fix(course): ICP journey fixes - branch-aware nav, legacy-page rewrites, citations, covers (40.17)

content/course/tech-for-non-technical-founders-2026/_index.md

@@ -41,7 +41,7 @@ Five modules take you from that idea to a signed paid pilot: test demand with st
 
 The first checkpoint comes fast. Two to three weeks in, at an evenings-and-weekends pace, you'll know whether the idea is worth building at all - a go/iterate/kill verdict on your **Founding Hypothesis** (the one-sentence version of your idea: who the customer is, what problem you solve, why they'd pick you) from real demand data instead of polite nods from friends.
 
-Need orientation first? See the [Quickstart](/course/tech-for-non-technical-founders-2026/quickstart/) (3 min), [FAQ](/course/tech-for-non-technical-founders-2026/faq/) (the most common blockers), or [How This Course Works](/course/tech-for-non-technical-founders-2026/how-this-course-works/) (full route overview).
+New here? Read [How This Course Works](/course/tech-for-non-technical-founders-2026/how-this-course-works/) once - the 10-minute route map - then start [Chapter 1.1](/course/tech-for-non-technical-founders-2026/form-your-founding-hypothesis-90-minute-sprint/). Short on time tonight? The [Quickstart](/course/tech-for-non-technical-founders-2026/quickstart/) (3 min) is the compressed version, and the [FAQ](/course/tech-for-non-technical-founders-2026/faq/) answers the most common blockers.
 
 *Returning? [Skip to the syllabus ↓](#module-map)*
 

content/course/tech-for-non-technical-founders-2026/agency-uses-ai-follow-up-questions/index.md

@@ -180,7 +180,7 @@ Three concrete moves for the next 24 hours, in order.
 
 ## Further reading
 
-- Veracode, [GenAI Code Security Report 2025](https://www.veracode.com/genai-security-report-2025) - 45% of LLM-generated code shipped at least one exploitable security flaw. The data behind Q3 (verification) and Q5 (accountability).
+- Veracode, [2025 GenAI Code Security Report](https://www.veracode.com/blog/genai-code-security-report/) - 45% of tested LLM-generated code samples carried at least one exploitable security flaw. The data behind Q3 (verification) and Q5 (accountability).
 - Stack Overflow, [2025 Developer Survey - AI section](https://survey.stackoverflow.co/2025/) - 84% of developers now use or plan to use AI tools. The market context behind why "we use AI" became a meaningless homepage line.
 - Bleeping Computer, [AI code suggestions sabotage software supply chain](https://www.bleepingcomputer.com/news/security/ai-code-suggestions-sabotage-software-supply-chain/) - the slopsquatting attack vector in plain English. The thing your agency must spot in PR review (Q4).
 - SecurityWeek, [AI Coding Agents Could Fuel Next Supply Chain Crisis](https://www.securityweek.com/ai-coding-agents-could-fuel-next-supply-chain-crisis/) - the practitioner read on why hallucinated package names are now the leading AI-era attack surface.

content/course/tech-for-non-technical-founders-2026/ai-token-bill-dev-shop-pass-through-cost/index.md

@@ -173,9 +173,9 @@ Three actions. In order.
 - [OpenAI API pricing](https://openai.com/api/pricing/) - the gpt-4o, o3, and o1 per-token rates. Note the cached input rates if your team uses prompt caching aggressively.
 - [Cursor pricing](https://cursor.com/pricing) - Pro, Business, and Ultra tiers. Business is the modal agency choice in 2026.
 - [GitHub Copilot pricing](https://github.com/features/copilot/plans) - Individual, Business, Enterprise. The fallback for teams that have not switched to Claude Code.
-- The Pragmatic Engineer, [The state of AI coding tools in 2025](https://newsletter.pragmaticengineer.com/p/state-ai-coding-tools-2025) - per-developer spend ranges from a survey of senior engineers, useful for sanity-checking the numbers above.
-- Simon Willison, [Pelican on a bicycle benchmarks](https://simonwillison.net/2024/Oct/25/llm-pricing/) - running cost comparison across models with public token-cost ranges.
-- Latent Space, [The AI engineer's stack 2026](https://www.latent.space/p/ai-engineer-stack) - which model and tool combinations large engineering orgs settle on, with cost commentary.
+- The Pragmatic Engineer, [Software engineering with LLMs in 2025: reality check](https://newsletter.pragmaticengineer.com/p/software-engineering-with-llms-in-2025) - per-developer spend ranges from a survey of senior engineers, useful for sanity-checking the numbers above.
+- Simon Willison, [LLM pricing calculator](https://simonwillison.net/2025/Apr/10/llm-pricing-calculator/) - running cost comparison across models with public token-cost ranges.
+- Latent Space, [The Rise of the AI Engineer](https://www.latent.space/p/ai-engineer) - which model and tool combinations large engineering orgs settle on, with cost commentary.
 
 Related course posts:
 - ["We Use AI": 5 Follow-Up Questions for Your Agency](/course/tech-for-non-technical-founders-2026/agency-ai-five-questions/) - sister chapter; the 5-question script for AI theatre detection includes the cost question this post expands.

content/course/tech-for-non-technical-founders-2026/build-path-decision-worksheet/index.md

@@ -40,7 +40,7 @@ She showed us the contract on a Tuesday. By Friday we had walked her through the
 
 Friday afternoon, alone, 30 minutes, before coffee runs out. Bring three documents: your filled-in [Validated Problem Statement](/course/tech-for-non-technical-founders-2026/validated-problem-statement-template/) from Chapter 2.5, your filled-in [one-page brief](/course/tech-for-non-technical-founders-2026/vibe-prd-template/) from Chapter 3.1, and a current bank statement showing months of runway. Pen on paper. Phone in another room.
 
-Answer the five questions in order. Each one is factual, not aspirational. "Number of paying pre-orders" is a count from your Stripe dashboard, not a vibe. "Months of runway" is cash on hand divided by monthly burn, not a guess. The matrix at the bottom routes you to Path 1, 2, 3, or 4 based on the five answers.
+Answer the five questions in order. Each one is factual, not aspirational. "Number of Mom Test interviews" is a count on your call log, not a vibe. "Months of runway" is cash on hand divided by monthly burn, not a guess. The matrix at the bottom routes you to Path 1, 2, 3, or 4 based on the five answers.
 
 If you spill past 30 minutes, you are negotiating with yourself. Stop, write the verdict the data supports, take it to one peer or advisor for a 20-minute call, and move on.
 
@@ -65,15 +65,21 @@ Months of runway in the bank: ________________________
 Q1. Is the problem validated?
 ------------------------------------------------------
 
-Counts as YES only if all 3 below are true:
+Counts as YES only if all 4 below are true:
  [ ] 10 or more Mom Test interviews complete
  [ ] Strong past-behavior signal in at least 7 of 10
- [ ] 2 to 5 pre-orders, paid pilots, or annual deposits
-     on the table (real money, refundable is fine)
+ [ ] Smoke test cleared the 6%+ "Promising" band -
+     the share of visitors who left their email on
+     your test page (Module 1, Chapter 1.4)
+ [ ] Prototype test: 4 of 5 people reached the right
+     screen without you coaching them (from Module 2)
 
 LinkedIn likes do not count. "They said they would
-buy" does not count. Money or a procurement-call
-calendar invite counts.
+buy" does not count. What people DID counts - the
+sign-ups on the smoke test and the prototype run, not
+what they said they would do. Pre-orders and paid
+pilots come later (Module 5); you do not need them
+to answer YES here.
 
 VERDICT: [ ] Yes    [ ] No
 
@@ -91,13 +97,21 @@ Q2. How backend-heavy is the build?
 ------------------------------------------------------
 
 Check every TRUE row:
- [ ] Real-time updates (WebSockets, server-sent events)
- [ ] Background queues with retry logic
- [ ] AI inference inside the request path,
-     >$0.01 per call
- [ ] Multi-tenant data with row-level security
- [ ] 5 or more third-party API integrations
- [ ] Regulated data (HIPAA, SOC 2, PCI scope)
+ [ ] Live updates that appear on screen without
+     anyone refreshing the page (like watching
+     someone type in Google Docs or Slack)
+ [ ] Heavy jobs that run in the background and keep
+     retrying if they fail (mass emails, big uploads)
+ [ ] An AI model runs every time someone uses the
+     feature, and each run costs real money (more
+     than a cent per use)
+ [ ] Many separate companies use the same app and
+     none can ever see another company's data
+ [ ] The app connects to 5 or more outside services
+     (payments, email, maps, calendars, and so on)
+ [ ] Regulated data - health records (HIPAA), stored
+     credit-card numbers (PCI), or enterprise
+     security audits (SOC 2)
 
 VERDICT:
  [ ] 0-1 boxes checked   = LIGHT backend
@@ -147,7 +161,7 @@ Money you can commit for at least 6 months:
  [ ] $5,000 to $30,000/mo (small team band)
  [ ] $30,000+ /mo (multi-person team band)
 
-If $0-$400 -> Path 2 (Self-serve / Module 4).
+If $0-$400 -> Path 2 (Self-serve / Ch 4.3 then 4.4).
               Paste one-page brief into Lovable.
 
