Execution Plan: Phase 0 → 1 → 2

March 2026 — Built for two working parents with a 4yo and 2yo, a full-time job, and 5-10 hrs/week of margin.

Builds-on: gap-analysis-henry-to-next-stage Builds-on: elite-overproduction-and-status-signaling Informs: Projects/sigil, Projects/tech-blog, Projects/polyphony

Guiding Constraint

Every action in this plan converts existing assets into visible, monetizable form. No new building unless the market demands it. The enemy is not lack of skill — it's the gap between what you've built and what anyone knows about.

Time budget: 5-10 hrs/week total across all phases. Not negotiable upward — you have two kids under 5.

Co-founder note: Phase 2 has a potential partner (friend with business ownership experience). This is noted where relevant and changes the math significantly.

Phase 0: Go Public (Weeks 1-4)

Goal: Establish visible proof that you know what you're talking about. Create the foundation that Phases 1 and 2 build on.

Time cost: 3-5 hrs/week

Week 1: Launch the Blog

You already have:

• Astro + AstroPaper blog scaffolded at /Projects/tech-blog
• One published post (Edge-LLM intro, Feb 2026)
• GitHub Pages deploy workflow (currently manual trigger)

Actions:

1. Enable auto-deploy on push — uncomment the push trigger in .github/workflows/deploy.yml. 15 minutes.

2. Publish the Edge-LLM post as-is. It exists. It's written. Ship it. Don't touch it again.

3. Write Post #2: "What I Learned Building a SaaS from My Own Agency Pain"

   • Angle: You ran y-designs. You know the proposal → delivery → approval workflow is broken. You built Sigil to fix it. Here's what the problem actually looks like and how AI-assisted spec mapping changes the game.
   • This is NOT a product launch post. It's a credibility post that happens to mention Sigil.
   • Length: 1,000-1,500 words. One evening.
   • End with: "I'm looking for agencies and PMs to test this with. DM me."

4. Set up minimal social distribution:

   • Post the Edge-LLM piece on LinkedIn with a short intro (2 paragraphs max). Tag it #LLM #WebGPU #AI.
   • Post the Sigil/agency piece on LinkedIn when ready. Tag it #agencylife #productmanagement #AI.
   • If you're on Twitter/X, same thing. If not, LinkedIn alone is fine.

5. Write Post #4: "I Made Different AIs Argue About a War"

   • Angle: MiroFish — you took an existing geopolitical simulation idea (original MiroFish by 666ghj, uses a single model for all agents) and rebuilt it with heterogeneous model casting and a tatemae/honne (public/private) dual-stack architecture. Different LLMs play different countries. The results were more realistic than single-model runs.
   • This is your most novel and shareable content. It's inherently interesting to non-engineers too (geopolitics, AI, emergent behavior).
   • The $25 cost and 23-round Hormuz scenario make it concrete and reproducible.
   • End with: applications beyond geopolitics (business scenario planning, red teaming, wargaming).

Deliverables by end of Week 1:

• Blog live with auto-deploy
• Edge-LLM post distributed on LinkedIn
• Sigil/agency post drafted
• MiroFish post drafted

Week 2: Write Post #3 + Start Outreach

Write Post #3: "Multi-Agent Orchestration in Production: Patterns That Actually Work"

• Angle: You built a multi-agent review framework at work. Most content about agents is toy demos. Write about what's different in production: fallback chains, error handling, cost management, when NOT to use agents.
• Anonymize Brightwheel specifics. Focus on patterns, not employer details.
• This is your "I'm a serious practitioner" post. It targets the engineering audience AND positions you for the education play later.
• Length: 1,500-2,000 words. Two evenings.

Start Sigil outreach (bridges to Phase 1):

• Make a list of 10 people from your network to show Sigil to. Prioritize:
  • PMs you know personally
  • Agency owners / freelancers / consultants
  • Friends who manage client delivery in any form
• Send 5 messages this week: "I built something from my y-designs days. Can I show you in 20 min and get honest feedback?"

Deliverables by end of Week 2:

• Post #3 drafted or published
• 10-person Sigil demo list created
• 5 outreach messages sent

Weeks 3-4: Publish, Distribute, Refine

Publish remaining posts. By now you should have 4 posts live:

1. Edge-LLM (browser-native LLM inference) — technical audience
2. Sigil/agency pain post — business/PM audience
3. Multi-agent production patterns — practitioner audience
4. MiroFish (heterogeneous LLM geopolitical sim) — broadest appeal, shareable to non-engineers

Distribute each on LinkedIn the day it publishes. One post = one LinkedIn share. Don't overthink the copy.

