| Quarter | RFS title | HSDLG contribution | IP-safe HSDLG tools to be applied | Impact |
|---|---|---|---|---|
| Winter 2027 | Startups in a World of Abundant Space Access | Compare mission/business alternatives and expose assumptions behind viability. | TUP; BP/BPL; Model/Theory Formation | Supporting: better feasibility decisions; aerospace models and mission evidence required. |
| Winter 2027 | The $15-an-Hour Expert | Turn expert reasoning into reviewable workflows, with escalation for uncertain judgments. | KAI; Knowledge Governance; BP/BPL; ValidityBench | Strong fit: more accessible expertise; task-specific competence must be demonstrated. |
| Winter 2027 | Robots Made in America | Review component provenance, supplier claims and maintenance evidence against requirements. | KAI; BP/BPL; Continuity | Supporting: clearer supply-chain decisions; manufacturing capacity remains external. |
| Winter 2027 | LLMs That Write Well | Maintain audience/style constraints and review writing against human preferences and factual evidence. | Context and Constraint Builder; BP/BPL; ValidityBench | Supporting: more consistent editorial quality; human preference evaluation required. |
| Winter 2027 | Plaid for Government Benefits | Connect authorized source records to eligibility claims, exceptions and reviewer decisions. | KAI; AAAM; Knowledge Governance; BP/BPL | Strong fit: reviewable eligibility evidence; agency integration, consent and legal interpretation required. |
| Winter 2027 | Wall Street for the Physical Buildout | Challenge utilization, cash-flow and asset assumptions across competing financing cases. | Model/Theory Formation; BP/BPL; Continuity | Supporting: stronger diligence; no underwriting or investment-performance claim. |
| Winter 2027 | Vertical Foundation Models | Structure domain hypotheses, training/evaluation evidence and meaningful falsification tests. | TAIE/EMSM; Model/Theory Formation; ValidityBench | Strong research fit: better evaluation foundations; domain transfer and model advantage need proof. |
| Winter 2027 | AI for City Hall | Trace permit or service recommendations to current evidence, rules and responsible officials. | KAI; TUP; BP/BPL; Continuity | Strong fit: clearer review and exception handling; local-rule and workflow qualification required. |
| Winter 2027 | Data Center Supply Chain | Review supplier evidence, dependencies and feasible alternatives against power/cooling constraints. | KAI; Model/Theory Formation; BP/BPL | Supporting: earlier identification of unsupported assumptions; real supplier and site data required. |
| Winter 2027 | AI for the Air-Gapped Economy | Provide evidence/review workflows alongside customers' chosen local intelligence. | TUP; BP/BPL; Knowledge Governance; Continuity | Strong architectural fit: governed local work; disconnected deployment and security qualification remain open. |
| Winter 2027 | Own Your Intelligence | Retain reviewed expertise and lessons across tasks and changing model choices under BYOI. | KAI; Knowledge Governance; Continuity; ValidityBench | Strong platform fit: reusable organizational knowledge; transfer benefit and interfaces need validation. |
| Fall 2026 | The Primer | Preserve learning context, identify misconceptions and review explanations against evidence. | KAI; Continuity; BP/BPL; ValidityBench | Potential learning continuity; teacher-led evaluation and child-appropriate design required. |
| Fall 2026 | The Future of American Defense | Support maintenance, logistics and acquisition evidence reviews with accountable decisions. | TUP; BP/BPL; Standing Grants; Continuity | Supporting: disciplined readiness decisions; domain/safety approval and qualified integration required. |
| Fall 2026 | A Cloud for Small Software | Carry requirements, behavioral checks and change histories with small applications. | Context and Constraint Builder; BP/BPL; Continuity | Potentially more maintainable small software; hosting and deployment integration required. |
| Fall 2026 | Multiplayer AI | Preserve shared evidence, outstanding questions and decisions across human/AI handoffs. | Continuity; Knowledge Governance; BP/BPL | Strong fit: less context loss and clearer responsibility; multi-user permissions/conflict handling required. |
