Shiseido builds its first AI-developed mist suncare with VOYAGER
Shiseido has used its AI formulation platform, VOYAGER, to develop a mist-type suncare product that blends SPF performance with fragrance and color. It's slated for summer 2026 under the fibona open innovation program.
The headline: AI didn't just optimize SPF. It helped shape sensory design-fresh, water-light feel-while guiding ingredient choices that align with consumer desire to make UV protection feel effortless.
What's new
- AI platform trained on 100+ years of expertise and 500,000+ formulation data points, including knowledge that's historically hard to digitize (emulsification, ingredient interactions, veteran heuristics).
- Upgraded recommendation engine generates tens of thousands of candidate formulations and narrows to several dozen for chemist review.
- First application where AI balances SPF function with fragrance and color tone to create a mist format that feels simple and enjoyable to use.
- Developed on VOYAGER, created with Accenture; platform built by ITOCHU Techno-Solutions.
How the VOYAGER workflow operates
- Consolidates past research into a searchable and generative system that proposes multi-approach formulations across categories.
- Creates a large candidate pool, scores options, and surfaces a short list for human evaluation.
- Enables junior researchers to produce commercially viable formulations by starting from AI-guided options.
- Supports co-creation: AI proposes structures (here, moving from biphasic beauty oil thinking toward a toner-like feel), while researchers tune for stability, feel, color, and scent.
Consumer insight that drove the brief
People want UV protection without the chore. The team prioritized a light, refreshing mist that's intuitive to use, pairing function (SPF) with uplifting fragrance and subtle color cues. Sensory enjoyment became a core requirement, not an afterthought.
Why this matters for product development teams
- Codify tacit knowledge: Capture chemist heuristics, lab notes, failure cases, and emulsification rules as structured features the model can learn from.
- Optimize across goals: SPF efficacy, sensorial targets (spread, volatility profile, freshness, post-feel), color stability, fragrance compatibility, cost, and regulatory constraints.
- Adopt an AI→human loop: AI creates a wide option space; automated screens prune; focused lab sprints validate; learnings feed back to the model.
- Mix categories intentionally: Use AI to recombine techniques from skincare, makeup, and suncare when the consumer job-to-be-done demands it.
- Level up juniors: Provide AI starting points, clear evaluation rubrics, and formulation version control to speed training and reduce rework.
- Measure what matters: Time-to-first-viable formula, iterations per week, stability pass rate, panel acceptance on feel/scent/color, and BOM cost variance.
- Guardrails: Ingredient safety, fragrance standards (e.g., IFRA), colorant regulations, regional sunscreen rules, and rigorous stability testing (photostability, packaging compatibility, microbial checks).
Process notes from the case
- Start from the consumer moment (using SPF should feel easy and pleasant) and let that steer both function and sensation targets.
- Let AI suggest structure types, then pressure test: in this case, a transition toward toner-like lightness to reduce friction and improve usage frequency.
- Use color and fragrance as cues that prime emotion and repeat use-validated through sensory research and usage studies.
What to watch next
- New categories that blend care, protection, and emotion-especially formats that make daily SPF feel instinctive.
- Broader application of AI-driven cross-category recombination in haircare, body, and makeup.
- How teams integrate AI with LIMS data, ingredient graphs, and lab automation to shorten feedback loops.
For teams building similar capability
- Data foundation: centralize raw materials, process parameters, stability outcomes, sensory panels, and cost data with consistent taxonomies.
- Model strategy: multi-objective scoring that can trade off SPF, sensory KPIs, cost, and compliance.
- People and ops: define reviewer roles, decision gates, and criteria for promoting AI candidates into lab work.
- Tooling: version control for formulations, experiment tracking, and a "design history" for audit and reuse.
VOYAGER shows what happens when institutional knowledge meets generative search at scale: more options, faster decisions, and products that people actually want to use every day.
If your team is upskilling for AI-enabled product development, see curated learning paths by role at Complete AI Training.
Note: Information reflects the state at the time of announcement and may change.
Your membership also unlocks:
