The Playbook: Tactical Moves for the AI Era

This is Part 2 of a 3-part series. Part 1 covered the bubble vs. build-out framework. This post explores what that means tactically for different roles.

---

If you accept the premise from Part 1—that both the financial bubble and technological build-out are real—then the question becomes: What does that mean for how I work? How I position myself? What I build?

I don’t have final answers. But I’m seeing patterns. And some of those patterns are working.

Here’s what I’m observing across different roles.

For Solo Devs / Small Teams: The Arbitrage Window

The financial bubble actually helps right now. Here’s why:

As long as the bubble inflates, enterprises are spending massive capex on AI infrastructure. That means:

• Compute is getting cheaper (more supply)
• Tools are getting better (more investment in dev tools)
• Clients are pre-sold on AI (you don’t have to convince them it’s real)

But you’re not betting on the bubble. You’re betting on productivity.

The Pattern I’m Seeing Work:

Pick a Vertical

Don’t be “an AI developer.” Be “the developer who uses AI to build X for Y industry.”

Examples I’m tracking:

• “I build compliance automation for healthcare using AI-assisted development”
• “I build real-time dashboards for logistics companies in 6 weeks instead of 6 months”
• “I deliver complete e-commerce platforms for DTC brands in 4 weeks”

You’re selling outcomes, not technology.

Build the Full Stack

This is the unlock. With AI, you can now deliver:

• The codebase (80 hours for a $2.5M platform)
• The go-to-market strategy (14 documents, competitive analysis, pricing)
• The market validation (financial models, ROI calculations)
• The technical architecture (production-ready, scalable, secure)

You’re not selling “AI development.” You’re selling “complete business solutions delivered at 50x speed.”

Price on Value, Not Time

Traditional team: $1.4M, 18 months, 8 people

You with AI: $80k in time cost, 80 hours, 1 person

Your price: $300k-$500k, 6 weeks delivery

This is arbitrage. You’re capturing the gap between:

• What the market expects it should cost (based on old models)
• What it actually costs you (based on AI-enabled speed)

Build Cash Flow, Not Valuation

Here’s where the bubble vs. build-out distinction matters:

What I’m avoiding:

• Raising VC money to “scale fast”
• Hiring a team before I have revenue
• Building for an exit/acquisition

That’s betting on the bubble. If the market corrects, funding dies and valuations crash.

What I’m trying instead:

• Charge for delivery upfront or milestone-based
• Keep the team tiny (1-3 people)
• Stack cash
• Build a portfolio of case studies

Building a business that survives a correction because it’s profitable from day one.

Timeline: The Uncomfortable Question

The arbitrage window (the gap between what’s possible and what the market knows) is probably 18-24 months.

But the technological transformation is 10+ years.

The strategy I’m testing:

• Months 0-24: Capture arbitrage. Deliver at 50x speed. Build cash reserves. Establish positioning.
• Year 2-5: The market catches up. The advantage narrows. But I’ve got case studies, cash, and relationships.
• Year 5-10: I’m not the “AI developer” anymore. I’m the established player in my vertical with proven delivery models.

If there’s a financial correction in year 2-3, I survive because I’m profitable and I deliver real value.

That’s the bet, anyway.

For Enterprise Leaders: The Brownfield Opportunity

If you’re a CTO or VP of Engineering, the bubble vs. build-out split changes the strategy.

The Risk I’m Watching:

If you’re betting on the bubble (buying AI stocks, waiting for “the next big thing”), you’re exposed when it corrects.

The Opportunity I’m Seeing:

If you’re investing in the build-out (actual productivity gains from AI-assisted development), a market correction actually helps.

Here’s why:

When the bubble pops:

• Compute gets cheaper (oversupply of data centers)
• Talent gets cheaper (laid-off AI engineers need jobs)
• Tools get better (companies compete on product, not hype)
• Executive pressure to “show ROI” intensifies (you’re already delivering it)

You become the person who says: “While everyone else was speculating, we were building. Here’s the 40% cost reduction we delivered.”

The Approach I’m Testing:

Phase 1: Map the Legacy (0-3 Months)

Use AI to comprehend your brownfield codebase.

This used to take 6 months and 5 senior engineers. Now it takes a weekend and 1 architect.

Tools: Claude Code, Cursor, Aider

Process:

1. Feed entire codebase to AI: “Map all services related to customer billing”
2. Ask for dependency graphs: “Show me every database call in the authentication flow”
3. Identify risk: “Find all security vulnerabilities in payment processing”

Output: Complete map of your legacy system. This is infrastructure. It survives a market correction.

Phase 2: Quarantine and Wrap (3-12 Months)

Don’t rewrite the legacy system. You’ll fail.

Instead:

1. Build the Facade: Modern API layer around legacy code
2. Use AI for the Noise: Data mapping, transformation, boilerplate (AI generates 90%)
3. Humans Manage Signal: Business logic, validation, edge cases (humans do 10%)

Why I Think This Works in a Correction:

Traditional approach: $50M, 3 years, 200-person consulting team

This approach: $5M, 12 months, 3-person team + AI

When budgets get cut (and they will in a correction), your project survives because the ROI is clear and the cost is manageable.

