Mayo Clinic Waypoints Program application.

Executive summary

The technology. A Neurochemical Cascade Architecture (NCA) for AI language models, with biology-grounded conductance gating (34 functional heads across four tiers), pharmacokinetic decay modeling, and a hash-verifiable specialist composition layer. Four pending USPTO applications (63/939,190; 63/962,385; 63/988,485; 64/034,536). Implemented in Scyla, a proprietary biology-native compiler wholly owned by Nexus Concordat.

Operational state. A live specialist cell (cardiac_gen3) serves inference behind a signed manifest verifiable by SHA-256 against a public registry at /cell-demo.html. A second specialist substrate (psych_gen1) is in active training on license-clean public corpus with per-source provenance manifests. The architecture is patent-pending and the runtime is in production.

The application we propose for Mayo. A training-simulation platform for psychiatry residency education. Each simulated patient is a software character with a persistent neurochemical state, populated from canonical diagnostic templates. Behavior under interview, perturbation (missed dose, sleep disruption, stressor), and simulated time is reproducible, replayable, and grounded in the architecture's biological substrate. Residents train against the simulator. Faculty evaluate resident judgment against known ground truth. Signed inference manifests make every training session auditable in a way that human standardized-patient actors structurally cannot match.

The framing. Training-simulation infrastructure. Non-device regulatory category. No real patient data interpreted. Future SaMD pathways identified as separate downstream tracks rather than the current ask.

The team. Founder + COO + JD/RAC-credentialed regulatory advisor on file. Active corpus discipline (SHA-256-stamped per source). Operational prototype. Existing TTRPG product (Chronicles of Apocalyptica) as the deployment substrate the simulator is built on top of.

What we would value from Waypoints. Structured introductions to Mayo Clinic faculty in psychiatry and medical education. Feedback from Mayo's innovation and regulatory staff on the proposed non-device positioning. The methodological grounding the Berg Innovation Exchange is set up to provide.

Application answers, Page 1 of 4 · PLACEHOLDER DRAFTS

Drafts only. Marjorie will fine-tune in the actual form, route through Jason and Robert, and submit personally.

1 · Job title / role

Founder & Principal AI Technologist, Nexus Concordat Inc.

2 · Leadership team

• Marjorie McCubbins · Founder & Principal AI Technologist. Inventor of record on all four pending patent applications. LinkedIn: [Marjorie to insert URL]
• Aislinn McCubbins · COO, Dynamic Hallucination, Inc. (the operating company; Nexus Concordat holds the patents). Co-creator of canon architecture and lore systems.
• Robert Michalik, JD, RAC · Regulatory advisor (pro bono, NDA signed January 2026). Twenty-plus years biopharma regulatory experience with specialization in AI-SaMD. Juris Doctor, Suffolk University Law School. Regulatory Affairs Certified. Found the work independently through LinkedIn and offered support before we met in person. Administrator of the LinkedIn group "Generative AI (AI ML) for CMC, QA & Regulatory Professionals in Pharmaceutical, Biologic & Med Tech."
• Jason Alba · Board of Directors, Dynamic Hallucination, Inc. and angel investor. Governance role, not operations.

3 · Company website

patent.nexusconcordat.com (architectural and IP documentation)
chroniclesofapocalyptica.com (the product platform)

4 · Pitch deck / executive summary

The Executive Summary section at the top of this page is the canonical 1-page summary for export. Export as PDF and upload.

5-10 · Company address

[Marjorie to confirm: physical operating address, or use Delaware C-Corp registered agent address]

Delaware registered agent (Harvard Business Services, on file as registered agent for Nexus Concordat Inc. since incorporation):

11 · Healthcare sector(s)

• ✅ Digital Health (primary · software training infrastructure)
• ✅ Healthcare Services (medical education service delivery)
• ☐ Medical Devices (intentionally avoided · non-device positioning)
• ☐ Diagnostics (intentionally avoided)
• ☐ Therapeutics (intentionally avoided)
• ☐ Other

12 · Problems addressed

Psychiatry residency training depends on standardized-patient actors to simulate the clinical presentations residents must learn to assess. The current approach has three structural limits.

First, cost. Major academic medical centers spend $100,000+ annually on standardized-patient actor programs. The cost scales linearly with cohort size and training hour.