 If $1.6K-$4K -> go to Q5.
@@ -172,11 +186,12 @@ Has shipped a backend at scale in the last 5 years.
  [ ] Yes, named: ___________________________________
  [ ] No
 
-If YES -> Path 2 (Self-serve / Module 4).
+If YES -> Path 2 (Self-serve / Ch 4.3 then 4.4).
           Use them for the monthly architecture call
           + worst-route code review.
 
-If NO -> Path 3 (Fractional CTO bridge).
+If NO -> Path 3 (Fractional CTO bridge - see the
+         hire-track supplementary reference).
          Buy the same insurance commercially.
 
 ======================================================
@@ -190,13 +205,13 @@ your Notion doc)
 | Path | Choose when | First action this week | Cost |
 |---|---|---|---|
 | **1. Validate without code** | Q1 = No, OR Q3 = less than 4 months | Ship Carrd page + Stripe checkout + Notion FAQ. Send to your 30-name outreach list. | $0 - $300 in tools + optional $100-200 in paid ads |
-| **2. Self-serve build (6A)** | Q1 yes, Q2 light, Q4 = $0-$400/wk, Q5 = senior eng in network | Paste one-page brief into Lovable. Hook Supabase + Stripe + Resend. | $200 - $1,200 / month |
-| **3. Fractional CTO bridge (5.2)** | Q1 yes, Q2 mid, Q4 = $1.6K-$4K/mo OR Q5 = no senior eng | Hire 5-10 hrs/wk Fractional CTO. Use for architecture, PR review, hiring, cost watch. | $1,600 - $4,000 / month |
-| **4. Hire a team (6B)** | Q1 yes, Q2 heavy, Q4 = $5K+/mo | Read draft SOW clause-by-clause. Confirm GitHub/AWS/domain ownership before kickoff. | $30K - $80K / month |
+| **2. Self-serve build** ([Ch 4.3](/course/tech-for-non-technical-founders-2026/self-serve-mvp-stack-lovable-supabase-stripe-2026/) then [4.4](/course/tech-for-non-technical-founders-2026/self-serve-mvp-stack-build-phases/)) | Q1 yes, Q2 light or mid, and either Q4 = $0-$400/wk of your own time, or Q4 = $1.6K-$4K/mo with a senior engineer in your network (Q5 yes) | Paste one-page brief into Lovable. Hook Supabase + Stripe + Resend. | $200 - $1,200 / month |
+| **3. Fractional CTO bridge** ([hire-track reference](/course/tech-for-non-technical-founders-2026/hire-track-supplementary-reference/#the-fractional-cto-bridge)) | Q1 yes, Q2 light or mid, and either Q4 = $1.6K-$4K/mo with no senior engineer (Q5 no), or Q4 = $5K-$30K/mo | Hire 5-10 hrs/wk Fractional CTO. Use for architecture, PR review, hiring, cost watch. | $1,600 - $4,000 / month |
+| **4. Hire a team** ([hire-track reference](/course/tech-for-non-technical-founders-2026/hire-track-supplementary-reference/)) | Q2 heavy (any budget), or Q4 = $30K+/mo | Read draft SOW clause-by-clause. Confirm GitHub/AWS/domain ownership before kickoff. | $30K - $80K / month |
 
 **Failure mode to watch for each path** ->
 - Path 1 -> 0 of 35 click. Pivot the pitch or the problem.
-- Path 2 -> Hits architectural ceiling at ~5K users. Route to 6B for next layer.
+- Path 2 -> Hits architectural ceiling at ~5K users. Route to the [hire-track reference](/course/tech-for-non-technical-founders-2026/hire-track-supplementary-reference/) for the next layer.
 - Path 3 -> CTO drifts into coder. 90-day review on hour allocation.
 - Path 4 -> Spaceship for the wrong moon. Friday demo + Org Chart audit catch it in week 3.
 
