Governed venture-development control plane for medical products
From a medical invention to an evidence-backed sequence of decisions
Decidian Intelligence is a governed venture-development control plane that converts a medical invention into an evidence-backed sequence of clinical, scientific, intellectual-property, regulatory, engineering, manufacturing, commercial, and capital decisions.
One secure intake becomes an explainable decision, an execution-ready blueprint, and, only for qualified projects, a permissioned path to experts, partners, and capital.
Decidian supports decisions. It does not make binding clinical, regulatory, legal, engineering, quality, or investment decisions, and it does not replace the professionals who do.
The problem
Fatal product problems are usually discovered late, and lateness is what makes them fatal
A classification error found after design freeze, a blocking patent found after tooling, a payment route that never existed, an endpoint no study can measure: none of these are exotic. They are ordinary findings that arrive after the money that could have answered them has already been spent.
- Medical-product development is a chain of interdependent decisions, and the common failure is loss of context rather than lack of expertise.
- Work is distributed across consultants, CROs, counsel, laboratories, manufacturers, spreadsheets, and static decks, which produces inconsistent assumptions and weak decision lineage.
- Different pathways impose different evidentiary standards, so a single undifferentiated workflow is unsafe.
- Investors face the mirror image: uneven materials that force regulatory, clinical, reimbursement, technical, and capital risk to be reconstructed from scratch.
The model
One secure intake, then decisions instead of documents
You submit material once. It is scanned, hashed, receipted, and normalized into a single project record that every later stage reads from, so the same facts are never re-extracted, re-argued, or quietly changed.
Secure single intake
Files are scanned, parsed in a sandbox, hashed, and receipted, and data rights and purpose are accepted before anything is analyzed.
Evidence-backed decisions
Every material conclusion links to the exact source span behind it, with authority, directness, recency, and applicability recorded separately.
Explainable, not oracular
Alternatives, assumptions, contradictions, and confidence are part of the record, so a reviewer can disagree with the reasoning, not just the answer.
Founder control of data and IP
You retain raw data and project IP. Decidian holds a processing right, and every disclosure is scoped, logged, and revocable.
Workflow sequence
Twelve state transitions from secure ingestion to outcome recalibration. The sequence advances one state at a time and can be paused or stepped manually.
Step 1 of 12 · Files scanned, hashed, receipted
Secure ingestion
Uploads are virus-scanned and parsed in a sandbox. Each file receives a hash and a receipt before anything reads it.
Text equivalent
- Secure ingestion — Files scanned, hashed, receipted. Uploads are virus-scanned and parsed in a sandbox. Each file receives a hash and a receipt before anything reads it.
- Extraction and normalization — Facts written to one record. Extraction writes into a normalized project schema. High-impact facts are held for confirmation before any stage depends on them.
- Evidence retrieval — Sources retrieved under policy. Analysis retrieves project and reference material under an explicit source policy. Ungrounded statements are not produced.
- Source-to-claim linking — Every claim carries its span. Each claim is bound to the retained source span, with authority, directness, independence, recency and applicability recorded separately.
- Parallel domain analysis — Eleven domains evaluated. Domains are analysed independently so a strong domain cannot mask a blocked one.
- Contradiction detection — Conflicts recorded, not averaged. Conflicts between sources, or within your own documents, are surfaced as records with both positions retained.
- Agent disagreement — Escalated, not resolved silently. When bounded agents disagree, the disagreement is escalated with both proposals intact.
- Human review — Named authority acts. A reviewer with the relevant authority approves, adds conditions, dissents, or overrides with a written rationale.
- Decision-gate transition — State recorded with version. The gate produces GO, CONDITIONAL GO, HOLD, PIVOT, STOP or ABSTAIN, versioned and appended to the ledger.
- Blueprint generation — Affected sections regenerated. Approved decisions assemble into the commercialization blueprint; changed decisions regenerate only the affected sections.
- Permissioned matching — Disclosure only with a grant. Investors see a project only where eligibility, mandate fit and an explicit founder grant all hold.
- Outcome recalibration — Forecasts scored against reality. Recorded outcomes are compared with forecasts and used to recalibrate by product type and stage.