Second, consistency. Two trainees who interview the same standardized patient get different presentations because actors are human. Faculty cannot evaluate resident judgment against a fixed ground truth.

Third, replay. A simulated patient cannot be reset to a specific neurochemical baseline on demand. A resident cannot meaningfully practice the same case three times to internalize a decision pattern. Longitudinal scenarios across multiple sessions cannot be constructed deterministically.

We address all three by replacing human standardized patients with software characters whose internal state is biochemically grounded, deterministically replayable, and auditable by signed inference manifest. Mayo's faculty curates the canonical patient panels and the encounter library. Our engine runs the simulator on top of that curation.

13 · Roles typically using the solution

• Psychiatry residents (primary trainees, PGY-1 through PGY-4)
• Psychiatry attendings and faculty (running structured training scenarios, evaluating resident performance)
• Medical education directors and curriculum coordinators (assembling scenario sets across residency cohorts)
• Standardized-patient program coordinators (integrating the simulator alongside existing standardized-patient infrastructure)

14 · Roles typically purchasing the solution

• Designated Institutional Officials (DIOs) overseeing Graduate Medical Education
• Department of Psychiatry residency program directors
• Vice Deans for Medical Education
• Chief Medical Education Officers at academic medical centers
• Hospital training budget owners with discretion over GME and CME spend

15 · Competitors

The closest functional competitors are standardized-patient actor programs (in-house departments and external companies like SimRecruit and SP Educator Network). These are institutional infrastructure we replace or augment, depending on the deployment model, rather than technology competitors.

In the AI-narrative space, AI Dungeon-style tools and LLM-wrapper roleplay products exist but do not address clinical-grade requirements: no biochemical substrate, no replayability across sessions, no signed-manifest auditability, no patent-defensible architectural claim, no regulatory advisory infrastructure.

In the simulation-training-software space, Simwerx, ScenarioMD, and similar provide medical-procedure training but not psychiatric residency training with persistent neurochemical state. Differentiation is the biochemical substrate and the architectural verifiability, both protected by pending patents.

No direct competitor offers a persistent character-state engine where chemistry drives behavior reproducibly across sessions with signed-manifest inference provenance.

Application answers, Page 2 of 4 · PLACEHOLDER DRAFTS

Drafts only. Marjorie will fine-tune in the actual form, route through Jason and Robert, and submit personally.

16 · Describe your innovation. How is it differentiated from competitors and current best practices?

The innovation is a TTRPG-based clinical training simulator powered by a patent-pending neurochemical AI architecture. A simulated patient is a software character whose 10-dimensional neurochemical state (cortisol, dopamine, serotonin, norepinephrine, oxytocin, GABA, acetylcholine, endorphins, glutamate, melatonin) sits on the character record. Behavior under interview, under perturbation (missed dose, sleep loss, environmental trigger, medication change), and across simulated time is generated through a trained specialist cell that has learned the relationships between chemistry, pharmacokinetics, and behavior, then rendered as clinical narrative through a language layer.

Differentiation against current best practices, three categories:

vs. standardized-patient actor programs (the institutional default in psychiatry residency training): actors cost academic medical centers $100,000+ annually, present inconsistently across sessions, and cannot be reset to a specific neurochemical baseline on demand. Our simulator costs effectively $0 marginal per session after deployment, is byte-exactly replayable across resident cohorts, and can be reset deterministically to any canonical state for longitudinal scenario design.

vs. AI roleplay and LLM-wrapper tools (AI Dungeon-class products and generic large-language-model chatbots): we are not improvising from a general-purpose model. Every emission is bound to a signed inference manifest. Every cell has a hash-chained lineage in a public Merkle DAG registry. Each architectural primitive is registered as a hashable opcode. Inference is byte-exact reproducible. Generic LLM tools structurally cannot match any of these properties.

vs. medical simulation software (Simwerx, ScenarioMD, and similar): these target procedural training (intubations, surgical steps, physical exams). None target psychiatric residency with persistent biochemical state. Our differentiation is the substrate, not the interface.

The architectural moat sits in four pending USPTO applications and a proprietary biology-native compiler (Scyla) owned in full by Nexus Concordat. A competitor would need to both infringe the pending claims and rebuild the compiler to reproduce the function.

