{"resourceId":"andromeda2-evidence-grounded-autonomous-lab-20260916","versions":[{"version":"external-141b06f08e475be9d0aba4a671d4459560a81fbb03a2ed0b04461b056e88de6e","resource":{"id":"andromeda2-evidence-grounded-autonomous-lab-20260916","title":"Andromeda 2: evidence-grounded agentic lab hits 50% high-AUC formulations vs 17% / 2% baselines (paclitaxel)","organization":"Intrepid Labs (Toronto) — arXiv preprint","sector":"Industry autonomous laboratory / pharmaceutical formulation R&D","geography":"Canada (Toronto); transferable methods with limits","publishedAt":"September 16, 2026 (arXiv)","publicationDate":"2026-09-16","eventDate":null,"sourceName":"arXiv 2609.19099","sourceLabel":"Academic preprint (arXiv)","sourceUrl":"https://arxiv.org/abs/2609.19099","evidenceClass":"academic-research","outcomeClass":"mixed","topics":["developers-agents","operating-model","knowledge-work"],"finding":"Andromeda 2 is an agentic formulation system that reasons over structured in-house experimental evidence and drives a miniaturized autonomous SEDDS lab. At a matched budget of 96 paclitaxel formulations, it achieved a 50% high-performance hit rate versus 17% for probabilistic Andromeda 1 and 2% for wet-lab DoE; 12 vs 6 vs 0 formulations met all four TPP objectives. Median AUC10–240 was 70.1 / 12.0 / 3.5 mg·min/mL. Ablation: withholding historical in-house evidence cut mean AUC by 34% (62.0→46.2). Authors note FaSSIF is not bioavailability; single campaign limits inference; industry lab setting.","sledRelevance":"University and national-lab AI-for-science programs evaluating autonomous labs need measured hit-rate gains from evidence grounding—not agent demos alone—and must separate in-vitro assay gains from clinical claims.","evidence":"arXiv abs 2609.19099 opened via HTTPS; hit rates, TPP counts, median AUCs, and 34% ablation taken from abstract/abs page text (paper dated 16 Sep 2026).","architectureImplications":"Agentic planner + deterministic feasibility engine + robotic wet-lab loop; historical evidence store separate from campaign measurements; human scientist reviews batches without redesigning compositions.","governanceImplications":"Evidence provenance and ablation transparency matter for claiming agent value; university transfer needs data-governance for reusable experimental evidence across PIs.","securityPrivacyImplications":"Formulation IP and experimental databases are sensitive; paper uses fully in-house models (no external commercial API), which is a deployment choice campuses may or may not mirror.","caveats":"Industry (Intrepid Labs) not a university lab; FaSSIF ≠ oral bioavailability; single campaign per arm; preprint not peer-reviewed journal version; paclitaxel-specific design space.","streamIds":["research"],"roles":{"sales":"Sales framing: Research leaders buying autonomous-lab or agentic-formulation capability need measured hit-rate lifts at fixed experimental budget, not chatbot demos. Andromeda 2’s paclitaxel benchmark (50% high-AUC hits vs 17% / 2%; 12 full-TPP formulations; 34% mean-AUC lift from historical evidence) supports a bounded value story around evidence-grounded agentic search. Do not sell clinical bioavailability or guaranteed cross-API generality—authors explicitly limit FaSSIF and note single-campaign statistics.","engineering":"Engineering framing: Fit is an agentic controller over a constrained composition grid with deterministic feasibility checks and robotic batch execution. Integrate an evidence store, assay data capture, and human safety review gates before execution. Prerequisites include structured historical experiments, clear TPP metrics, and executable composition constraints. Proof-of-value should be matched-budget hit rates plus an ablation with and without historical evidence—not agent narrative alone.","delivery":"Delivery framing: Stand up evidence schemas, a batch-review SOP, and a matched-budget evaluation harness before claiming agent superiority over DoE or Bayesian baselines. Ownership: lab platform ops own robots and assays; scientific leads own the TPP; agents only propose under feasibility gates. Risks include over-generalizing from one API, treating FaSSIF as pharmacokinetics, and skipping replicate campaigns needed for campaign-level inference."},"roleDetails":{"sales":{"customerProblem":"Formulation and autonomous-lab teams spend scarce wet-lab budget without learning from prior experimental evidence.","stakeholders":["VPR / research computing for AI-lab platforms","Core facility directors","PI formulation / chem teams","Research integrity / data governance"],"discoveryQuestions":["Is historical experimental evidence structured and reusable across campaigns?","What matched-budget baselines exist (DoE, Bayesian optimizer, prior agent)?","Which TPP metrics are assay-valid vs translational?","Are models and evidence kept in-house or via external APIs?"],"valueHypothesis":"Evidence-grounded agentic design can raise high-performing hit rates at fixed wet-lab budget versus probabilistic optimization or DoE when historical evidence is available.","potentialEngagement":"Matched-budget pilot on one hard API with ablation of historical evidence and pre-registered TPP.","unsupportedClaims":["In vivo bioavailability improvement","Guaranteed gains on every API","Peer-reviewed clinical endpoints","University-lab identical ops cost"]},"engineering":{"fit":"Agentic proposal/critique loop over constrained SEDDS design space with robotic execution and feasibility engine.","architectureIntegration":"Wire evidence DB, assay LIMS, composition validator, and batch review UI; keep labels/metrics outside unconstrained agent writes where possible.","prerequisites":["Structured historical experiments","Robotic liquid handling","Defined TPP assay","Feasibility constraint engine","Human batch review SOP"],"constraints":["Industry preprint setting","Single-campaign statistics","FaSSIF endpoint only","API-specific design space"],"security":"Protect formulation IP and experimental databases; decide on-prem vs external LLM APIs before campus deployment.","proofOfValue":["Matched-budget hit rate vs baselines","Full-TPP count","Evidence ablation delta","Replicate campaign if claiming generality"]},"delivery":{"work":"Implement evidence schema, agent tools, feasibility checks, and evaluation harness; run matched-budget pilot.","dependencies":["Lab robotics","Assay analytics","Data engineering for evidence store","Scientific reviewers"],"ownership":"Platform team owns robots/software; PIs own scientific claims; research integrity owns disclosure of AI role.","skillsAdoption":"Train scientists to set TPPs and audit agent rationales; train operators on batch review without silent redesign.","governanceCheckpoints":["Pre-register TPP and budget","Ablation plan review","Endpoint interpretation review (no PK overclaim)"],"acceptanceCriteria":["Documented hit-rate vs baseline at fixed N","Evidence ablation completed","Human review logs retained"],"risks":["Assay misread as clinical success","Evidence store biased to past APIs","Single-run variance","Vendor lock on closed evidence"]}},"retrievedAt":"2026-09-26T21:59:40-05:00","enrichedAt":"2026-09-26T22:08:00-05:00","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Raises need for data-curation and assay specialists who can maintain reusable experimental evidence—not only GPU engineers.","procurementImplications":"Buy or build evidence platforms and evaluation harnesses with ablation requirements; do not procure agent seats without matched-budget acceptance tests.","operatingModelImplications":"Shift from one-off campaigns to accumulating institutional experimental evidence that improves subsequent agentic search—requires cross-PI data sharing norms.","sourceVerification":{"openedUrl":"https://arxiv.org/abs/2609.19099","referenceExcerpt":"Andromeda 2 achieved a 50% high-performance hit rate versus 17% for Andromeda 1 and 2% for DoE","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}