AI is leading startups toward smaller executive teams, new roles like Head of AI, and leaders who can manage both people and automated systems. VCs advising on hiring now weigh a candidate's AI fluency and ability to run lean teams alongside traditional experience.
I hear a version of this from talent leads all the time. A founder pings you about a VP of Sales search, and the real question underneath is different. Do they need a VP at all, or a strong player-coach with good tooling? Should the marketing lead also own RevOps? And who owns the AI work that three teams are now doing on their own?
The old advice was simple: hire the person who already scaled the function at a company one stage ahead of you. I don't think that advice is wrong. But it misses something now. A leader who built a 60-person sales org may build a 60-person sales org again by reflex, even when the company needs 15 people and a well-run stack. Your job on the platform team is to help founders see that difference before they sign an offer.
What is actually changing in the executive team
Experience still matters. It just isn't the whole test anymore. Here's what keeps coming up when I talk to platform leads and founders.
Fewer executives, hired later. Founders hold off on a full C-suite longer. They hire senior individual contributors first and add a layer of management only when the work truly needs one.
Wider roles. You see more combined seats. A Head of Growth owns marketing and sales development. An operations lead also owns finance systems and internal automation. A product lead owns the AI roadmap.
Mixed spans of control. A leader might manage five people plus a set of automated workflows and agents that do work a junior team once did. Managing that second group is a real skill. Somebody has to define the work, spot-check the output and fix it when it drifts. In the good setups, the tools do most of the work and a person reviews the rest.
Data first, agents second. This is the order most leaders get backwards. No one has good data. An agent running on a messy CRM produces slop, just faster. The leaders getting real results fixed their data before they built agents. A leader who gets this plans for the cleanup. A leader who doesn't promises the board a headcount cut that never shows up.
New titles. Head of AI, AI engineer, applied AI lead. Some of these are real jobs. Some are titles a founder adds because an investor asked about the AI strategy (you can usually tell which in one conversation).
A scorecard your platform team can hand to founders
Most founders don't have a structured way to compare executive candidates. A simple scorecard fixes that, and it gives your talent team a shared language across the portfolio. Score each dimension from 1 to 4:
- 1: No evidence.
- 2: Talks about it well, but no concrete examples.
- 3: Clear examples from past roles.
- 4: Clear examples, plus the candidate taught others to do it.
| Dimension | What good looks like | One interview question |
|---|---|---|
| Hands-on AI fluency | Uses AI tools daily in their own work, not only through their team | "Show me a workflow you changed with AI in the last six months." |
| Lean team design | Adds headcount only after the process is clear | "How would you run this function with half the team you had before?" |
| Systems thinking | Documents processes, measures them, improves them | "Walk me through a process you built that still runs without you." |
| People leadership | Develops managers, handles conflict, keeps good people | "Tell me about someone you promoted and how you got them ready." |
| Judgment on risk | Knows where automation is unsafe for customers or data | "Where would you not let AI touch your function, and why?" |
| Data instincts | Fixes the inputs before automating, can trace a number to a source | "When an AI workflow gave you a wrong answer, how did you find out?" |
| Domain depth | Deep knowledge of the buyer, market or craft | Role-specific |
Ask founders to agree on the weighting before the first interview. For a first sales leader, domain depth and people leadership might count double. For an operations lead, systems thinking and AI fluency might matter most. If you agree first, nobody gets to tune the weights afterward to fit the candidate they already like (and people really do this).
How to update your talent playbook
You don't need to rebuild your whole talent function at once. A crawl, walk, run rollout works fine here.
Crawl: update the templates. Go through your job description templates and interview kits. Add one AI-fluency question to every leadership kit. Cut the lines that assume large teams, like "built and managed a team of 40+." Replace them with outcomes, like "grew pipeline while keeping the team small."
Walk: add org design to office hours. Before a founder opens an executive search, offer a 30-minute org design session. Use three questions:
- What outcomes does this leader own in the first 12 months?
- Which of those outcomes need people, and which need systems?
- What does this function look like at twice the revenue?
This is my favorite part. A lot of searches change shape after this conversation. Some become a senior IC hire. Some become two part-time advisors.
Run: build a bench, and move on the signal. In your talent network, tag operators who have run lean teams with heavy automation. Then don't wait for the founder to ask. The most important leadership roles often aren't public. They sit deep in a board deck, or in a hiring problem that shows up in three monthly updates in a row. Treat that as the trigger: signal → two or three names → the founder's inbox, the same day the role takes shape. If you're still building out the talent function, I wrote more about what portfolio companies need from VC talent teams.
Mistakes worth flagging to founders
A few of these show up often enough that your team should name them early.
- Hiring a Head of AI as a signal. If the title exists to answer an investor question, the person will struggle to find real ownership.
- Copying a later-stage org chart. A Series A company with a Series C org chart burns cash on coordination.
- Hiring a leader who only delegates. On a small team, a leader who can't do the work can't judge the work, whether a person or a tool did it.
- Ignoring the manager of the machines. If agents now handle support triage or outbound research, somebody needs to own their quality and spot-check the output. Write that ownership into a role.
Here's a message your team can adapt when a founder asks for a search:
Happy to help with the VP Marketing search. Before we start, can we do 30 minutes on what this person owns in year one? A few founders in the portfolio found they needed a senior hands-on marketer plus better tooling, not a VP with a big team. Might be the same for you, might not. Worth checking before we write the job spec.
Where human leadership still matters most
AI changes how much work a small team can do. It doesn't change the parts of leadership that are about people: hard conversations, firing, setting culture, earning a big customer's trust in person, staying steady in a bad quarter. Those still need a person with judgment and credibility. And I expect the face-to-face part to matter for a long time.
When you advise founders, keep both halves in view. A candidate with great AI fluency and weak people skills is still a risky executive hire, which is why people leadership stays on the scorecard. The same goes for your own team. I covered that in how AI is changing the role of a VC platform team.
Do startups need a Head of AI?
Most early-stage startups don't need a dedicated Head of AI. If AI is the product, the CTO or a senior engineer should own it. A separate Head of AI makes sense when AI work spans several teams and nobody clearly owns model quality, vendor choices, cost and risk.
Here's a short checklist to use with a founder. A dedicated role starts to make sense when most of these are true:
- Three or more teams run AI workflows or features in production.
- Model and API costs are a visible line in the budget.
- Customers or regulators ask how AI handles their data.
- The CTO's time on AI decisions is crowding out the core roadmap.
If only one or two are true, a clear owner inside an existing role is usually enough. Write that ownership into the person's goals so it doesn't fall between teams.
How should VCs evaluate executive candidates in the AI era?
Ask candidates to show, not tell. Have them walk through a workflow they rebuilt with AI tools, what changed in speed or headcount, and what broke along the way. Then ask how they would run their function with a smaller team than their last company gave them.
Reference checks need an update too. Add two questions to your standard reference script:
- "What did this person change about how the team worked, not only what the team delivered?"
- "When a new tool or process failed, how did they handle it?"
The first question separates leaders who build systems from leaders who mostly add headcount. The second tells you whether they'll manage automation with the same care they bring to managing people.