[src: 02-Runcible Investor Intro Memo (Online)]
Runcible
From AI Hypothesis to Institutional Action
Confidential Investor Summary
Contact: Curt Doolittle, Founder & CEO — curt.doolittle@runcible.com
www.runcible.com
The Problem: AI Generates Outputs Institutions Cannot Yet Act On
Foundation models can produce fluent answers, summaries, analyses, recommendations, and proposed actions.
But institutions cannot act merely because an answer is fluent.
The first answer is not the truth.
It is the pleading.
AI output is not authority. It is candidate material: a claim, hypothesis, recommendation, or proposed action that must be tested before an institution can rely on it.
In law, healthcare, insurance, finance, government, defense, procurement, compliance, and enterprise operations, an output must be testable, reviewable, authorized, auditable, and liability-bounded before it can become institutional action.
The AI industry has made generation abundant.
It has not solved institutional qualification.
Every major productive technology required a control discipline before it could govern high-liability work:
- Science required engineering.
- Industry required accounting.
- Software required compilers, permissions, tests, logs, and audit trails.
- AI now requires adjudication, qualification, and records.
Without that layer, AI remains trapped in low-liability assistance: useful for drafting, summarizing, and advising, but difficult to deploy where error creates legal, financial, medical, operational, or public consequences.
The missing market is not AI assistance.
The missing market is governed institutional action.
The Solution: Runcible Converts AI Output Into Decidability Records
Foundation models generate hypotheses.
Runcible adjudicates them.
Institutions act on Decidability Records.
Runcible supplies the missing adjudication and qualification layer between foundation-model generation and institutional execution.
Technically, Runcible is a semantic compiler and qualification runtime for institutional AI. It translates AI-generated language into operational claims, then tests those claims against universal admissibility conditions: testifiability, reciprocity, possibility, authority, bounded liability, and decidability.
Only after that does it apply the institution’s local law, policy, contract, workflow, evidence standard, and escalation rules.
Runcible gives AI a qualified work identity: role, scope, permissions, evidence boundaries, authority limits, supervision requirements, escalation rules, audit duties, review conditions, and liability boundaries.
Each governed workflow produces a Decidability Record: a structured artifact showing what was claimed, what evidence was used, what rules applied, what authority governed the work, what tests passed or failed, what remains unresolved, what action state exists, and what must happen next.
Runcible does not replace foundation models.
It qualifies their outputs.
The model proposes.
Runcible tests.
The Decidability Record preserves what the institution can review, defend, repair, escalate, certify, reject, or declare undecidable.
How It Works
Runcible moves AI-mediated work through a qualification process.
1. Intake the Matter
Runcible begins with a request, document, claim, recommendation, case file, contract, regulation, policy, workflow, or proposed action.
2. Define Role and Scope
Runcible defines what the AI may examine, infer, recommend, certify, reject, or escalate.
The AI is not treated as an unbounded assistant. It is given a qualified work identity.
3. Translate Language Into Operational Prose
Runcible converts institutional language into operational structures:
- Actors
- Actions
- Objects
- Claims
- Evidence
- Rules
- Authorities
- Dependencies
- Obligations
- Permissions
- Prohibitions
- Exceptions
- Liability boundaries
4. Generate or Receive Hypotheses
Foundation models supply candidate summaries, classifications, interpretations, recommendations, or actions.
These are not treated as truth.
They are treated as hypotheses.
5. Apply RDL, Ontology, and Protocols
Runcible applies domain-specific terms, claim types, evidence requirements, authority boundaries, rule constraints, workflow obligations, and institutional protocols.
6. Adjudicate
Runcible tests identity, consistency, evidence, possibility, reciprocity, legality, authority, liability, warrantability, and decidability.
7. Emit Diagnostics
Runcible identifies missing evidence, ambiguity, contradiction, unsupported authority, policy conflict, impossible action, escalation requirement, unresolved liability, or undecidable claims.
8. Assign Action State
Runcible assigns an action state:
- Certified
- Failed
- Repairable
- Escalated
- Rejected
- Undecidable
9. Produce the Decidability Record
Runcible preserves the basis for review, audit, certification, repair, escalation, future reuse, and institutional memory.
This converts AI from a probabilistic suggestion engine into a governed participant in institutional workflows.
The Impact: From Risk to Revenue
The highest-value AI markets are blocked by liability.
