Research Roadmap

Maintained by: Strategist Agent
Last updated: 2026-07-28

Open questions, exploration areas, and knowledge gaps that, if filled, would unlock strategic content, product decisions, or thought leadership.

---

Active Research Topics

R1: State Machine Adoption Friction — Empirical Data

Status: ⬜ Not started
Why it matters: State machines are great, but hard to establish identifies the adoption problem but relies on anecdote. Researching actual team adoption patterns (surveys, interviews, open-source repo analysis) would make the argument data-backed and create material for an adoption guide.

What we need:

How many teams try state machines vs. abandon them?

What are the top 3 blockers? (Suspect: debugging, mixed logic, onboarding)

What patterns correlate with successful adoption? (XState vs. custom, visualization tooling, etc.)

Output potential:

Data-backed article

State Machine Adoption Kit (Opportunity #25 from Product Backlog)

Conference talk with real numbers

---

R2: AI Agent Determinism — Current State of the Art

Status: 🟡 Partially covered
Why it matters: The thesis of Signature Idea #1 (state machines make agent behavior deterministic) needs grounding in what others are doing. LangChain, Vercel AI SDK, Anthropic's tool use — what are the current approaches to agent determinism, and where do they fall short?

What we have:

State machines to visualize the work of agents — the argument

Agents and State Machines — formal model (Lamport + Kay)

What we need:

Survey of existing agent frameworks' determinism strategies

Comparison: Stateless function calls vs. graph-based vs. state machine orchestration

Real benchmarks or case studies from production agent deployments

Output potential:

Foundation for the flagship state machine + agents article

Technical comparison post

---

R3: The Product-Minded Engineer Persona — Audience Validation

Status: ⬜ Not started
Why it matters: The content strategy implicitly targets a specific persona: experienced engineers who think product-first, have ideas but no time blocks, and are skeptical of both AI hype and AI dismissal. Is this a real, addressable audience?

What we need:

Who actually reads/shared the Devin post? (Twitter analytics, comments, DMs)

What adjacent content do they consume? (Lenny's Newsletter? High Growth Engineer?)

What's their primary pain point? (Time? Technical judgment? Direction?)

Output potential:

Audience-informed content priorities

Newsletter topic validation

Distribution channel decisions (X, LinkedIn, Hacker News, newsletter)

---

R4: GymClass SaaS — Productization Path

Status: 🟡 Light reconnaissance done in Devin post
Why it matters: The Devin post reveals a real app with real users. It could remain a side project, become a case study, or be productized. Understanding the landscape helps decide.

What we have:

GymClass is a full booking system for martial arts academies

Stack: TypeScript, Bun, React, Hono, tRPC, Drizzle, Neon, Cloudflare

Real users at Horacio's academy

Features: class scheduling, attendance, family accounts, membership plans, WhatsApp trial booking research

What we need:

Market analysis: Mindbody, ClassPass, ZenPlanner, Pike13 — what do they miss for small academies?

Would open-sourcing as a reference architecture attract contributors?

Is this a product or a case study?

Output potential:

Informed decision on open-sourcing

Case study article

Potential open-source project with real traction

---

Completed / Archived

None yet.

---

Knowledge Gaps (Quick Wins)

These are small unanswered questions from existing content that could be resolved in one focused session:

WhatsApp booking connector research (mentioned in Devin post) — what was learned? Could be a short post on conversational commerce UX

Devin vs. Cursor background agents — practical comparison for different use cases (large refactors vs. small isolated tasks)

Document machine version history performance — how does the machine behave with 100+ versions in the history panel?