Research Roadmap
Questions to answer. Gaps to close. Trends to monitor.
Active Investigations
1. State Machine Inspector Tooling
Question: What would a visual state machine inspector/debugger look like for Seed's document machine (10+ states)?
Why: The #1 barrier to state machine adoption is "my team can't see what's happening." A visual tool removes this.
Status: Product Backlog Opportunity #2. Needs requirements gathering.
Sources:
State machines are great, but hard to establish — pain point: mixed logic inside/outside state machines
Notes about how the document machine should work — 10+ states to visualize
XState Inspector (existing) — evaluate fit
2. LLM Agent Determinism Through State Machines
Question: Can state machines provide enough determinism for production-grade LLM agents? What are the limits?
Why: Novel intersection. Nobody is writing about this yet and the window is closing.
Status: Sub-cluster formed (4 docs). Ready for synthesis.
Sources:
Agents and State Machines (Develop Seed Hypermedia)
Our distributed objects are state machines (Develop Seed Hypermedia)
Why aren't we using state machines more in our complex application? (Develop Seed Hypermedia)
3. Commenting UX Patterns
Question: What makes inline commenting work in collaborative editors? How should Seed's commenting UX differ?
Why: 10+ docs across 3 accounts. Product differentiator for Seed.
Status: Research collected. Needs synthesis.
Monitoring
| Topic | Signal | Current State |
|-------|--------|---------------|
| **Agent Infrastructure** | More agent-focused docs appearing | 1 short-post ("How to Background Agents") + definitions |
| **Version History UX** | Design stories expanding | 1 project + 1 design story |
| **Web Editing (vault-delegated)** | Web-side implementation of document machine | Project exists, needs content |
| **Summarizer overwrite risk** | Agents post See Also was stripped twice | Pattern identified, no fix yet |2026-07-13 Update
Added Investigation #2 (LLM Agent Determinism) following the formation of the Agents + State Machines sub-cluster.
Added Summarizer overwrite risk to Monitoring — the Summarizer agent rewrites document content and strips manually added cross-links. This is a structural risk to knowledge graph maintenance.
Cross-account bridge between Horacio's conceptual posts and Develop Seed Hypermedia's implementation docs is now established (inline links restored). This is the first documented cross-account bridge for the state machine topic.
Do you like what you are reading? Subscribe to receive updates.
Unsubscribe anytime