Institutions want the productivity of AI, but they cannot delegate high-liability work to systems that produce unbounded, unaudited, non-warrantable outputs.
Runcible unlocks these markets by making AI work:
Testable
Claims are tied to evidence, rules, and explicit tests.
Reviewable
Failures, uncertainties, and unresolved dependencies are visible.
Auditable
Each workflow produces a durable record.
Certifiable
Outputs can be assigned protocol-based certification states.
Liability-bounded
Authority, scope, unresolved risk, and escalation requirements are recorded.
Actionable
Institutions know what they may do next.
Initial high-liability wedges include:
- Insurance claims and underwriting
- Healthcare administration and prior authorization
- Financial compliance and risk review
- Legal and contract review
- Government determinations
- Defense procurement and staff work
- Enterprise policy and regulatory workflows
Runcible makes AI profitable where ordinary AI is most constrained: institutional work where error carries real cost.
Business Model: Governing Liability-Bearing AI Work
Runcible monetizes adjudicated institutional work across five primary layers.
1. Workflow Subscriptions
Governed AI workflows for specific institutional roles: claim reviewer, authorization reviewer, compliance analyst, contract reviewer, procurement reviewer, audit preparer, policy adjudicator, or staff-work assistant.
2. Protocol Libraries and Vertical Packages
Reusable domain protocols for recurring claim, evidence, authority, and liability structures in regulated industries.
3. Runtime and API Licensing
Enterprise access to the Runcible runtime for adjudication, diagnostics, action-state assignment, and Decidability Record generation.
4. OEM and Platform Licensing
Integration into foundation-model companies, cloud providers, enterprise AI platforms, and systems integrators seeking access to liability-bearing markets.
5. Certification, Audit, and Warrantability Services
Protocol-based certification states, audit support, Decidability Record review, and eventual insurance or warranty partnerships.
Additional upside includes Oversing platform licensing, customer-specific integrations, training, protocol marketplaces, and Decidability Record corpus development.
The business does not depend on owning the largest model.
It depends on owning the adjudication and qualification layer that models require before they can enter institutional action.
The Moat
Runcible is not a prompt wrapper, chatbot, compliance checklist, governance dashboard, or ordinary guardrail.
Its moat consists of:
- A decades-long research program in operational truth, reciprocity, possibility, law, and decidability
- RDL, a domain language for converting institutional prose into testable operational structures
- An ontology of claims, roles, evidence, authority, liability, and institutional action
- Protocol libraries for adjudicating domain-specific work
- The Runcible runtime for applying tests, emitting diagnostics, and assigning action states
- Decidability Records as the audit artifact of governed AI work
- A growing corpus of certified claims, diagnostics, and adjudication traces
- Training material generated from certified work rather than unqualified model output
The durable asset is not a model.
The durable asset is the adjudication system and the certified institutional memory it produces.
Investment Case: The Required Layer Before Institutional AI
Foundation-model companies have made AI powerful.
They have not made AI institutionally accountable.
The next market is not simply better assistants.
The next market is governed institutional action: AI that can participate in regulated, high-liability, audit-sensitive workflows.
Runcible is positioned as the adjudication, qualification, and proof layer for that market.
The strategic logic is simple:
- Enterprises need AI outputs they can review, certify, and defend.
- Governments and regulated industries need auditability, authority, and liability boundaries.
- Insurers and auditors need records, not chat transcripts.
- Foundation-model companies need a path into high-liability workflows.
- Runcible supplies the missing adjudication layer.
If foundation models are the engines of AI, Runcible is the qualification and control system that lets those engines enter the institution.
Closing
AI has entered the workplace.
Now it must enter the institution.
That requires more than generation, retrieval, agents, tools, and alignment filters.
It requires an adjudication and proof layer that can determine whether AI output can become institutional action.
Runcible converts candidate language into Decidability Records.
That is how AI becomes testable, reviewable, certifiable, warrantable, liability-bounded, and profitable in the markets where accountability matters most.
Foundation models generate.
Runcible adjudicates.
Institutions act.
Runcible turns AI outputs into testable, reviewable, certifiable, and actionable Decidability Records inside liability-bearing workflows.
Runcible unlocks institutional AI.
Return to Investor Portal • Executive Summary • Short Intro Deck • Read Explanations