11. Appendix: How to Build Your IT Team in the AI Era (For Startups and Small Businesses)

Foreword: This Is Not an “Expansion Guide”

In the AI era, the act of “building an IT team” itself has fundamentally changed.

It is no longer a question of “how many people to hire, what roles to divide, what tech stack to use”, but a more essential question: “Who do you want to be responsible for your system?”

If you are:

• Preparing to establish your first technical team
• Or your existing team has fewer than 5 people
• Or you are hesitating “whether to hire more people”

Then the goal of this appendix is singular: To help you avoid walking the old path of “looking professional but actually being extremely expensive” in the AI era.

I. A Counter-Intuitive Conclusion

In the AI era, the vast majority of companies do not need a “fully configured” IT team.

What you truly need is often just:

• A very small number of people who can make judgments
• Plus AI as the default executor

If you rush to build a full suite of “Frontend, Backend, Ops, Database, QA” at the very beginning of your team, you are most likely not building capability, but manufacturing Coordination Costs in advance.

II. First Principles: What Problem Are You Solving?

Before considering “hiring”, I suggest you answer three questions first:

1. Is this system an internal efficiency tool, or an external product?
2. What is the cost of failure?
   • Can you tolerate incomplete features?
   • Will a single data error be fatal?
3. Who will maintain this system in the long run?

If you cannot clearly answer the third question now, it means you are not ready to hire an “executor”.

III. The First Tech Role in AI Era: Not Engineer, But “Owner”

For a 0 → 1 team, the first technical role is critical.

❌ Not Recommended Choices

• Outsourcing that only writes code: Often lacks long-term commitment to the business.
• Executor who only knows one stack: Difficult to cope with full-link technical challenges.
• Tech person to “use cheaply for now”: The invisible technical debt caused by low judgment is often much more expensive than the salary difference.

✅ More Reasonable Choice

A person who can be responsible for the “System as a Whole”.

This person may not be the fastest coder or the one who knows a framework best, but they must have three things:

1. Can independently complete a closed loop of a complete system: From requirement understanding, technology selection, to launch and maintenance.
2. Possess AI Driving Capability: Knows when to use AI to speed up, and when absolutely not to trust AI.
3. Willing to bear responsibility for results.

You can call this role: Product Engineer, System Engineer, or more bluntly: Tech Lead / Head of Engineering.

IV. A Realistic Minimum Team Structure

Structure 1: 1 Person + AI (Very Early Stage)

• Suitable for: Startups, internal tools, MVP validation stage.
• Configuration: 1 Fully Responsible Engineer + AI as default collaborator.
• Key Points:
  • All code must be “human-readable”.
  • AI-generated content must be strictly reviewed.
  • Do not pursue perfection, only pursue maintainability.

Structure 2: 2–3 Person Small Team (Most Recommended)

• Suitable for: Validated requirements, starting to have users, need certain stability.
• Configuration: 1 Tech Lead + 1–2 Empowered Engineers + AI as default execution accelerator.
• Key Points:
  • No clear frontend/backend boundaries.
  • Everyone has their own “responsibility module”.
  • No “handover”, only “I own this”.

V. About “Whether to Hire Junior Engineers”

This is a question you must be extremely cautious about.

In the AI era: Junior engineers are not “cheap labor”, but “high-risk assets”.

If you do not have the ability to:

• Provide complete context
• Conduct high-quality reviews
• Cover for their judgment errors

Then please do not rush to hire junior engineers.

Compared to “hiring a new person”, what you are more likely to need is: Clearer requirements, more stable system boundaries, and more mature decision-making mechanisms.

VI. Don’t Rush to “Specialize”, Ensure “Accountability” First

Many companies fall into a misunderstanding too early: “Wait until the scale grows, then split the responsibilities.”

But in the AI era, the more reasonable order is reversed:

1. First ensure that every part of the system has someone responsible for it.
2. Then introduce specialized support when necessary.

Roles like Ops, Security, Performance, Data can be specializations, but should not be isolated positions from the start.

VII. The Real Duty of Managers in Small Teams

If you are a founder or business leader, please remember:

In the AI era, the easiest reason for small teams to fail is not technical incompetence, but loss of control over expectations.

Your three most important things are:

1. Distinguish Demo, System, and Product: Stay sober at all times.
2. Don’t treat “AI can do it” as “It should be done now”: Understand physical constraints of engineering.
3. Shield unreasonable time pressure for the Tech Lead: Protect the team’s judgment.

VIII. Finally, A Calm but Sincere Advice

If you read this far and only want to remember one sentence:

In the AI era, building an IT team is not for “faster delivery”, but for “not making fatal errors in faster execution”.

Speed can be left to AI; Judgment, Choice, and Responsibility must still be completed by humans.