Watch for signals:

• Which post gets the most engagement? That tells you where the audience is.
• Who reaches out? Those are warm leads for Phase 1 or 2.
• What questions do people ask? Those become future posts or workshop topics.

Optional but high-leverage:

• Cross-post to dev.to or Hashnode (takes 10 minutes per post, expands reach)
• If multi-agent post gets traction, pitch it to a newsletter (The Pragmatic Engineer, TLDR, ByteByteGo)

Phase 0 Exit Criteria:

• 3 blog posts live and distributed
• Blog getting some traffic (even if modest)
• Sigil outreach in progress (5-10 conversations initiated)
• You have a public URL you can point anyone to that shows what you know

Phase 1: Sigil Validation (Weeks 3-8, overlapping with Phase 0)

Goal: Find out if Sigil is a product people will pay for, or a tool you built for yourself. Either answer is valuable.

Time cost: 3-5 hrs/week (demos + follow-up)

The y-designs Advantage

Most founders build products and then go looking for the problem. You ran the business (y-designs), lived the pain, and built the solution. This means:

• You can speak the customer's language, not just the engineer's language
• You know which features matter and which are vanity
• Your demo isn't "look at this tech" — it's "remember when this happened to you? here's how I fixed it"

Use this in every conversation. Lead with the y-designs story, not the tech stack.

Week 3-4: First 5 Demos

Target: 5 demos from your outreach list. Mix of PMs, agency people, freelancers.

Demo structure (20 minutes max):

1. (2 min) "I ran an agency called y-designs. The worst part was the gap between what the client approved and what we actually delivered. Proposals lived in one tool, specs in another, delivery in a third. Clients would say 'that's not what I asked for' and we'd have no receipts."
2. (10 min) Walk through Sigil: proposal → client approval → AI-assisted stage mapping → staged delivery with sign-offs
3. (5 min) "What does this look like in your world? Where does your process break?"
4. (3 min) "If this existed and worked, what would you pay for it?"

What you're listening for:

• "I need this" — potential early customer. Follow up within 48 hours.
• "This is cool but I wouldn't pay" — ask why. The "why not" is more valuable than the praise.
• "I'd need X feature" — write it down. If 3+ people ask for the same thing, that's signal.
• "I know someone who needs this" — get the intro. This is the most valuable response.

Week 5-6: Refine Based on Feedback + 5 More Demos

Adjust the pitch based on what you heard in weeks 3-4. Maybe PMs care about something different than agency owners. Maybe the AI mapping is the hook, or maybe it's the client approval trail.

Second batch of 5 demos. These should include:

• At least 2 people from referrals (not your direct network) — referral interest is stronger signal than friend interest
• At least 1 person who matches your ideal customer profile from the Sigil docs (boutique agency, 2-15 people)

Week 7-8: Decision Point

By now you have 10 data points. Score them:

If total score is 15+: Sigil has legs. Move to paid pilot (offer 3-month access at $50-100/month to 3-5 early users). This isn't about revenue — it's about getting real usage data.

If total score is 5-14: Sigil solves a real problem but the product-market fit isn't tight yet. Keep it alive but don't pour more time in. Shift energy to Phase 2.

If total score is below 5: Sigil is a personal tool, not a product. That's fine — you learned a ton building it and it powers your credibility. Archive the go-to-market effort and go all-in on Phase 2.

Phase 1 Exit Criteria

• 10 demos completed
• Clear signal on whether to pursue Sigil as a business or archive it
• If positive: 3-5 paid pilot users
• If negative: clear articulation of why not (this feeds your blog and education content)

Phase 2: LLM Education for Business Owners (Weeks 6-16, overlapping with Phase 1)

Goal: Validate and launch a paid education offering. This is where the co-founder changes everything.

Time cost: 5-8 hrs/week (split with co-founder = 2.5-4 hrs each)

Why This Works for You

The Product: What You're Selling

Not a coding bootcamp. Business owners don't want to learn to code. They want to:

1. Understand what's possible with LLMs (so they stop getting sold snake oil)
2. Know how to evaluate AI tools and vendors (so they make good bets)
3. See real examples of LLM integration in businesses like theirs (so they can act)
4. Have a network of peers going through the same transition (so they're not alone)

Format options (choose one to start):

Recommended start: Half-day workshop. Lowest risk, fastest to market, generates testimonials for bigger formats.

Week 6-8: Curriculum Design (With Co-Founder)

Split the work:

• You: Technical content. What LLMs actually do, what they don't, real demos, production patterns, common failure modes, how to evaluate vendors.
• Co-founder: Business framing. ROI models, implementation roadmaps, case studies, networking facilitation.