17 · Stage of maturity

Prototype.

The architecture is live and serving inference behind a signed manifest. A specialist cell (cardiac_gen3) is queryable at /cell-demo.html with verifiable byte-exact provenance against a public registry. A second specialist substrate (psych_gen1) is in active training as of June 2026. The TTRPG product (Chronicles of Apocalyptica) is in production. The Mayo collaboration through Waypoints would move us from prototype to Mayo-curated training-platform deployment.

18 · Proof points

Each item below is independently verifiable at the URL given, or via the public artifact named.

• Live cardiac specialist cell with signed inference manifest emission. Type a clinical phrase, receive an autoregressive output bound to a SHA-256-verifiable manifest. patent.nexusconcordat.com/cell-demo.html
• Public cell-lineage registry for byte-exact external verification. patent.nexusconcordat.com/registry/cardiac-registry.json
• Architectural plan documenting the cell-as-opcode-tree unification. patent.nexusconcordat.com/jason-plan.html
• Training plan + trajectory document describing how the trained cells become the substrate the TTRPG runs on. patent.nexusconcordat.com/sister-cell.html
• Company overview hub linking every operational surface. mayo.nexusconcordat.com
• Four pending USPTO provisional applications covering the architecture, the memory weighting, the cell-as-opcode-tree unification, and hash-verifiable specialist composition: 63/939,190, 63/962,385, 63/988,485, 64/034,536.
• Operating company in production built on the architecture. dynamichallucination.com
• TTRPG platform live on the substrate the simulator would deploy. chroniclesofapocalyptica.com
• Vanilla baseline serving alongside the chemistry-conditioned cell for apples-to-apples emission comparison under matched seed, temperature, and top-k parameters. Visible on the cell-demo page as the right-hand column.
• Robert Michalik, JD, RAC. Regulatory advisor, 25 years biopharma regulatory experience, mutual NDA signed January 2026. Independent peer validation: he found the work through LinkedIn and offered support before we met in person.
• Corpus discipline. Per-source SHA-256 manifests with verbatim query terms, license attribution, retrieval timestamps. Raw and cleaned data separation enforced at directory level. Documented in CORPUS_PURPOSE.md at the corpus root.

Application answers, Page 3 of 4 · PLACEHOLDER DRAFTS

Drafts only. Marjorie will fine-tune in the actual form, route through Jason and Robert, and submit personally.

19 · Most recent milestones achieved

• June 22, 2026. Invited by Mayo Clinic Berg Innovation Exchange to apply for the September 2026 Waypoints Program cohort. Personalized application link issued.
• June 22, 2026. psych_gen1 corpus pulled (PubMed targeted, OpenFDA labels + FAERS, ChEMBL, ClinicalTrials.gov, NIMH RDoC) with full per-source SHA-256 manifests. Tokenized to 51.5 million tokens against a 76,325-row vocabulary. Vanilla baseline now in active training on local A5000. 4th provisional patent application (64/034,536, hash-verifiable specialist composition) filed.
• June 21, 2026. cardiac specialist cell + Qwen 2.5 14B language bridge shipped as a production service. 6-to-8-second end-to-end latency per inference, signed manifest emission, byte-exact verification against a public registry. Per-character clinical pipeline integrated.
• June 1, 2026. Public cell-demo page goes live with signed inference manifests and SHA-256 verification recipe published.
• May 31, 2026. cardiac_gen3 specialist cell completes end-to-end training through the 34-head Hodgkin-Huxley cascade on 20 million tokens of curated cardiology corpus. Vanilla baseline trained on identical data for apples-to-apples comparison.
• April 30, 2026. Nexus Concordat Inc. incorporated as Delaware C-Corporation (File 10605960) via Harvard Business Services.
• April 2026. AetherBrain Cascade checkpoint trained (32.6 million parameters, 34-head cascade, Vast.ai H200). Clean reload verified, zero missing keys, zero unexpected keys.
• January 2026. Robert Michalik, JD, RAC, joined as regulatory advisor on a pro bono basis. He found the work independently through LinkedIn and offered support before we met in person. Mutual NDA signed.

