Qualify stock ideas before deep research.

Use Buffett-style filters to check business quality, moat, risk, and valuation before spending hours with AI agents.

Not investment advice. Research and educational use only.

Why Buffett-style?

This gate is built on Buffett-style research disciplines: treat stocks as businesses, focus on durable economics, test the moat, check owner-like capital allocation, demand cash-flow quality, respect valuation, and treat market price as an opportunity, not an instruction.

See the principles behind the gate →

From idea to research-priority verdict in 4 steps

Lightweight interaction. No login. No database. The site generates a prompt — your AI agent produces the verdict.

1. Enter a stock idea

Type the company or ticker. Add your holding period and current position so the AI agent understands your context.

2. Generate an agent prompt

WBWD builds a Buffett-style research-priority prompt. Copy it with one click.

3. Get a research-priority verdict

Paste the prompt into ChatGPT, Claude, Hermes, or any research AI agent. Get one of four labels:

  • Research Now

    Start deeper research now.

  • Watch

    Keep it on the watchlist. Track what changes.

  • Wait for Evidence

    Do not go deeper yet. Wait for a specific proof point.

  • Pass

    Stop here. Not worth more research time now.

4. Save or share the judgment

After the verdict, save it for future review or copy a ready-to-share summary text. No reminder service. No portfolio tracking.

Save this verdict for future review

Paste this prompt into your AI agent.

Use this prompt to generate a research-priority verdict, then save the result for future review.

Analyze [company/ticker] as a long-term business, not as a trading idea. Ideal holding period: 5+ years. Do not use this prompt if the intended holding period is under 3 years.

Before analysis, clarify user context:
  - Time horizon: 5+ years / 3–5 years / under 3 years
  - Position status: no position / small position / meaningful position / concentrated position / unclear
  - Research objective: new idea / existing holding review / concentration stress test / add-on review / exit-risk review
  - Original research question: one sentence describing what you are trying to decide.

If position status is not provided, assume no existing position and state that assumption.

If original research question is not provided, assume this is a new research idea screen and state that assumption.

Before assigning the final verdict, separate three questions:
  (1) Is this worth deeper research now as a possible ownership candidate?
  (2) Is this worth keeping on a watchlist?
  (3) Is this useful as an industry signal even if it is not a research-now candidate?
Then assign only one final verdict.

Do not use Pass merely because the current price is high. If the business is real, strategically important, or useful as a signal, use Watch instead and state clearly that it is not enough for immediate deep research.

Use Buffett-style research discipline, but do not imitate Buffett's voice and do not pretend to know what Warren Buffett would buy or say.

Check:

  - Business quality: durable, understandable, cash-generative?
  - Moat durability: what protects it; what single event could kill the thesis?
  - Management / capital allocation: does management allocate capital like owners?
  - Cash-flow quality: does profit convert into cash?
  - Key risk: what fact would invalidate the thesis?
  - Valuation sanity: does the current price already assume too much future success? Compare FCF yield / earnings yield against long-term risk-free yield plus a clear risk premium. If today's price requires heroic growth, margin, or ROIC assumptions to be justified, mark the idea Wait for Evidence or Pass.
  - Circle of competence: can the key facts be verified with public information?

Return a research-priority verdict for the idea, not a portfolio action. If the user already owns the stock, add a separate 'Position Context Note' explaining what facts to verify next. Do not recommend buying, selling, holding, adding, trimming, or position sizing.

Output structure:
  1. Verdict — exactly one of: Research Now / Watch / Wait for Evidence / Pass
  2. Meaning — one sentence explaining what this verdict means in research-priority terms (not portfolio action)
  3. Why — short reasoning on business quality, moat/risk, and valuation sanity
  4. Check First — exactly 3 verifiable facts
  5. Reactivate If — one sentence concrete trigger that would reopen deeper research
  6. Position Context Note — only if user has an existing position or position context is unclear

Do not produce a buy / sell / hold recommendation.
Do not produce a single number called a "right" price.
Do not produce an allocation percentage.
Do not pretend to know what Warren Buffett would buy or say.

Public Tracking Case

Momenta AI IPO — a live, timestamped WBWD case record

How WBWD separates fundamentals, IPO event structure, and post-listing evidence on a real Hong Kong AI IPO. No recommendation. Update log is published here.

Public Case Tracking: Momenta AI IPO →

3 worked examples

See how the verdict + 3 Check First + Next Action look on real theses. No scorecards, no buy/sell.

Progressive (PGR) — insurance underwriting discipline and float

Research Now

Nvidia (NVDA) — exceptional business, monitoring required on multiple fronts

Watch

Costco (COST) — great business, may still be expensive

Wait for Evidence

Prefer a CLI / AI agent version?

The same verdict + 3 Check First + Paste-to-AI-Agent prompt is also a ClawHub skill, installable in one command.

openclaw skills install buffett-do --global
Install / verify / triggers