← Back to all playbooks

Thariq

THARIQ SHIHIPAR: Inside Claude Code
Start: 06:02:20
Sketch note for Thariq's session
TL;DR

AI engineering moats are built on 'vectors, not points,' requiring a ruthless willingness to delete custom orchestration code as model capabilities advance to keep the implementation 'native.'

Key Insights

The 'Delete Code' Engineering Mandate

06:17:01

Avoid over-engineering complex orchestration for current model limitations; when a new model 'unhobbles' a capability, you must delete your custom code to prevent technical debt from blocking native model performance.

Shift from Ephemeral To-Dos to Persistent Tasks

06:19:06

Move away from flat, session-specific to-do lists toward 'Tasks' (V2) which support dependency graphs and persist state across multiple agents and sessions.

Elicitation as a Model 'Unhobbling' Strategy

06:04:50

Instead of the model guessing intent, implement specific 'ask_user' tools that allow the model to pause and resolve ambiguity, effectively expanding its 'space in the box' for complex planning.

Moats are Vectors, Not Points

06:24:05

Competitors can clone a specific feature (a point), but they cannot clone the 'vector'—the institutional knowledge of what was tried and discarded, and the direction of the next iteration.

Systems

Spec-Driven Long-Running Agent Loop

  1. Spend 30 minutes manually writing a high-density technical specification.
  2. Use 'Spec Mode' to reduce all possible ambiguity for the agent.
  3. Hand off the spec to a high-reasoning model (e.g., Opus) for long-duration execution.
  4. Allow the model to run autonomously for extended periods now that the 'ambiguity floor' is lowered.
Tools: Claude Code, Opus

Engineer-Led Ground Truth Feedback

  1. Eliminate dedicated QA and Support teams for dev tools.
  2. Require engineers to read GitHub tickets and social feedback directly to maintain 'vibe' and 'feel'.
  3. Identify 'meta-patterns' in unstructured user feedback.
  4. Synthesize patterns into 'composable building blocks' rather than one-off feature requests.
Tools: GitHub, Twitter

Tools

Claude CodeOpus 4.5Sonnet 4.5GitHubConductor