20 · Next milestones to achieve

• September 20, 2026. Submit completed Waypoints Program application.
• By Q3 2026. Complete psych_gen1 4-brain Scyla cascade training run (cardiac mirror, four learning rules plus vanilla baseline). First checkpoint per learning rule, signed-manifest verifiability against the public registry.
• Q4 2026. If selected for Waypoints, execute the 12-week structured engagement (October 5 through end of December 2026). Targeted introductions to Mayo Clinic psychiatry and medical education faculty. Joint scoping of the simulated-patient training-platform pilot.
• Within the 12-month provisional window for each pending application. Convert the four pending provisionals (63/939,190; 63/962,385; 63/988,485; 64/034,536) to non-provisional applications.
• By end of Q4 2026. Third specialist cell trained (domain to be chosen based on what closes the cleanest generalization argument). Endocrine, neurology, or GI are the leading candidates.
• 2027 (downstream). Mayo-curated simulated-patient training pilot. Joint validation publication.

21 · How Mayo Clinic can help

• ✅ Clinical Insights · Mayo psychiatry faculty curating canonical patient state panels and diagnostic archetypes is the highest-leverage Mayo contribution.
• ✅ Co-development · Joint validation paper, Mayo-branded simulator instance, methodology aligned with Mayo training standards.
• ☐ Clinical Trials · Not applicable in the Waypoints engagement. We are explicitly non-device, non-CDS, and do not interpret real patient data.
• ✅ Research · The simulator itself becomes a reusable research substrate for resident training research and longitudinal scenario reproducibility studies.
• ✅ Product Testing · Mayo residents training against the simulator IS the product testing, in the deployment we are scoped for.
• ✅ Data · Mayo's expert-validated state taxonomy and anonymized residency simulation scenario library are the data inputs the engine needs to move from public-corpus baseline to clinical-grade training.
• ☐ Other

22 · Current engagement with Mayo Clinic

No formal engagement currently. The communication channel established to date is:

• David Kim, Business Analyst, Mayo Clinic Berg Innovation Exchange. Initial outreach: March 22, 2026 (inquiry about Waypoints Program submission). Response: June 22, 2026 (formal invitation to apply for the September 2026 cohort with personalized application link).

23 · Most recent funding classification

Bootstrapped.

Nexus Concordat Inc., the patent-holding entity submitting this application, is self-funded. The operating company built on the Nexus Concordat patent portfolio (Dynamic Hallucination, Inc.) has accepted angel investment from Jason Alba, who serves on its Board of Directors. The Nexus Concordat IP holding entity itself has not raised institutional capital.

Application answers, Page 4 of 4 · PLACEHOLDER DRAFTS

Drafts only. Marjorie will fine-tune in the actual form, route through Jason and Robert, and submit personally.

24 · Do you have funding for your Research study?

No.

Nexus Concordat Inc. has no research-study budget earmarked at this time. The Waypoints engagement is meant in part to help scope what a research project would look like and what its funding requirements would be. Any subsequent research collaboration would require its own funding round, which we expect to be a separate conversation downstream of Waypoints.

25 · Mayo Clinic experts for Product Testing

We are not yet certain which specific Mayo Clinic experts we would engage. The characteristics and experience we are looking for:

• Psychiatry residency program director or vice chair for education (the deployment owner for a residency simulation pilot)
• Mayo Clinic faculty with active experience designing and running standardized-patient programs in psychiatry (the people who know what the current state of practice is and where it falls short)
• Medical simulation specialists with experience integrating software-based simulation alongside traditional standardized-patient infrastructure
• Mental health curriculum design SMEs (for canonical scenario authoring)
• Mayo Clinic Berg Innovation Exchange innovation and technology-transfer staff (for engagement structure, methodological grounding, and pathway navigation)
• Regulatory affairs SMEs at Mayo familiar with non-device positioning of medical-education infrastructure (complementary to our advisor, Robert Michalik, JD, RAC)

Rationale: the proposed engagement is a psychiatry residency training-simulation pilot. The expert mix we describe maps onto the people who would own the pilot from the Mayo side (program director, curriculum design), validate the methodology (psychiatry faculty), integrate it operationally (medical simulation specialists), and pressure-test the regulatory framing (innovation and regulatory staff).

