Biotech Drug Discovery AI Needs Institutional Memory
A practical Connectory use case for biotech and drug discovery teams that need shared org memory, PR evidence, and agent governance across scientific software.
Drug discovery companies should use Connectory when AI-assisted discovery depends on software teams, wet-lab scientists, computational biologists, and coding agents sharing the same operational memory. The hard problem is not whether an agent can draft a ligand scoring utility or edit a data pipeline. The hard problem is whether the organization can prove which assay definition, model context of use, owner, SOP, validation expectation, and prior decision governed the change before it merged.
That is a Connectory problem. FDA and EMA guidance now points drug developers toward human-centric governance, risk-based context of use, data governance, documentation, life cycle management, and clear information. Connectory maps to that control gap with existing tools: Genie institutional memory, RepoWatch, OrgWatch, SlopBuster review, Guardian merge gates, and advisory MCP checks.
Why This Is A Connectory Problem
Biotech AI teams already have specialized scientific tools. They have ELNs, LIMS, data lakes, cheminformatics libraries, notebooks, workflow engines, model registries, SOPs, and domain experts. Connectory should not pretend to replace those systems.
The missing layer sits above them. A discovery program may have the assay protocol in an ELN, the raw result in a LIMS, the normalization logic in a Python repo, the feature contract in a data platform repo, the model card in a wiki, the deployment rule in GitHub, and the scientific rationale in a Slack thread or a principal scientist's memory.
Local tooling cannot make all of that context appear in the next pull request. A coding agent reading one repository cannot know that the pIC50 normalization rule changed last month because an external CRO changed assay plates. A reviewer cannot reliably remember which data split is approved for one target family and which is forbidden for another. A compliance lead cannot audit a decision that was never turned into durable, queryable memory.
Connectory's fit is precise: keep scientific systems where they are, but turn the organization-level knowledge around them into typed memory that humans, PR review, merge gates, and agents can all read.
Composite Example: Oncology Discovery Pipeline
Consider a composite oncology biotech. No proprietary source code is needed for the example.
The company is training models to prioritize kinase inhibitors. The pipeline spans seven practical surfaces:
1. Assay results arrive from a CRO into LIMS.
2. Scientists review plate-level quality and annotate exceptions in an ELN.
3. Data engineers normalize assay values and publish curated datasets.
4. Computational chemists generate fingerprints and molecular descriptors.
5. ML engineers train ranking models and uncertainty estimates.
6. Platform engineers deploy an internal scoring service.
7. Program leads use the model output during weekly candidate review.
Now add coding agents. One agent edits the ingestion connector. Another writes a transform for a new assay format. A third updates feature generation. A fourth changes the scoring API. Each change can look reasonable in its own repo. The scientific risk appears only when those changes are read together.
The question leadership should ask is not, "Can our engineers review this?" The better question is, "Can our review process see the same scientific and organizational memory that the program depends on?"
What The Team Can Handle
A strong biotech engineering group can solve many local problems without Connectory.
It can build CI checks. It can add CODEOWNERS. It can enforce unit tests. It can document SOPs. It can require model cards. It can use Great Expectations, dbt tests, MLflow, Argo, Nextflow, Airflow, Dagster, GitHub branch protection, and internal quality review. It can write good local agent instructions.
Those are necessary controls. They are not a shared memory control plane.