Phase 3: Retrain, Don’t Replace (12-24 Months)

Your existing developers aren’t obsolete. They’re using the wrong tools.

The developers who know your 20-year-old business logic are your most valuable asset. You can’t hire that knowledge. You can train AI fluency.

Training Program I’m Experimenting With (8 weeks):

• Week 1-2: AI fundamentals (what LLMs can/can’t do)
• Week 3-4: Intent-driven development (describing intent, validating outputs)
• Week 5-6: Code review for AI outputs (spotting gaps, security issues)
• Week 7-8: Systems thinking (architecting with AI as infrastructure)

Why I Think This Survives a Correction:

When layoffs come (and they will), companies that fired their institutional knowledge and tried to replace it with “AI developers” will be screwed.

You kept your knowledge. You just upgraded the tools.

Phase 4: Shift the Metrics (24+ Months)

Stop measuring:

• Lines of code written
• Hours worked
• Tickets closed

Start measuring:

• Business value delivered
• Cycle time (intent → production)
• System complexity reduced

When the CFO asks “what are we getting from this AI investment?”, you have an answer: “40% faster delivery, 60% reduction in maintenance costs, zero increase in headcount.”

That survives any market condition.

For Consultants: Adapt or Die (No, Really)

If you’re in consulting, the bubble vs. build-out distinction is existential.

The Old Model (Dead in a Correction):

200-person “transformation” team. $50M engagement. 3 years. Selling process, not outcomes.

This model depends on:

• High corporate spending (bubble money)
• Long sales cycles (executives with budgets)
• Tolerance for “strategic work” with unclear ROI

When the correction comes, this dies first.

CFOs look at $50M, 3-year projects with no guaranteed outcome and say “no.”

The Model I’m Building:

1-3 person team. $300k-$1M engagement. 6-12 weeks. Delivering outcomes, not process.

Team Size: 1-3 people

Team Composition:

• 1 Architect (systems thinker, business strategist, technical validator)
• 1 Utility Engineer (AI wrangler, infrastructure specialist, deployment expert)
• 1 Translator (client interface, requirements gatherer, stakeholder manager)

Often one person plays multiple roles.

Service Offering:

“I’ll deliver a production-ready solution in 6-12 weeks.”

What I’m Trying to Deliver:

• Working code (deployed, monitored, production-ready)
• Strategic documentation (why this solution, how it fits the business)
• Knowledge transfer (training their team to maintain it)
• Roadmap (what comes next, how to scale)

Pricing:

$300k-$1M per engagement. Fixed scope. Outcome-based.

Why I Think This Survives a Correction:

When budgets tighten:

• $50M, 3-year projects get canceled
• $500k, 12-week projects with guaranteed delivery get approved

You’re 50x faster and 90% cheaper. And you deliver actual working software, not PowerPoint decks.

The Split That Matters:

Parrot Consultants (Die in a Correction):

• Sell “AI strategy” without building anything
• Deliver “roadmaps” and “frameworks”
• Staff large teams of junior consultants doing process work
• Depend on executive budgets staying inflated

Architect Consultants (Survive a Correction):

• Deliver working solutions
• Prove ROI in weeks
• Work as tiny, highly skilled teams
• Price on value delivered, not hours worked

If you can’t deliver in the second model, you were a parrot all along. The correction will expose that.

For Junior Engineers: The New Baseline

If you’re starting your career, the bubble vs. build-out split is actually good news.

Why?

Because in a correction, the signal separates from the noise faster.

The Old Path (Dead):

Graduate → junior developer role → write CRUD apps for 2 years → promotion

This path assumed:

• Companies need bodies to write boilerplate
• You learn by grinding through repetitive work
• The ladder has rungs

AI has removed the bottom rungs. Companies don’t need “junior developers” to write boilerplate.

The Path I’m Seeing Work (Harder, But Higher Leverage):

Year 1: Junior Architect

You’re not writing boilerplate. AI writes that. You’re learning:

• Systems thinking (how does this component affect the whole system?)
• Intent clarity (can I describe what I want clearly enough for AI to build it?)
• Domain knowledge (pick an industry, learn it deeply)
• Code review (can I spot gaps in AI-generated code?)

You work with AI from day one.

Year 2-3: Directing Complexity

You’re directing AI to build increasingly complex systems. You’re making architectural decisions. You’re validating AI outputs against business requirements.

Year 4+: Full Architect

You can take a business problem, design a solution, direct AI to implement it, and validate the output.

You’re delivering value at 10-50x the rate of traditional developers.

Why I Think This Survives a Correction:

When layoffs come, companies cut:

• Junior developers doing boilerplate (AI does that now)
• Mid-level developers who never learned to work with AI (competing in a dead category)

They keep:

• Architects who deliver business value at 10x speed
• People who understand the business domain + AI fluency

The bar is higher. But if you clear it, you’re more valuable—and more recession-proof—than traditional developers ever were.

The Uncomfortable Truth:

Not everyone will clear this bar. This path is harder than the old one.

But the ones who do are building careers that survive market cycles.

---

This is Part 2 of a 3-part series. Part 1 covered the framework. Part 3 will explore the real patterns I’m seeing and what I’m actually doing.

These are just the patterns I’m noticing—in real time, with incomplete data. If you’re seeing something different, I’d be curious to hear it.