@@ -206,9 +221,9 @@ your Notion doc)
 
 > Bad: *"I posted on LinkedIn and got 47 likes and 12 comments. People love the idea."*
 
-> Good: *"I ran 12 Mom Test calls (May 2026 sample). 9 of 12 described the exact problem in past-tense with a number attached (hours per week, dollars per month). 3 of 9 said yes to a $200 refundable deposit on the spot. 2 wired the deposit by Friday."*
+> Good: *"I ran 12 Mom Test calls (May 2026 sample). 9 of 12 described the exact problem in past-tense with a number attached (hours per week, dollars per month). My smoke-test page converted 8% of visitors to email sign-ups, above the 6% 'Promising' bar. 4 of 5 prototype testers reached the sign-up screen without me coaching them."*
 
-The good answer is countable: 12 calls, 9 strong signals, 3 yes-on-the-spot, 2 wired money. The bad answer is a vibe metric (likes) and a hypothetical (love the idea). Likes do not pay invoices. The matrix routes the bad answer to Path 1 regardless of how confident the founder feels, because the data is not there.
+The good answer is countable: 12 calls, 9 strong signals, an 8% smoke-test click rate, 4 of 5 testers through the prototype. The bad answer is a vibe metric (likes) and a hypothetical (love the idea). Likes are not behavior. The matrix routes the bad answer to Path 1 regardless of how confident the founder feels, because the data is not there.
 
 **Q4 - Monthly engineering budget**
 
@@ -228,7 +243,7 @@ The good answer names the person, the relevant experience (Stripe payments backe
 
 ## What to do after
 
-- **Write your verdict at the top of a fresh Notion doc.** One line: *"Path X (name). Started: [date]. Next module: [6A / 6B / 5.2 / Module 2 retake]."* Pin the doc. Re-read the line every Monday for 4 weeks.
+- **Write your verdict at the top of a fresh Notion doc.** One line: *"Path X (name). Started: [date]. Next: [Ch 4.3 self-serve / hire-track reference / back to Module 2 to re-validate]."* Pin the doc. Re-read the line every Monday for 4 weeks.
 - **Share the worksheet result with one peer or advisor in a 20-minute call.** Not three. One. Ask them: *"If the matrix said Path X, what would change your mind?"* Their first answer is the assumption you should pressure-test next. If they cannot name one, the matrix verdict holds.
 - **Schedule the first action for tomorrow morning.** Path 1: ship the Carrd page. Path 2: paste one-page brief into Lovable. Path 3: 3 LinkedIn outreach messages to Fractional CTOs in your industry. Path 4: read the SOW clause-by-clause with a printed pen, not in Google Docs. Calendar the action with a hard start time. The worksheet routes the decision; the calendar invite routes the work.
 

content/course/tech-for-non-technical-founders-2026/channel-selection-before-outbound/index.md

@@ -241,7 +241,7 @@ Pick the channel your interviews already named, commit to it long enough to read
 - Lenny Rachitsky, [How today's fastest growing B2B businesses found their first ten customers](https://www.lennysnewsletter.com/p/how-todays-fastest-growing-b2b-businesses) - the channel breakdown across Figma, Stripe, Slack, Notion. Channel choice is the first lever, not the last.
 - Brian Balfour, [Why Product Market Fit Isn't Enough](https://brianbalfour.com/essays/product-market-fit-isnt-enough) - the case for channel-model fit as a distinct requirement from product-market fit. Strong reading for any founder who has a working product but no acquisition motion.
 - Andrew Chen, [The Law of Shitty Clickthroughs](https://andrewchen.com/the-law-of-shitty-clickthroughs/) - why every channel degrades over time and why early-stage founders need to pick the channel that works now, not the channel that worked for a more established company 3 years ago.
-- First Round Capital, [From 1 to 1,000 Users](https://review.firstround.com/from-1-to-1000-users-tactics-from-airbnb-tinder-etsy-reddit-and-more/) - channel selection stories from Airbnb, Tinder, Etsy, Reddit, including how each picked their first acquisition channel based on where their early customers actually lived online.
+- Lenny Rachitsky, [How the biggest consumer apps got their first 1,000 users](https://www.lennysnewsletter.com/p/how-the-biggest-consumer-apps-got) - channel selection stories from Airbnb, Tinder, Etsy, Reddit, including how each picked their first acquisition channel based on where their early customers actually lived online.
 
 > **Done:** your worksheet is filled, one channel is chosen, and you have written your commitment statement (channel name + why + evaluation criteria).
 >