| Fall 2026 | Compute at Sea | Compare credible architectures and test cooling-led economics against delivered useful work. | TUP; Model/Theory Formation; BP/BPL; bounded mathematical checks | Studied: exposed scenario deficits and constraint interactions; actual site/SME evidence remains essential. |
| Fall 2026 | AI-Powered Consumer Products for 1 Billion People | Supply reusable evidence and continuity functions to a selected consumer workflow. | Context and Constraint Builder; Continuity; BP/BPL | Indirect: application quality support; consumer demand, distribution and unit costs remain separate. |
| Fall 2026 | AI for the Aging Population | Reconcile care observations and retain caregiver decisions with escalation on uncertainty. | KAI; AAAM; Continuity; BP/BPL | Potential coordination benefit; care expertise, consent, accessibility and field evidence required. |
| Fall 2026 | New Operating Systems for the Physical World | Connect asset evidence, resource-feasible plans and bounded authority to outcome records. | TUP; Standing Grants; BP/BPL; Continuity | Studied: explicit constraints and control checks; no incremental final scheduling gain beyond optimizer. |
| Fall 2026 | The Best Time to Build in Crypto | Structure operational reconciliation and review evidence for disputed or uncertain transactions. | AAAM; BP/BPL; Continuity; bounded mathematical checks | Supporting: clearer exception/reconciliation work; financial-domain and legal qualification required. |
| Fall 2026 | Data for the Real World | Define observation quality and isolate sensing-policy effects from model/workflow changes. | Model/Theory Formation; BP/BPL; ValidityBench | Studied prospectively: stronger experiment design; empirical sensing benefit remains untested. |
| Fall 2026 | Proving You're Human | Bind authenticity claims to specific signals, threat models and permitted decisions. | AAAM; BP/BPL; ValidityBench | Studied prospectively: scoped trust claims and falsifiable evaluation; no deployed identity product. |
| Fall 2026 | AI-Native Compliance Infrastructure | Maintain the evidence supporting vendor reliance and identify changes requiring review. | KAI; AAAM; BP/BPL; Knowledge Governance | Studied: concrete scope analysis and local mechanics; customer benefit requires a paired pilot. |
| Fall 2026 | Self-Maintaining APIs | Support scoped repairs, behavioral verification and accountable handling of uncertain effects. | TUP; BP/BPL; Standing Grants; Continuity | Studied: repaired application passed 12/12 selected checks; general repair effectiveness unestablished. |
| Summer 2026 | AI for Low-Pesticide Agriculture | Compare intervention hypotheses and link field observations to treatment decisions. | Model/Theory Formation; KAI; BP/BPL; ValidityBench | Supporting: better experiment and evidence design; agronomic outcomes need controlled field trials. |
| Summer 2026 | AI-Native Discovery Engines | Structure hypothesis/experiment/review cycles and preserve failed hypotheses and revised theories. | Model/Theory Formation; TAIE/EMSM; BP/BPL; Knowledge Governance | Strong fit: disciplined discovery loops; laboratory integration and empirical novelty remain to prove. |
| Summer 2026 | AI-Native Service Companies | Deliver reviewed work products with evidence, open decisions and repeatable customer acceptance. | TUP; KAI; BP/BPL; Continuity | Strong commercial fit: an avenue for founder-led delivery now; quality and margins require measurement. |
| Summer 2026 | AI Personalized Medicine | Organize patient-specific evidence and competing explanations for qualified clinical review. | KAI; AAAM; BP/BPL; ValidityBench | Supporting only: traceable clinical reasoning; no clinical efficacy or treatment authorization implied. |
| Summer 2026 | Company Brain | Retain reviewed organizational knowledge with provenance, changes and unresolved contradictions. | KAI; Knowledge Governance; AAAM; Continuity | Strong platform fit: reusable institutional context; access control and later-task benefit need testing. |
| Summer 2026 | Counter-Swarm Defense | Support requirements, sensor-evidence evaluation and non-operational readiness review. | Model/Theory Formation; BP/BPL; ValidityBench | Indirect: evaluation and assurance support; no targeting, interception or weapons-control capability claimed. |