Environments
Four environments, one protected core
Innovators, investors, expert reviewers, and execution partners work in role-specific environments over the same evidence, policy, decision, and governance core.
Innovator Venture Room
MedTech, diagnostic, SaMD, combination-product, and pharmaceutical innovators
Turn one secure intake into traceable decisions, a living commercialization blueprint, and disclosure you control.
Read the specificationInvestor Intelligence Portal
Investors, corporate venture teams, family offices, venture studios, and strategic acquirers
Mandate-driven discovery over evidence-linked projects, with the reasoning behind every match visible.
Read the specificationExpert Control Room
Clinical, scientific, regulatory, IP, quality, engineering, reimbursement, and manufacturing experts
Review where it matters most, with sources one click away and your authority recorded.
Read the specificationPartner Network
Laboratories, manufacturers, CROs, design firms, commercialization partners, and qualified service providers
Qualified introductions after diligence, driven by validated requirements rather than a directory listing.
Read the specificationDecidian operating core
Four environments, one protected core. Select an environment to see exactly what it reads from and writes to the core.
Core contents
- Evidence and provenance
- Sources, spans, hashes, contradictions, and quality dimensions.
- Policy and permission
- Data rights, disclosure scope, role authority, retention.
- Decision and dependency
- Decision objects, gates, assumptions, invalidation edges.
- Governance and audit
- Approvals, dissent, overrides, versions, immutable ledger.
- Environment
- Selected path
- Protected core
Text equivalent
- Innovator Venture Room (Founders and product teams) reads project record, decisions, blueprint, gaps, disclosure log; writes intake material, confirmations, assumptions, permissions. The core holds evidence and provenance, policy and permission, decision and dependency, governance and audit.
- Investor Intelligence Portal (Professional and institutional investors) reads permissioned decision evidence, readiness vector, match rationale; writes mandates, diligence requests, expressions of interest. The core holds evidence and provenance, policy and permission, decision and dependency, governance and audit.
- Expert Control Room (Domain reviewers with authority) reads review queue, agent proposals, evidence spans, contradictions; writes approvals, conditions, dissent, overrides with rationale. The core holds evidence and provenance, policy and permission, decision and dependency, governance and audit.
- Partner Network (CROs, CDMOs, labs, notified-body advisors) reads scoped requirement packages, capability requests; writes credentials, availability, responses, delivered outcomes. The core holds evidence and provenance, policy and permission, decision and dependency, governance and audit.
Operating workflow
Twelve governed stages
Each stage answers one decision, produces a gate result, and records the evidence and authorization behind it.
- 01
Secure intake and normalized project record
Do we hold the rights to this material, and can it become one reliable project record?
- 02
Product and jurisdiction classification
What kind of product is this, and which authority governs it where you intend to sell?
- 03
Clinical need and scientific plausibility
Is the clinical problem worth solving, and is the proposed mechanism scientifically plausible?
- 04
Market, adoption, pricing, and reimbursement
Who buys it, who is paid for it, and through which payment mechanism?
- 05
Patent landscape and preliminary freedom-to-operate hypotheses
Is there protectable space, and which third-party rights need qualified legal review?
- 06
Regulatory pathway and scenario mapping
Which submission pathway is defensible, and what evidence does each scenario demand?
- 07
Engineering and design controls
Can the intended use become a safe, verifiable design under design controls?
- 08
Manufacturing, supply chain, DFM, and quality readiness
Can this be made consistently, compliantly, and at a cost the market supports?
- 09
Preclinical, clinical, and verification-and-validation evidence strategy
Which studies and tests actually answer the questions regulators, payers, and clinicians will ask?
- 10
Milestone capital, probabilistic cost and timeline, and risk-adjusted valuation
What will the next milestone cost, how long will it take, and what is the venture plausibly worth?
- 11
Investor, partner, and vendor matching
Who is genuinely able to finance or execute the next milestone, and may they see this project?
- 12
Continuous audit, outcome capture, and model calibration
What has changed, which prior decisions are now invalid, and how well were we calibrated?
