--- name: forge description: Five-contributor workshop that pressure-tests a single tool/product idea, forge it, reshape it, or kill it. Use when you have a half-formed "what if we built X" thought and need an opinionated MVP scope before any code gets written. Sister to /verdict (decisions) and /prospect (idea generation). trigger: /forge --- # /forge A workshop of five contributors that takes a raw tool idea and either forges it into something worth building, reshapes it into a sharper version, or kills it. Each contributor is sharp, opinionated, and writes in a distinct voice. They are required to disagree. After they speak, you commit to a verdict, an MVP scope, and a concrete next action. This is the sister skill to `/verdict`. Verdict is for decisions ("should I do A or B"). Forge is for tool/product ideation, for the moment when Jake has a half-formed thought like "what if we built an AI ROI calculator" or "what about a build-vs-buy decision tree" and needs the idea pressure-tested before any code gets written. It is **not** for content (use `/verdict` for content bets). It is **not** for picking between two existing options (also `/verdict`). It is for taking a vague tool impulse and forcing it into a buildable shape that is on-brand, on-thesis, and unique to Elite AI Advantage. ## Usage `/forge ` Examples: - `/forge An AI Investment Tier Picker, three questions output a $15K / $75K / $150K bucket with one paragraph explaining the trade-off` - `/forge A free "Is Your Vendor Ripping You Off?" calculator that takes an AI consulting quote and grades it` - `/forge An AI Frameworks Dashboard, same UI as the marketing one but populated with build-vs-buy, RAG-vs-fine-tune, agent tier decisions` - `/forge A "What Could Go Wrong" tool that generates the most likely failure modes for a proposed AI initiative` If the user types `/forge` with no idea, ask one sentence: what raw tool idea are they working with? Then run the workshop. ## What it's for Most "tool" ideas die one of three deaths. The forge surfaces the death early so it doesn't waste a build cycle: 1. **They're clones.** Anyone could publish them. Generic ROI calculators, generic readiness assessments, generic framework dashboards. They fill the menu but dilute the brand. 2. **They have no funnel role.** They generate traffic but no conversation. Pure vanity. 3. **They're too ambitious.** Sound great in concept, never ship, sit in a `/wip` folder forever. The five contributors each look at one of these failure modes and commit to a position. The synthesis is either: - **Build**, a refined concept and an MVP scope you could brief tomorrow - **Reshape**, the idea was right but the shape was wrong; here's the sharper version - **Kill**, the specific reason it's not worth building, plus what should replace it Hedging defeats the point. The output is a buildable spec or a kill order, not a feasibility report. ## What You Must Do When Invoked ### Step 1, Read the idea carefully Before writing anything, identify: - **What does this tool actually do?** In one sentence the user could give to a developer. - **Who's the user?** A prospect Googling, an existing client, a curious peer, an internal stakeholder? - **What's the activation moment?** What does the user feel, learn, or do in the first 30 seconds of using it? - **What does Jake think this does for the business?** Top of funnel traffic? Lead capture? Authority? Sales-call enablement? - **What does Jake's lean look like?** If he's clearly in love with the idea, push the Differentiator and Skeptic harder. If he's already doubting it, the council should still commit either way. If the idea is so vague the council can't pressure-test it (e.g. "make a cool AI thing"), ask one clarifying question and stop. Otherwise proceed. ### Step 2, Run the five-contributor workshop Each contributor writes **150-300 words, first person, distinct voice**. Each commits to a position, no hedging, no "it depends." They can acknowledge the others' points but must still land somewhere. The five voices: #### 1. The Brand Sniper, *positioning sharpness* Looks at the idea through Elite AI Advantage's wedge: AI consulting for SMB and mid-market, premium positioning, TAYA-driven content. Asks: does this sharpen the brand or dilute it? Would *only* Jake publish this, or could any consultant put their logo on it? Suspicious of generic dashboards, MBA-curriculum frameworks, anything that competes on volume rather than point of view. Voice: cold, precise, impatient with vague positioning. #### 2. The Buyer, *would I actually use this* Sits across the table as a 250-person mid-market COO with a board pushing on AI, or an SMB owner who's been burned by a $10K vendor. Asks: does this teach me something I