The gap appears when knowledge crosses boundaries:
| Discovery Risk | Where It Appears | Why Local Tools Miss It | Existing Connectory Control |
|---|---|---|---|
| Assay schema drift | LIMS connector and normalization repo | CI sees fields, not scientific meaning | Genie stores the current assay decision; SlopBuster checks PRs against it |
| Context of use drift | Model training and scoring service | A model card may not affect API review | Genie keeps the approved context visible to reviewers and agents |
| Ownership gaps | Shared libraries and target programs | CODEOWNERS does not capture scientific owner changes | OrgWatch maps people, teams, repos, and bus-factor risk |
| Agent-authored risk | Any repo touched by AI coding tools | Generic review cannot separate human, agent, and hybrid activity | OrgWatch separates activity patterns and SlopBuster reviews with org context |
| Evidence gaps | Audit prep, vendor review, IND support | Tickets and PRs do not explain why a change was allowed | Guardian exceptions and review outcomes create decision evidence |
| Repeated reinvention | Cheminformatics utilities and pipeline helpers | One repo cannot see another team's accepted pattern | RepoWatch and org context reveal existing utilities and conventions |
This is the part most companies cannot maintain by hand unless they choose to become a software governance vendor themselves. A wiki can explain the rule. It cannot make every future coding agent ask the rule before planning. A model registry can store model metadata. It cannot stop a pull request in a different repo from violating an assay constraint. A Slack channel can answer a question once. It cannot preserve that answer as a typed object that the next review can use.
What Connectory Adds With Existing Tools
Connectory adds the shared operating layer around the discovery software.
Genie is the institutional memory. It stores the people, teams, projects, sources, beacons, questions, events, and relationships that define how the organization works. In a biotech setting, those beacons can represent approved assay assumptions, model context of use, data restrictions, validation expectations, program ownership, and known scientific caveats.
RepoWatch understands each repository before review. That matters when a discovery program has ingestion jobs, feature pipelines, scientific libraries, internal APIs, dashboards, and infrastructure repositories. The same change pattern may be harmless in a prototype notebook and unacceptable in a pipeline feeding a candidate selection meeting.
OrgWatch keeps the cross-repo people picture current. It helps leaders see contributor patterns, bus factor, quality trajectory, and human versus agent activity. In a biotech, that exposes the painful cases: one computational chemist is the only person who understands a descriptor library, or agent-heavy changes are concentrating in a regulated evidence path.
SlopBuster is the review surface. It uses live organizational context when a pull request is opened or updated. The review can ask whether a change matches the repo, the program, the policy, and the prior decision, not only whether the code compiles.
Guardian is the enforcement surface. When a policy requires a block, approval, or exception, Guardian turns the rule into a merge gate with evidence.
Advisory MCP tools let coding agents ask the organization before they act. Existing tools such as check_idea, check_plan, and check_code are advisory, so agents can receive guidance without mutating the graph.
Review-Time Evidence Matrix
For biotech leaders, the most useful product story is not "Connectory writes better code." It is "Connectory makes the right organizational memory available at the moment code is approved."
| Biotech Question | Memory Needed | PR Evidence To Capture | Connectory Surface |
|---|---|---|---|
| Does this transform preserve the approved assay interpretation? | Assay decision, owner, exception notes | Finding, reviewer action, linked policy | Genie plus SlopBuster |
| Is this model still being used in the approved context of use? | Model purpose, target population, output use | PR comment, approval, exception reason | Genie plus Guardian |
| Did the right scientific owner review the change? | Person, team, project, repo relationship | Requested reviewer and outcome | OrgWatch plus review metadata |
| Is an AI agent changing regulated evidence paths? | Human, agent, and hybrid contribution signals | Author pattern and risk tier | OrgWatch plus Guardian |
| Are we repeating a solved pattern from another repo? | Repo purpose, shared utility, accepted convention | Finding that points to existing pattern | RepoWatch plus SlopBuster |
| Can compliance read the story without installing a coding tool? | Policy, decision, event, question, result | Dashboard state and audit trail | Genie dashboard |
This is especially relevant because FDA's 2025 draft guidance on AI for regulatory decision-making centers on a risk-based credibility assessment for a particular context of use. Even when a discovery tool is not itself part of a submission, the discipline is useful: define the purpose, preserve the evidence, manage changes, and show who was accountable.
