work.

  3. 03

    Pick the right route

    Fermentation, partner formulation, licensing, or own brand.

We expect some molecules to fail. That is the point of the loop: kill weak candidates in software, send fewer to the bench, and let the food team spend time on the ones with a real shot.

Today

Food is inherited from evolution and patched by industry.

2040

Every gram can be designed around nutrition, cost, and taste.

We are designing food proteins in public, today.