26 · Do you have funding for Product Testing?

No.

The Waypoints engagement model is benefactor-funded at no cost to participating innovators. The product testing we are proposing (residents training against the simulator) sits inside that engagement window. Beyond Waypoints, a Mayo-deployed pilot would require funding scoping that we expect to address through subsequent conversations with Mayo's innovation staff or jointly identified partners.

27 · Data needs specifications and quantities

The data we would seek from Mayo is curated educational content. We are not requesting patient records, diagnostic case histories, or any identifiable patient data.

• Approximately 12 canonical neurochemical and behavioral state archetypes covering the high-frequency psychiatric diagnostic clusters: major depressive disorder (severe), major depressive disorder (moderate), generalized anxiety disorder, panic disorder, bipolar I depressive episode, bipolar I manic episode, bipolar II, PTSD, complex PTSD, schizophrenia spectrum, borderline personality disorder, substance use disorder. For each archetype: Mayo's expert-validated definition of the canonical neurochemical signature and behavioral presentation pattern.
• Approximately 30 to 50 anonymized residency simulation scenarios drawn from Mayo's existing standardized-patient training library. For each scenario: presenting complaint, encounter shape, decision tree, expected resident response.
• Mayo's expert-validated state taxonomy. Documentation describing how Mayo defines and distinguishes adjacent diagnostic states (e.g., the criteria Mayo uses to distinguish complex PTSD from MDD-comorbid-PTSD).

Format: structured documentation, training-scenario templates, and reference taxonomies. No protected health information, no identifiable patient data.

28 · Do you have funding to purchase or license data?

No.

We are not proposing a data purchase or licensing transaction. The data described in question 27 is curated educational content we would seek through the Waypoints co-development structure. If Mayo's policies require a separate licensing or data-use agreement for curated training content, we are open to scoping that.

29 · Other areas of support needed

• ✅ Legal/Regulatory Guidance · For the conversion of pending provisional patents to non-provisional applications, and for Mayo's regulatory perspective on the proposed non-device positioning.
• ☐ New Market Entry
• ☐ Manufacturing/Development
• ✅ Partner · Mayo as the validation partner for the simulated-patient training pilot.
• ☐ Office Space
• ☐ Specialized Space (Lab or Otherwise)
• ☐ Specialized Equipment
• ☐ Investment · Not the Waypoints ask. May be relevant in a separate downstream conversation.
• ✅ Other · Academic credibility and publication path. Joint validation publication co-authored with Mayo psychiatry faculty would convert our prototype into infrastructure that other academic medical centers can adopt.

30 · Current engagement with Mayo Clinic

No formal engagement currently. The communication channel established to date is:

• David Kim, Business Analyst, Mayo Clinic Berg Innovation Exchange. Initial outreach: March 22, 2026 (inquiry about Waypoints Program submission). Response: June 22, 2026 (formal invitation to apply for the September 2026 cohort with personalized application link).

31 · Most recent funding classification

Bootstrapped.

Nexus Concordat Inc., the patent-holding entity submitting this application, is self-funded. The operating company built on the Nexus Concordat patent portfolio (Dynamic Hallucination, Inc.) has accepted angel investment from Jason Alba, who serves on its Board of Directors. The Nexus Concordat IP holding entity itself has not raised institutional capital.

Application drafts complete

Placeholder drafts for all four pages of the Waypoints application are above. Walk into the actual application form with these as starting clay, edit personally, route the finished draft through Jason and Robert, then submit. Application deadline: September 20, 2026. Selection decisions: September 28, 2026. Program begins: October 5, 2026.

Supporting links for the Waypoints reviewers

• /cell-demo.html · live cardiac_gen3 inference, signed manifests, public verification recipe
• /jason-plan.html · architectural plan (cell as opcode tree)
• /sister-cell.html · training plan, the TTRPG, and the trajectory
• /opcodes.html · the registered opcode table
• /registry/cardiac-registry.json · public cell-lineage registry

What Marjorie needs to provide before submission

• Marjorie's LinkedIn URL (question 2)
• Confirmation of operating address vs. registered agent address (questions 5 through 10)
• Executive summary exported as PDF (question 4 file upload, 10 MB max)