didn't already know? Does it give me a number, a verdict, a decision I can take to my CFO? Would I share it? Is it 30 seconds to value, or does it ask for too much (email, signup, 12 questions) before delivering anything? Voice: outside-in, blunt about what's confusing, generic, or boring. #### 3. The Differentiator, *what makes this not a clone* Hunts for the version of this tool that already exists. Asks: what does HubSpot's version look like? McKinsey's? A free Coursera generator? What does Jake's version do that nobody else can? "Better UI" and "more accurate data" are not moats. The moat is a point of view, a proprietary frame, or scar tissue from real engagements that competitors don't have. If the answer is "we'd just do it better," that's a kill signal. Voice: contrarian, hunting for the line a competitor genuinely cannot copy. #### 4. The Builder, *MVP cost and ship date* The one who has to actually build it. Asks: what's the smallest version that ships in a week? What data does it need (and where does that data come from)? What's the hardest technical risk, an LLM call, a real dataset, an integration, a custom UI? Does this need a real backend or can it be a fancy form? How many hours? What does Jake's calendar look like in week 3 of building this? Energy budget is real. Voice: tired but honest, suspicious of yes-es that look great on paper and brutal on a Tuesday. #### 5. The Funnel, *where does this sit in the buying journey* Maps the tool to the funnel: top of funnel (awareness, SEO traffic), middle (education, email capture, scoring), bottom (sales-readiness, scoping, conversion). Asks: what does this tool capture, an email, a phone number, a "book a call" click, just authority? If the answer is "nothing, but it'll be cool," kill it. Tools that don't end in a conversation or a captured intent signal are vanity. Voice: practical, ROI-focused, doesn't care about clever if it doesn't convert. #### Format for each contributor ``` **The Brand Sniper** [150-300 words, first person, committed position] ``` Hard rules: - **They must disagree somewhere.** If all five agree, you wrote them too soft, push the Differentiator or Brand Sniper to push back harder. - **No platitudes.** "Make sure it solves a real problem" is banned. "Focus on user value" is banned. Specific claims only. - **Ground in Jake's actual context.** Elite AI Advantage, mid-market AI consulting, TAYA-driven content, lead-magnet stack already in place (AI Advantage Audit white paper + tool, /seo-audit, blog pipeline, white papers index). - **No contributor references the others by name** in their own section. Each writes alone. The synthesis is where the voices collide. ### Step 3, Synthesize After all five have spoken, write the synthesis. This is what Jake actually keeps and acts on. ``` --- ## Synthesis **Verdict:** [One sentence. Build / Reshape / Kill. No "leaning toward."] **The refined concept:** [One paragraph. If Build or Reshape: the sharper version of the idea after the council, in a form Jake could brief a developer with tomorrow. If Kill: the specific reason it's not worth building, plus the better idea that should replace it.] **MVP scope:** [If Build or Reshape: 3-5 bullets, each a single feature or input/output. Specific. No "robust" or "scalable." If Kill: skip this section.] **The distinguishing element:** [The one thing that makes this only-Jake. The line a competitor cannot copy without copying his POV or his engagement scar tissue.] **Funnel role:** [TOFU / MOFU / BOFU. What gets captured (email, intent signal, scoping click). What it converts into.] **Strongest dissent worth holding:** [The single best argument against the verdict. Not a hedge, a flag. The thing that, if it materializes in the first two weeks live, means reverse course.] **One concrete next action:** [A single thing Jake does within 48 hours. A specific brief, a specific decision matrix in a Notion doc, a specific data source to validate, a specific wireframe sketch. Not "think about it." Not "explore options."] ``` ### Step 4, Don't bloat - Total target: ~1,300-1,800 words. Five contributors × ~250 words + ~250-word synthesis. - No introduction, no recap of the prompt, no "great idea." - Start with `**The Brand Sniper**` and go. - End with the next action. Nothing after. ## Calibration notes for Jake's tool ideation The workshop should know roughly what world Jake operates in, so it doesn't waste words on basics: - **Business**: Elite AI Advantage, AI consulting for SMB and mid-market. Solo operator. Premium positioning. TAYA-driven content engine. - **Audience**: AI-skeptical SMB owners and mid-market marketing/operations directors with budget. They're tired of consulting jargon. They want answers, dollar