Decision states
Including the state most systems refuse to have
When evidence is insufficient, the correct output is ABSTAIN — not a confident guess.
GO
Evidence supports advancement within stated conditions.
CONDITIONAL GO
Advancement is reasonable only if named conditions are satisfied.
HOLD
Pause pending further evidence or required authority.
PIVOT
Change intended use, design, population, pathway, market, or execution plan.
STOP
A current blocker makes further investment unjustified or unsafe.
ABSTAIN
No defensible conclusion can be reached from the available evidence.
Decision-gate timeline
Decision states are recorded, not overwritten. Each transition carries its trigger, the required approval, anything it invalidated, and the version it produced.
- GO
- Evidence supports advancement within stated conditions.
- CONDITIONAL GO
- Advancement is reasonable only if named conditions are satisfied.
- HOLD
- Paused pending further evidence or required authority.
- PIVOT
- Change use, design, population, pathway, or plan.
- STOP
- A current blocker makes further investment unjustified.
- ABSTAIN
- No defensible conclusion can be reached from available evidence.
Currently expanded entry: Decision reopened by an evidence change
Text equivalent
- Day 0 — ABSTAIN: Classification decision withheld. Trigger: Intake complete; jurisdictional evidence below the completeness threshold. Approval: Regulatory reviewer confirmed abstention rather than a provisional answer.
- Day 14 — HOLD: Pathway analysis paused pending predicate evidence. Trigger: Two candidate predicates identified with contested comparability. Approval: Regulatory reviewer set a review date and named the resolving experiment.
- Day 31 — CONDITIONAL GO: Advance on a narrowed indication. Trigger: Predicate comparability evidenced for the narrowed wording only. Approval: Regulatory reviewer approved with two written conditions.
- Day 47 — HOLD: Decision reopened by an evidence change. Trigger: The field performance dataset was superseded; dependent assumptions failed. Approval: Automatic re-evaluation; queued to the original approver with rationale. Invalidated v3 conditions 1 and 2, and the dependent reimbursement decision.
- Day 52 — PIVOT: Change the intended-use population. Trigger: Re-test showed the endpoint is measurable only in the narrower population. Approval: Clinical and regulatory reviewers approved jointly; founder acknowledged.
Governance
Human-governed gates, and a blueprint that keeps living
Software prepares the analysis. People with the relevant authority approve, add conditions, dissent, or override with a written rationale — and every one of those actions is part of the permanent record.
Intelligence
Retrieval, classification, analysis, simulation, matching, and monitoring.
Governance
Provenance, permissions, policy, human approvals, electronic signatures, decision history, and deterministic replay.
Execution
External experts, laboratories, CROs, engineering firms, manufacturers, regulatory representatives, and authorized intermediaries.
Living commercialization blueprint
Approved decisions assemble into an execution-ready plan covering pathway, evidence, design controls, manufacturing readiness, and capital milestones. When a decision changes, the affected sections are re-issued rather than left to age inside a static deck.
Permissioned discovery
Investors and execution partners reach a project only after it qualifies and its owner approves a disclosure package. Access is scoped, recorded, and revocable, and matching follows diligence rather than replacing it.
Why it is different
Structural choices, not features
One ingestion, many decisions
Documents are normalized once and reused across the lifecycle.
Readiness vector, not an opaque score
Domain readiness stays visible instead of collapsing into a single number.
Uncertainty is a product feature
Every forecast carries distributions, assumptions, and sensitivity drivers.
ABSTAIN is enforceable
The system refuses to produce conclusions the evidence cannot support.
Highest-value next experiment
HOLD and PIVOT states identify the experiment most likely to change the decision per dollar and unit time.
Continuous diligence
New evidence triggers re-evaluation instead of a new static report.
Ninety-day pilot
Test the decision value before anyone scales the promise
A scoped, paid pilot runs historical replay, a blinded expert challenge, and live shadow mode against your own portfolio, measured against a baseline you define. Every KPI is a measurement target, not an achieved result.
Bring one project. See where it would abstain.
A scoped demonstration shows the evidence trail, the gate logic, and the exact conditions that unblock the next decision.