LLM Wiki 101Starter Kit

Ask through curated Wiki Lens.

Ask naturally; the system routes to topic packs and exposes the source-backed reasoning path.

Try the Lens demoBuild the first Wiki

Grounded answer packet

1. Recommendation

2. Lens route and reasoning path

3. Wiki pages and source caveats

Context assets

Wiki packs become reusable context objects instead of one-off prompt folders.

Automatic routing

The question selects the relevant packs; users do not need to pick an expert mode first.

Traceable answers

Every response shows the lens route, pages used, caveats, and next move.

The product move: context cards, not character cards.

JanitorAI proves that reusable context objects can drive repeat usage. This version swaps fictional characters for serious Wiki Lens Packs that help with product, pricing, founder decisions, and agent design.

Universal ask
router/v0

Available Lens Packs

First Principles42 pages / 18 sources
Product Taste36 pages / 21 sources
Pricing Power31 pages / 14 sources
Founder Decisions48 pages / 26 sources
Wiki context route
2 lenses
Pricing Powerscore 2Agent Builderscore 1

Direct answer

Use Pricing Power + Agent Builder before giving advice. The product should answer the user's question directly, then show which wiki context shaped the answer.

Reasoning path

  1. 01Normalize the question into a decision, critique, research, or build task.
  2. 02Route the task to the smallest useful set of Lens Packs instead of asking the user to choose first.
  3. 03Answer with a clear recommendation, then expose the supporting wiki pages and unresolved assumptions.
  4. 04Offer one next move that can be saved back into the user's own LLM Wiki.

Wiki pages used

  • Principles/Value Metric.md
  • Playbooks/One-Time Pack vs Subscription.md
  • Architecture/Agent Loop Anatomy.md
  • Checklists/Eval Before Launch.md

Caveats and next move

  • - Do not impersonate living people or imply endorsement from named founders.
  • - Treat celebrity or founder references as source-backed context, not as a simulated identity.
  • - If the relevant Lens has weak source coverage, the answer should say what is unsupported.

Create a saved decision note with the routed lenses, cited pages, caveats, and the user's final choice.

How the Lens system works

The user experience stays as simple as a chat box. The differentiation happens behind the answer: curated Wikis route, constrain, and cite model reasoning.

01

Universal Ask

The user asks directly. They do not need to choose a theme, folder, or expert mode before writing the question.

02

Context Router

The router selects the smallest useful set of Wiki Lens Packs, such as First Principles, Pricing, or Product Taste.

03

Grounded Answer

The answer shows the lens route, reasoning path, wiki pages used, caveats, and the next move to save.

First Lens marketplace preview

These are not bots pretending to be famous people. They are structured knowledge packs with pages, sources, usage boundaries, and decision templates.

First Principles Lens

Reduce a vague decision to constraints, primitives, and falsifiable assumptions.

Lens
42wiki pages
18source notes
  • Product bets
  • Technical tradeoffs
  • Business model clarity

Product Taste Lens

Review product surfaces through clarity, density, hierarchy, trust, and taste.

Lens
36wiki pages
21source notes
  • Landing page critique
  • Feature polish
  • UX positioning

Pricing Power Lens

Turn willingness to pay, packaging, proof, and switching cost into a pricing decision.

Lens
31wiki pages
14source notes
  • Pricing pages
  • Subscription decisions
  • Paid pack packaging

Founder Decisions Lens

Judge a move by opportunity cost, operating leverage, cash flow, and compounding learning.

Lens
48wiki pages
26source notes
  • Startup focus
  • Market choice
  • Founder operating cadence

AI Agent Builder Lens

Separate model, tools, memory, workflow, evaluation, and user-facing failure modes.

Lens
39wiki pages
17source notes
  • Agent products
  • RAG workflows
  • AI feature architecture

Business model: sell the asset, subscribe to the router.

The first sale can still be a downloadable Wiki Pack. The recurring product is Hosted Ask over many premium Lens Packs and private user Wikis.

Free

Universal Ask demo, a few public Lens Packs, and SEO pages for high-intent concepts.

Paid packs

Downloadable Wiki Context Packs for Obsidian, Markdown, and prompt workflows.

Hosted Ask

Subscription access to premium Lens routing, saved decisions, and private user Wikis.

Implementation boundary for the first build

The first version should prove the interaction and the product story. Real retrieval, accounts, payments, creator submissions, and private uploads can come after the Lens metaphor works.

No celebrity impersonation, voice cloning, or fake endorsement.
Named founders can appear only as source-backed context inside a broader topic.
Every premium Lens needs source indexes, caveats, and unsupported-claim handling.
The first build stays deterministic; real LLM retrieval comes after the product story is proven.