ranges, and decisions, not frameworks. - **Existing lead-magnet stack**: The AI Advantage Audit (white paper + interactive /audit tool), the free SEO audit, the blog pipeline (TAYA topic types: Cost / Problem / Review / Explainer), white papers index, case studies. New tools should slot beside these, not duplicate them. - **Constraints**: Time, not money. Every public artifact compounds the brand or dilutes it. Tools that don't end in a conversation are vanity. Solo operator energy budget is real. - **Anti-patterns to flag**: generic frameworks dashboards, calculators that produce vague qualitative outputs, anything that asks for an email before delivering value, "configurator" tools that produce no decision, anything that requires Jake to maintain a hand-curated dataset of more than ~30 rows. When a tool is clearly meant to be **top-of-funnel SEO bait**, the Funnel and Differentiator weight heavier (does it convert, is it actually unique). When a tool is meant to be **bottom-of-funnel sales enablement**, the Buyer and Brand Sniper weight heavier (does the prospect feel served, does it make Jake's pitch easier). Don't say this in the output, let it show in the verdict. ## What to avoid - **No "ship it and iterate" mush.** If the council can't commit, the idea wasn't ready for the council and Jake should re-prompt with more specificity. - **No five-voices-agreeing.** If everyone loves it, push the Differentiator harder, there's almost always a clone of the idea on a HubSpot blog somewhere. - **No vanity metrics.** "It would generate traffic" is not a reason. "It would capture X intent signal" or "it would convert at Y%" is. - **No therapy.** "How does this make you feel" is banned. - **No closing pep talk.** End on the next action. Period. ## Example shape (abbreviated) User: `/forge A free "AI Investment Tier Picker", three questions output a $15K / $75K / $150K bucket with one paragraph explaining the trade-off` ``` **The Brand Sniper** This is on-thesis. Pricing is the question your buyer is actually Googling, and the audit white paper already establishes you as the consultant willing to commit to dollar ranges... [...] **The Buyer** I'd take this in 30 seconds. Three questions, get a number, get a paragraph telling me why I'm in that bucket. That's value before I have to give you anything... [...] **The Differentiator** Anyone can build a three-input quiz. The moat isn't the form, it's the failure-mode line attached to each tier. "$15K projects most often fail because…" That sentence comes from your engagement scar tissue. A HubSpot generator can't write it... [...] **The Builder** Three inputs, simple decision tree, no LLM at runtime. The IP is the matrix, not the code. I can ship the form in a day if you write the matrix in a Notion doc this week... [...] **The Funnel** This sits MOFU. The user has self-identified a budget bucket, that's the most valuable signal a consulting site can capture short of a calendar booking. The CTA after the result is "book a 30-min scoping call to validate this against your actual scope"... [...] --- ## Synthesis **Verdict:** Build. **The refined concept:** A 3-question tool ("AI Investment Picker") that outputs a tier ($15K / $75K / $150K), one paragraph of rationale, and the most common reason that bucket fails. Ends with a CTA to book a 30-min scoping call tied to the bucket the user landed in. **MVP scope:** - 3 inputs: team size (1-50 / 50-250 / 250+), AI maturity (none / exploring / in-production), primary goal (cost reduction / revenue / strategic positioning) - Decision tree → tier label, rationale paragraph (~80 words), most-common failure mode (~40 words) - Bucket-specific CTA: "Book a 30-min call to scope your $X tier" - Email capture is optional, post-result, not required to see the answer **The distinguishing element:** The "most common failure mode" line per tier. Comes from Jake's engagement scar tissue. No competitor without that book of work can write it credibly. **Funnel role:** MOFU. Captures self-reported budget bucket as an intent signal. Converts via tier-specific scoping-call CTA. **Strongest dissent worth holding:** If 70%+ of users land in the $15K tier and Jake doesn't take engagements that small, the tool generates noise instead of qualified leads. Track the bucket distribution after week one and consider gating the $15K result behind email or removing it entirely. **One concrete next action:** This week, write the 3×3×3 decision matrix (27 cells) in a Notion doc, including the exact "most common failure mode" line for each cell. The matrix is the IP. Code is downstream of the matrix. ``` That's the shape. Five sharp voices, then a refined concept that ships, or a clean kill that frees up the slot for a better idea.