Three regulatory pressures are forcing CPG food reformulation at the same time, and none of them are optional.
FDA’s phase-out of six petroleum-based synthetic dyes — Red 40, Yellow 5, Yellow 6, Blue 1, Blue 2, Green 3 — targets end of 2026. An expected federal definition of ultra-processed foods under MAHA is likely to force reformulation review across most of the functional category.
Class-action suits are now targeting functional claims like “prebiotic” and “gut healthy.” They’re suing not because an ingredient is absent but because there isn’t enough of it to do what the label promises.
Together, they represent more reformulation volume than most CPG R&D teams can absorb on their current timelines, and the demand for CPG food reformulation tools that can compress the cycle has never been more urgent.
Six startups in the Halo network are built to help. Each addresses a distinct part of the problem — not as a general AI pitch, but as a specific answer to what the current regulatory environment is forcing R&D teams to solve.
Six solutions for the backlog
Alchemyst: R&D literature intelligence, compressed
Before any formulation work begins, someone has to know what’s been tried, what’s been approved, and what consumers have already rejected. On a complex dye replacement project, that literature review alone can consume weeks.
Aleksandar Boskovski’s Alchemyst compresses that front end into an agentic workflow. The system indexes 4.3 million scientific papers, the full USFDA and EFSA regulatory canon, USDA and FooDB composition data, and 16 million consumer reviews covering 1.3 million products. The pitch is direct: your scientists are formulation experts, not literature specialists.
→ See Aleksandar’s full profile on Halo
amofor — Predictive natural colorant stability modeling
Natural colorants are less stable than synthetic dyes under heat, pH swings, and light. Predicting how a candidate pigment will behave in a specific food matrix, under a specific processing regime and across a specific shelf life, has traditionally required extensive bench testing. That’s where the months go.
Christian Lübbert is the Managing Director of amofor, which brings pharmaceutical-grade thermodynamic modeling (PC-SAFT) to food problems: natural colorant stability, creatine solubility in beverages, functional ingredient bioavailability. The team has listings on Halo at TRL 9 for model-guided stabilization of functional ingredients in food systems, deployment-ready rather than theoretical.
→ See Christian’s full profile on Halo
Biovit Technologies — Stabilized functional ingredients
Creatine has moved fast — from supplement aisles to ready-to-drink retail in roughly 18 months, with major players shipping into 10,000-plus stores in early 2026. The formulation problem hasn’t kept pace. Creatine degrades to creatinine in acidic aqueous environments, which describes most RTD formats, and stability is what’s standing between the market opportunity and mass-market execution.
Nicholas Deeney’s Biovit Technologies is working on exactly that problem: stabilized creatine delivery systems designed for food and beverage applications. Where amofor models the physical chemistry of stability, Biovit is building formulated solutions for it. The company is actively seeking CPG partners through Halo.
→ See Nicholas’s full profile on Halo
Tarion.AI — Computational sugar reduction and creatine protection
Ed Sosa’s path into food modeling started with a first-principles question about why sugar tastes the way it does. The result — developed under the original ZeroIN Foods banner — was a next-generation sweet-tasting dietary fiber matching sugar’s sweetness at roughly 10x lower cost, without high-intensity sweetener aftertaste. The project earned recognition at the Swiss Foodtech competition and through the Creative Destruction Lab.
His current entity, Tarion.AI, extends the same computational methodology into a digital-twin platform for sweetener refining operations and CreaLock, a microcrystalline wax micro-reservoir system that protects creatine from moisture until consumption. If the UPF definition lands as expected and triggers reformulation review across the functional category, sugar reduction without taste compromise is one of the first workstreams in the queue.
→ See Ed’s full profile on Halo
DyeConverter — Screening natural dye replacements before bench work begins
Kelly Anderson, Founder of DyeConverter, built a predictive reformulation platform for screening natural colorant candidates against the constraints that actually determine success: pH environment, thermal processing, shelf-life targets, dosage ranges, and regulatory compliance across U.S., EU, and Canadian markets — all before bench testing begins. The output is a prioritized set of viable alternatives sized to reduce trial-and-error cycles. The company holds one patent and is actively seeking CPG R&D partners for sponsored pilots.
→ See Kelly Anderson’s full profile on Halo
Raw Edge — Fermentation-based reformulation that survives consumer testing
Reformulation projects fail at consumer acceptance more often than they fail in the lab. Reducing sugar, salt, or fat without breaking the sensory expectations that drove adoption is a harder problem than the chemistry suggests and it’s the problem the UPF backlash is about to force at scale.
Kristel Vene leads the food science and technology research group at Tallinn University of Technology, 15+ years studying how flavor perception, sensory expectation, and fermentation shape food acceptance. She co-founded Raw Edge, an EIT Food-backed startup producing fermented low-sugar functional beverages with lactic acid bacteria, while licensing the technology and production model to local manufacturers globally. The fermentation approach matters beyond the startup itself: fermentation-based sugar reduction can create flavor complexity and sensory satisfaction that direct substitution often cannot, particularly in beverages. As brands search for reformulation paths that don’t crater taste, that distinction becomes increasingly important.
→ See Kristel’s full profile on Halo
What the 2026 reformulation crunch means for CPG R&D
The pattern across all six is the same: specialized tools absorbing the predictable portions of formulation work and concentrating human expertise on the problems that actually require it.
That trajectory is familiar. In pharma, contract research organizations absorbed the predictable parts of preclinical work over a decade ago — not because large pharma companies lacked the capability, but because the volume and pace of work exceeded what internal teams could absorb without structural change. In semiconductors, the same logic drove the separation of design from fabrication. Food CPG is at that inflection now.
Brands that clear their dye backlog in 2026 using external partnerships are building process infrastructure that will outlast the deadline. No single CPG will build all of these capabilities internally, nor should they. The question isn’t whether to use external tools. It’s which ones are ready to deploy now.
