Ideas
Everything being explored at various stages — from early seeds to actively developed projects.
Active
2Actively developed and maintained
Elements of Agentic System Design
projectWhat are the concrete code patterns behind "intelligent" agent behaviors? Framework complete, expanding to book.
Idyllic Runtime
projectWhat's the minimal infrastructure layer for agent prototypes? Stateful sessions, streaming, real-time sync on Cloudflare Durable Objects.
Exploring
5Researching and prototyping
Cortex
projectWhat representations work for turning unstructured text into connected knowledge?
mdagent
projectCan agent behavior be fully specified in markdown? Prototype exploring the boundaries.
Primitives > Vertical AI Apps
thesisMost vertical AI agent startups (2024-2026 cohort) will fail. The value accrues to composable building blocks, not applications.
Users Will Compose Their Own Agents
thesisThe default mode shifts from "buy agent product" to "assemble agent from primitives." AI makes custom software composable by end users.
Wiki/Belief Engine
projectHow does an AI-native research org externalize its cognition?
Seed
6Early ideas worth investigating
Agent Primitives Empirical Study
projectSurvey existing agent systems, extract recurring patterns, test which deserve elevation.
Delegation Model Experiments
projectPrototype different human-agent handoff patterns, measure effectiveness.
Dynamic Context Loading
patternSuccessful agent primitives provide a mechanism for the model to decide to load context dynamically.
Legible Representation Prototypes
projectBuild different visualization/inspection tools for agent state, see what makes systems understandable.
Minimal Constraint Pattern
patternPrimitives that give model more freedom outperform those that prematurely constrain. Skills > MCP.
System Capability Bindings
patternPrimitives that bind prompts to system capabilities (tools → function calls, skills → code execution, heartbeat → cron).
Archived
0Paused or completed
Nothing here yet