Workshop outline (half-day, 4 hours):

Materials needed:

• Slide deck (you both contribute)
• 1-page "AI Readiness Assessment" handout (co-founder drafts)
• "Vendor Evaluation Checklist" handout (you draft from real experience)
• Post-workshop email sequence (3 emails: recap, resources, upsell to next offering)

Week 9-10: Pre-Sell the First Workshop

You do NOT need a website. You need:

1. A Google Doc or Notion page with: title, date, what they'll learn, price, "register" link (Stripe payment link or even Venmo)
2. Messages to your combined networks

Outreach script (adapt to your voice):

"Hey [name] — [co-founder] and I are running a half-day workshop for business owners on how to actually use AI in your business (not the hype, the real stuff). I've built production AI systems for [years] and [co-founder] has [their background]. We're capping it at 15 people. [Date]. $750. Interested?"

Target: 10-15 attendees at $500-1,000 each = $5-15k gross revenue (split with co-founder).

Where to find them:

• Your dev/PM/designer network → ask them: "Who do you know who runs a business and is trying to figure out AI?"
• Co-founder's network (this is why having a partner matters — 2x the reach)
• LinkedIn post: "We're running a workshop. Here's what we'll cover. DM for details."
• Wife's BID network (business improvement district = literally a directory of local business owners)
• The friend who paid $8k — ask them to refer people who would have benefited from a better version

Week 11-12: Deliver the First Workshop

Before:

• Dry run with each other (1 hour). Time each block. Cut what doesn't fit.
• Send attendees a pre-workshop survey: "What's your biggest AI question?" — this lets you customize live and shows you care.

During:

• Record it (with permission). This becomes content and proof.
• Collect testimonials at the end. Even 2-3 quotes are gold.
• Take photos (for social proof on LinkedIn).

After:

• Send recap email within 24 hours
• Post on LinkedIn: "We just ran our first workshop. Here's what business owners are actually asking about AI." (This is your next blog post too.)
• Follow-up call 1 week later

Week 13-16: Iterate or Scale

Based on Workshop #1 data:

If it works, the scaling path is:

1. Half-day workshop ($750) → proves the content
2. 2-day intensive ($2,500) → adds networking, deeper content
3. Cohort program ($5,000) → recurring revenue, community
4. Corporate training ($10k/day) → highest per-session revenue
5. Eventually: recorded course as passive income layer (but NOT first — live is where you learn what works)

The Combined Timeline

Week  Phase 0              Phase 1              Phase 2
──────────────────────────────────────────────────────────────
1     Blog live + Post #1
2     Post #2 + outreach    Outreach begins
3     Post #3              First demos
4     Distribute           More demos
5                          Refine pitch
6                          Second batch          Curriculum design
7                          Decision point        Curriculum design
8                          Pilot users?          Pre-sell workshop
9                                                Pre-sell workshop
10                                               Finalize materials
11                                               DELIVER WORKSHOP #1
12                                               Follow-up + iterate
13-16                                            Scale or pivot

The Math: What Success Looks Like at Week 16

Conservative scenario (things go okay):

• Blog: 3-5 posts live, modest LinkedIn traction, a few hundred readers
• Sigil: 2-3 paid pilot users at $50-100/mo = $100-300/mo
• Workshop: 1 delivered, 10 attendees at $750 = $7,500 (split = $3,750 your share)
• Total new income: ~$5k
• Real value: public proof + market data + a co-founder relationship tested under pressure

Optimistic scenario (things click):

• Blog: 1 post gets real traction, drives inbound
• Sigil: 5 pilot users + clear signal to pursue
• Workshop: Sold out, second one scheduled, corporate training inquiry
• Total new income: ~$15-25k
• Real value: a second income stream that doesn't depend on Brightwheel + a network that knows what you do

What both scenarios give you that you don't have today:

• A public identity beyond "staff eng at Brightwheel"
• Revenue that isn't salary
• Data on which opportunity (Sigil vs. education vs. something else) has the most pull
• A co-founder relationship with reps
• A network that sees you as a builder, not just an employee

Risks and Mitigations

Week 1 Checklist (Start Here)

• Enable auto-deploy on tech blog
• Publish Edge-LLM post to LinkedIn with 2-paragraph intro
• Draft Sigil/y-designs blog post (one evening)
• Check Brightwheel moonlighting/non-compete clause
• Text co-founder: "Let's do the workshop. Here's my rough plan. Coffee this week?"
• Make the Sigil demo list (10 names, PMs and agency people)
• Send first 3 outreach messages for Sigil demos

That's 7 actions. All can be done in one focused week at 5 hours.