Cockpit Story Ch 6 · Three pitches. Three audiences. Now.
External signals · impact on AVY
Loading external signals…
1 2 3 4 5 6 7 8
CHAPTER 6 · WHERE THE MONEY IS

Three pitches. Three audiences. Now.

Each is board-ready · funded this quarter · 90-day proof point.

SITUATION TODAY
Why sequence matters · who funds what
Seven use cases qualify. Three are board-ready today; four are next-wave once these three prove out. Sequencing matters — each pitch lands with a different exec who controls a different budget. Run them in parallel, not in series.
Head of R&D
Pitch 1 owner
Budget: R&D / programme · 2025 cap $11M
CIO / CDO
Pitch 2 owner
Budget: enterprise data · 2025 cap $8M
CSO
Pitch 3 owner
Budget: ESG + regulatory · 2025 cap $4M
90 days
Time from yes → 1st proof
Each pitch has the same cadence
$12–18M
Combined Y1 ROI
Across the three
BENEFIT
3 pitches · 3 owners · 90-day proof each · $12–18M Y1
Each lands with a different exec · funded this quarter · proves out in parallel
PITCH 1
Formulation & Next Best Experiment
−30 to −50%formulation cycle time
Bayesian optimiser turns historical PfS data into the next experiment.
AI / ML / LLM
Bayesian opt (BoTorch)Gaussian Process surrogateActive learning DoEqNEHVI multi-objective acquisition
DATA SOURCES
PfS · 8 formulations · 13 results · 21 prior runsSupplier monomer specs (Stora Enso · BASF · Wacker)Customer requirement library (Diageo · Pernod · Nike)Plant runnability constraints
→ +25 pts first-pass success · $3.2M / programme · 11 weeks back
PITCH TOHead of R&DMaterials Group · Hassan Rmaile org
Set up 60-min R&D session →
PITCH 2
Scientific KG & R&D Copilot
1grounded answer · with citations
Every lab, patent, PDF, customer requirement — one query, cited.
AI / ML / LLM
Property graph (Neo4j-class)Hybrid RAG (BM25 + embeddings)Azure OpenAI gpt-4.1Provenance per token
DATA SOURCES
PfS Experiments + ELN free-text (3,100/yr)Patent corpus (USPTO + EPO · 28k AVY-adjacent)SharePoint + lab PDFs (~120k docs)Customer requirement spec library
→ Tribal knowledge preserved · onboarding halved · zero duplicate research · $2.8M/yr saved
PITCH TOCIO / CDOCross-functional · spans every BU
Set up 45-min CIO session →
PITCH 3
Sustainable Materials Intelligence
Pre-labsustainability scoring
PFAS · SVHC · LCA · supplier ESG · ESPR readiness — before the beaker.
AI / ML / LLM
Multi-source hazard classifier (sentence-transformer)LCA inference (regression + KG)LLM regulation parserSubstitution recommender
DATA SOURCES
PfS Material master + composition (13 materials)REACH SVHC · FDA · ESPR · Prop 65 feeds (auto-updated)Supplier ESG ratings (EcoVadis + AD audits)AD 2030 ESG baseline + targets
→ EU DPP-ready · 100% pre-lab screening · $0 regulatory surprise · sustainability score on every Sample
PITCH TOChief Sustainability OfficerAligned with AD 2030 ESG goals
Set up 30-min CSO session →
NEXT-WAVE · once these 3 prove out
Root-cause AI · failure × supplier × climate correlationEnterprise scientific search · self-serve GenAIRegulatory intelligence · auto-substitution on new bansDigital thread · beaker → atma.io-tagged SKU
HOW IT WORKS
What each pitch buys · what proves it
1
Pitch 1 → R&D leadership
Bayesian Next Best Experiment engine, trained on PfS history. 90-day proof: HELIX-v1.5 hits peel-curve target on first try.
▣ BoTorch + GP + qNEHVI · live on HELIX wedge
2
Pitch 2 → CIO / CDO
R&D Copilot grounded in PfS + patents + PDFs + customer requirements. 90-day proof: 80% of 'has anyone done this?' questions answered with citations in <2 min.
▣ Knowledge Graph + hybrid RAG + Azure OpenAI gpt-4.1
3
Pitch 3 → CSO
Every PfS formulation auto-scored on PFAS / SVHC / LCA / supplier ESG / ESPR before lab. 90-day proof: zero PFAS-restricted material lands in a new programme.
▣ Hazard classifier + LCA inference + LLM regulation parser
BEFORE / AFTER
What changes for each persona
DimensionTodayWith SCIKIQ + ThermoΔ
Head of R&D Trial-and-error formulation Bayesian-guided NBE −40% cycle
CIO / CDO 12-tab scientist workflow 1 cited answer in 90 sec −$4.3M tab tax
CSO Post-hoc ESG audit Pre-lab ESG scoring EU DPP-ready
Combined Y1 $0 from this stack $12–18M ROI Payback <8mo
WHO WINS
"We don't have to choose. Each pitch lands with a different exec and a different budget. They prove out in parallel. By Q4, all three are funded."
Three execs · three budgets · same lever · R&D · CIO · CSO
⚠ IF WE DON'T
Cost of running them in series instead of parallel
  • Pitch 1 funded alone → R&D speed up, but ESG still audited post-hoc · customer trust erodes
  • Pitch 2 funded alone → great answers, but no NBE recommendation · scientists still bottlenecked on DoE
  • Pitch 3 funded alone → sustainability scored, but no upstream R&D acceleration
  • All three in parallel → compounding effect · $12–18M Y1 vs ~$4M each in isolation
3pitchesBoard-ready today
3ownersR&D · CIO · CSO
90dto proofPer pitch · in parallel
$12–18MY1 ROICombined · payback <8mo
Four more sit in the next wave once these three prove out. Start with the three.
Three pitches. Three owners. Funded this quarter. $12–18M Y1.
AI / ML / LLM IN THIS CHAPTER
See per-pitch AI/ML/LLM stacks in the cards below
DATA SOURCES
See per-pitch data sources in the cards below