← Ben Lai

Vision ledger

The first divergence pass over databookman.com, preserved as a durable planning artifact: raw ideas, finalists, reconciliation notes, and the current shipped state.

Product at its best

Public identity engine

Databookman.com is a builder's public identity engine: sharp essays about the agent economy paired with proof that the author ships what he writes about. The design, one-URL-per-thing discipline, and self-grading surfaces make the work feel permanent, inspectable, and worth revisiting.

Finalists

Where each pick landed

Raw list

52 ideas kept intact

L1 — 10x the core loop

  1. Every quantitative essay ships a runnable artifact.
  2. Live essays: number-driven posts recompute nightly from live data.
  3. Receipts appendix: transcripts, logs, and code behind each essay claim.

L2 — Kill a load-bearing constraint

  1. Reader marginalia and annotations at the edge.
  2. Other people and agents rebut essays in public.
  3. Auto-translated JP, KR, and zh editions with per-locale URLs.
  4. Ninety-second narrated audio editions and a podcast feed.
  5. Project rail auto-updates from repo telemetry.
  6. Named readers see a research-notes layer inline on public URLs.

L3 — Cross-domain transplant

  1. /now dashboard as Bloomberg terminal for current building, reading, and shipping.
  2. Public ship-something-weekly streak from GitHub activity.
  3. Essays display their own revision diffs and dated claim changes.
  4. Newsletter with memory of what each reader opened.
  5. One sharp micro-take per day feeding the essay queue.
  6. Dated falsifiable predictions graded annually.

L4 — New user class

  1. AI agents as readers through llms-full.txt and a machine-readable claims feed.
  2. Recruiter and investor proof page with signed commits, live products, and benchmarks.
  3. Per-essay founder-register explanation toggle.
  4. Every essay tagged with a try-this-on-your-repo exercise.

L5 — Future press release

  1. Essays cited in frontier-lab system cards or YC required reading.
  2. The blog that graded its own predictions five years straight.

L6 — Gasp demo

  1. Talk to the corpus in the author voice, grounded only in essays, with citations.
  2. Paste your own API bill into The output token tax.

L7 — Compounding asset

  1. Claims registry: every falsifiable claim extracted, dated, and tracked.
  2. Curated agent-economy bibliography per tag with sources and counter-evidence.

L8 — Agent-native

  1. Site as MCP server with an inbox for other agents.
  2. Red-team agent writes the strongest rebuttal before publication.
  3. The site registers as an actex-connect agent with a passport.

L9 — Audience taste x owner insight

  1. Steal-this endings and a toolbox of real versioned configs.
  2. Problem-first entry for readers arriving with a live pain.
  3. Paste-your-situation router to relevant essays.
  4. Reader-runnable receipts for testing claims on your own stack.
  5. Examples that adapt to a pasted stack.
  6. Unlisted bonus notes or counter-evidence after an essay.
  7. Reply-to-RSS threaded answers that become public letters.
  8. Since-you-were-here diff with no accounts and no tracking.
  9. Series recaps based on what a reader already read.
  10. Subscribers get essays early plus the steelman draft.
  11. Ninety-second hire dossier with receipts.
  12. Evidence-only dossier Q&A.
  13. What I would do first at your company.
  14. Prove-me-wrong bounties with credited rebuttals.
  15. Linkable agent-economy glossary.
  16. Monthly office-hours slots bookable from the site.
  17. Claims API with resolving citations.
  18. Functional easter egg addressed to agents that fetch the site.
  19. Weekly travel photo with a three-line story.
  20. 404 page serving a random sharp paragraph.
  21. One-question-a-day widget from the corpus.
  22. Print and PDF mode for paper reading.

Reconciliation

What this ledger is not