| Summer 2026 | Dynamic Software Interfaces | Bind generated interfaces to user intent and test that changes preserve required behavior. | Context and Constraint Builder; BP/BPL; ValidityBench | Potentially clearer, more dependable adaptation; usability and runtime integration require testing. |
| Summer 2026 | Electronics in Space | Review design evidence against thermal, power, radiation and mission requirements. | TUP; Model/Theory Formation; BP/BPL | Supporting: more explicit engineering tradeoffs; hardware design/qualification remains external. |
| Summer 2026 | Hardware Supply Chain | Track component evidence, substitutions and dependencies behind procurement decisions. | KAI; AAAM; BP/BPL; Continuity | Supporting: traceable sourcing decisions; verified commercial and technical data required. |
| Summer 2026 | Industrial Capabilities in Space | Compare process and mission hypotheses and define the evidence needed before capital commitment. | Model/Theory Formation; TUP; BP/BPL | Indirect: stronger feasibility screening; physical processes, launch and operations remain external. |
| Summer 2026 | Inference Chips for Agent Workflows | Define representative workloads and assess throughput, energy and total-system tradeoffs. | TUP; Model/Theory Formation; ValidityBench | Supporting: decision-relevant benchmark design; no chip design or silicon advantage established. |
| Summer 2026 | SaaS Challengers | Build customer workflows around retained evidence and reviewed outcomes using chosen intelligence. | TUP; BP/BPL; Knowledge Governance; Continuity | Potential product foundation; competitive differentiation and switching value need customer proof. |
| Summer 2026 | Software for Agents | Supply bounded work planning, review, authority and continuity functions for agent applications. | TUP; Standing Grants; BP/BPL; Continuity | Strong fit: reusable agent-work infrastructure; full MCP/client integration remains in progress. |
| Summer 2026 | Startups That Want to Sell to Huge Companies | Prepare evidence-backed deliverables and acceptance records for a tightly scoped enterprise problem. | Context and Constraint Builder; KAI; BP/BPL | Strong delivery fit: clearer enterprise acceptance; sales cycles, procurement and willingness to pay remain external. |
| Summer 2026 | Supply Chain 2.0 for Semiconductors | Connect supplier/production evidence to dependency changes and operational decisions. | KAI; AAAM; Continuity; BP/BPL | Supporting: clearer disruption analysis; trustworthy supply-chain feeds and domain models required. |
| Summer 2026 | The AI Operating System for Companies | Provide reusable evidence, review and continuity functions on which business applications can build. | KAI; Knowledge Governance; TUP; BP/BPL; Continuity | Strong platform fit: aligns directly with the EOS direction; composed business workflows require qualification. |
| Spring 2026 | Cursor for Product Managers | Connect customer evidence to competing product hypotheses, requirements and acceptance tests. | KAI; Context and Constraint Builder; BP/BPL; ValidityBench | Strong fit: a reviewable basis for what to build; product success still needs customer testing. |
| Spring 2026 | AI-Native Hedge Funds | Separate observations from investment hypotheses and challenge backtests, leakage and alternative explanations. | Model/Theory Formation; KAI; BP/BPL; ValidityBench | Supporting research discipline; no return, alpha or investment-advice claim. |
| Spring 2026 | AI-Native Agencies | Turn client briefs into scoped work with review, correction and evidence of acceptance. | Context and Constraint Builder; TUP; BP/BPL; Continuity | Strong service fit: repeatable deliverables; software-like margins must be demonstrated. |
| Spring 2026 | Stablecoin Financial Services | Structure transaction evidence, exceptions and accountable review across service workflows. | AAAM; BP/BPL; Continuity; Standing Grants | Supporting operational rigor; payment infrastructure, licensing and custody remain separate. |