A 30-Day Pilot Plan
Start with one discovery program, not the whole company. Pick a pipeline where software changes can alter scientific interpretation: assay normalization, feature generation, cohort selection, model scoring, or candidate ranking.
| Week | Work | Output | Connectory Tool |
|---|---|---|---|
| 1 | Identify the critical repos and owners | Project, repo, team, and owner map | OrgWatch and RepoWatch |
| 1 | Capture the current scientific decisions | Assay policy, model context, data restrictions | Genie beacons |
| 2 | Connect review to the highest-risk repos | PR findings grounded in program context | SlopBuster |
| 2 | Define required approval paths | Merge rules for assay, model, and evidence changes | Guardian |
| 3 | Let agents ask before writing | Plans checked against org memory | Advisory MCP checks |
| 4 | Run an evidence replay | One change traced from policy to PR to approval | Genie dashboard and review history |
The pilot should answer four questions:
1. Can a reviewer see the relevant scientific decision in the PR?
2. Can an agent check its plan against the current policy before writing?
3. Can leadership see bus factor and agent activity around the discovery program?
4. Can compliance reconstruct why a high-impact change was allowed?
If the answer is yes, the organization has more than local automation. It has a working memory loop.
FAQ
Q: Is this only for regulated software?
A: No. The strongest starting point is any discovery workflow where software changes can alter scientific interpretation. Regulated evidence paths raise the stakes, but the memory problem starts earlier.
Q: Does Connectory replace LIMS, ELN, model registry, or data catalog tools?
A: No. Those systems remain the systems of record for scientific data and workflows. Connectory preserves the organizational memory around them and uses it during review, merge decisions, leadership visibility, and agent guidance.
Q: Why not just write better SOPs?
A: SOPs are necessary, but they are prose. Connectory turns decisions, owners, exceptions, policies, and questions into typed memory that can be read by reviewers and agents at decision time.
Q: Does Connectory need to store our source code?
A: No. Connectory is designed for security-sensitive engineering organizations and does not need to store source code permanently to preserve org memory, review evidence, and governance context.
Q: What is the strongest first use case for a biotech?
A: Start with an AI-assisted assay normalization or model scoring pipeline. Those workflows have clear owners, clear scientific risk, and enough cross-repo activity to show why local review is insufficient.
Summary
Biotech and drug discovery companies do not need Connectory because they lack scientific tools. They need Connectory because their most important software decisions sit between scientific systems, engineering repos, AI agents, reviewers, and compliance owners.
The existing Connectory toolset gives them a practical control loop. Genie stores the organization's scientific and engineering memory. RepoWatch and OrgWatch keep repository and people context current. SlopBuster applies that memory in pull request review. Guardian enforces the rule when review needs a gate. Advisory MCP checks let agents ask the organization before acting.
That is the part a biotech usually cannot build for itself without turning a discovery software team into a governance platform company. The business value is not another coding assistant. It is shared memory that governs the software changing the science.
References
[1] FDA, "Artificial Intelligence for Drug Development," accessed August 10, 2026. https://www.fda.gov/about-fda/center-drug-evaluation-and-research-cder/artificial-intelligence-drug-development
[2] FDA, "Guiding Principles of Good AI Practice in Drug Development," January 2026. https://www.fda.gov/about-fda/artificial-intelligence-drug-development/guiding-principles-good-ai-practice-drug-development
[3] FDA, "Considerations for the Use of Artificial Intelligence To Support Regulatory Decision-Making for Drug and Biological Products," January 2025 draft guidance. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/considerations-use-artificial-intelligence-support-regulatory-decision-making-drug-and-biological
[4] Wilkinson, M. D. et al., "The FAIR Guiding Principles for scientific data management and stewardship," Scientific Data, 2016. https://www.nature.com/articles/sdata201618
[5] Pistoia Alliance, "From Data Access to Data Readiness," published June 18, 2026. https://pistoiaalliance.org/resource-library/from-data-access-to-data-readiness/
[6] FDA, "Part 11, Electronic Records; Electronic Signatures - Scope and Application," September 2003 guidance. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/part-11-electronic-records-electronic-signatures-scope-and-application