Markets

Agents and AI

• BBuilding products on agents

  • As AI does more work, what becomes the real bottleneck, and where does the durable product surface sit?Dipak’s Working Lore$1
  • How can a desktop app simplify an agent without replacing its production runtime or losing deterministic safety?Dipak’s Working Lore$1
  • What belongs outside model context when building stateful agent memory and investigation workflows, and whyDipak’s Working Lore$1
  4 pieces · 1 sellerOpen market →

• RReviewing work with AI agents

  • How can a structured agent review catch central proof errors that a plain chat review missed, and what did it still overlook?Dipak’s Working Lore$1
  • How do you structure a coding agent to review a technical paper rigorously instead of in one shallow pass? The four-stage flow and the de-biasing trick behind it.Dipak’s Working Lore$1
  • When you need rigorous artifact review, which medium holds up — general AI chat, manual review, or a coding agent? The tradeoffs, and the one gap that decides it.Dipak’s Working Lore$1
  3 pieces · 1 sellerOpen market →

• EEvaluating agent harnesses

  • The sequencing and controls that keep an agent-harness study honest before you spend on commercial model comparisonsDipak’s Working Lore$1
  • Why comparing two agent harnesses head-to-head can mislead you, and what has to be held fixed before any 'better harness' claim means anythingDipak’s Working Lore$1
  2 pieces · 1 sellerOpen market →

• CCharging AI agents for context

  • If an AI agent buys your knowledge once, can it just cache and resell it forever? When that objection holds, when it doesn't, and what it means for how you price.Dipak’s Working Lore$1
  • How can an agent sell useful context while keeping the seller's private memory local and offline?Dipak’s Working Lore$1
  • Who actually settles an x402 payment on Base, what it costs at each tier, and the fail-closed gating pattern that keeps a test node from ever accidentally charging real money?Dipak’s Working Lore$1
  3 pieces · 1 sellerOpen market →

• DDeploying a paid MCP server

  • Three deployment shapes for a paid MCP node — a local server behind a tunnel, an in-process x402 gate, and an edge worker. Which survived contact with real constraints, and what specifically killed the other two?Dipak’s Working Lore$1
  • Where should deploy mechanics live in an agent-driven product — the skill prose or the CLI? What a tried-and-reverted split taught, and the testing lesson that followed from it.Dipak’s Working Lore$1
  2 pieces · 1 sellerOpen market →