| Spring 2026 | AI for Government | Link administrative recommendations to source records, applicable rules and responsible review. | KAI; TUP; BP/BPL; Continuity | Strong fit: inspectable processing and exception handling; agency adoption/data permission required. |
| Spring 2026 | Modern Metal Mills | Compare production/maintenance options against resource constraints and actual operating evidence. | Model/Theory Formation; TUP; BP/BPL | Supporting: feasible-plan and assumption review; mill engineering and process control remain external. |
| Spring 2026 | AI Guidance for Physical Work | Keep guidance tied to current procedures, worker competence and explicit stop/escalation conditions. | Context and Constraint Builder; KAI; BP/BPL; Continuity | Strong conceptual fit: more context-appropriate guidance; field safety and effectiveness require validation. |
| Spring 2026 | Large Spatial Models | Define physical/spatial consistency tests and evaluate outputs against domain references. | Model/Theory Formation; ValidityBench; BP/BPL | Supporting evaluation infrastructure; no spatial foundation model or geometry capability claimed. |
| Spring 2026 | Infra for Government Fraud Hunters | Organize source-linked allegations, contradictions and investigative questions for competent legal review. | KAI; AAAM; BP/BPL; Continuity | Strong evidence-work fit: traceable case preparation; allegation truth and legal sufficiency require separate judgment. |
| Spring 2026 | Make LLMs Easy to Train | Preserve dataset/model provenance, evaluation criteria and failure lessons across training iterations. | KAI; Knowledge Governance; ValidityBench; Continuity | Supporting: more reproducible training decisions; training compute and orchestration remain separate. |
TUP: Tool Use Planning. BP/BPL: Backpass review and bounded repair/recheck. KAI: knowledge acquisition/integration. Knowledge Governance: reviewed admission, revision and reuse of knowledge. Continuity: preservation and handoff of task state and decisions. Context and Constraint Builder: explicit purpose, context and operating boundaries. Standing Grants: bounded recurring authority under the applicable implementation and issuer policy. Model/Theory Formation: explicit hypotheses, assumptions and rival explanations. AAAM: candidate claim/evidence association support. ValidityBench: candidate evaluation methodology/tooling; task-specific references and qualification remain necessary. TAIE/EMSM: geological research/theory and mathematical resources; transfer requires an explicit domain mapping and new validation. Bounded mathematical checks: selected calculations or local helpers where appropriate, rather than a claim that the full system is executing.
Tool names describe functions that could be selected for each application. The table does not assert that every named component is callable, universally qualified, or already wired into these domains. Source-code availability, local test evidence, empirical efficacy and production admission remain separate. Document-guided methods can support founder-led work today; an available interface cannot replace missing domain evidence, expertise or action authority.
RFS titles and batch assignments were checked on 2 November 2026. The live YC page supplies Winter 2027. Historical lists were reconstructed from dated reproductions and cross-checks because retrieved YC pages default to the current batch. Some secondary lists omit entries or use abbreviated titles. We included Summer’s AI-Native Discovery Engines and Spring’s Large Spatial Models and Make LLMs Easy to Train. Titles use the common short form where reproductions differ.
HSDLG basis: HSDLG_V2 Capability and Standards Overview, TUP 1.0, BP integrated workflow 1.0 and the five retained project studies. Review route: same-assistant, document-guided planning and Backpass; no new native runtime invocation or domain experiment. Review checked title coverage, application-specific contribution, tool-role boundaries, fit strength, study attribution and disclosure. No proprietary implementation, contracts or internal source identifiers are included.
Additional technical documentation, standards, test records and implementation details are available for scoped diligence under an appropriate NDA or contract.