I’ve just been watching a talk from Tommy Keeley a growth PM at Amplitude, on integrating AI into product workflows. And look, whilst everyone’s talking about AI, many teams are still just using it for summarising docs and writing Slack messages.
The gap between what’s possible with AI today and what most startups are actually doing is widening. And that gap is a competitive disadvantage whether you want to acknowledge it or not.
Below are three things you can implement this week to close that gap. No additional headcount required.
Stop Letting AI Agree With Your Bad Ideas
AI is a chronic people-pleaser.
Keeley tested this by asking ChatGPT about a genuinely terrible idea (an alarm clock that wakes you every 5 minutes and never turns off), and it enthusiastically supported it rather than telling him it was garbage.
As he puts it: “By default AI is going to tell you what you want to hear, not what you should hear. And that’s really, really dangerous, especially as PMs, because as PMs, we need friction.”
So what do you do about it? Engineer disagreement into your prompts by creating what he calls a “board of dissenters”, specific personas that challenge your thinking from different angles.
You pick 2-3 of these perspectives depending on your decision:
- The Skeptic (questions assumptions and ROI)
- The User Champion (focuses relentlessly on user needs)
- The Executor (asks about implementation reality)
- The Visionary (pushes for bigger thinking)
- The Systems Thinker (examines second-order effects)
But make it specific. Assign AI an actual name and background. For example: “You are Don Norman, the design thinking expert. I’m deprecating features to simplify our product. Challenge my approach from a user psychology perspective.”
The key is selecting advisors who are as far from your natural tendencies as possible. You’re not looking for validation, you’re engineering conflict to make better decisions. One clever tactic Keeley mentions is pitting two advisors against each other (like Andy Grove vs. Elon Musk on a resourcing decision) and observing where the tension lies.
Learn to “Vibe Code” (Yes, Even If You’re Non-Technical)
Growth PMs and founders have always been scrappy.
AI just turned that scrappiness into a superpower, but only if you actually use it.
Keeley’s core argument is that building with AI isn’t really about replacing engineers or replacing PMs. It’s about removing friction between your ideas and your learning.
For startups, this matters because:
- You gain understanding of what’s easy vs. hard, which leads to better prioritisation
- You build empathy for your engineering team, which leads to faster decisions and better partnership
- You ship prototypes before PowerPoints, which means faster feedback loops
So how do you start if you’ve never coded?
Open ChatGPT and literally say: “I’ve never coded before. I don’t know where to get started. Give me step-by-step instructions.”
Then create a playground. Replit, Bolt, or Lovable work for prototyping. Cursor or Claude Code if you want to get into actual coding.
Start small. Build a toy app or website to learn the basics. Then graduate to your company’s codebase with small copy updates or design prototypes.
But be warned about the confidence curve. Expect your confidence to spike when you build your first toy apps, then crater when you enter your company’s real codebase. That’s actually healthy, it builds respect for the complexity your team deals with daily.
One more thing. Just because you can build something doesn’t mean you should. Keley’s reminder is important: “Speed does not equal success. We still need to be delivering value for our customers and our business.” Don’t fall in love with your castle and build something nobody needs.
Use AI to Find the Hidden Growth Levers
Most teams use AI for surface-level productivity, writing queries, summarising dashboards.
But the real opportunity is using AI to sharpen your judgement and identify non-obvious opportunities.
Four ways to do this:
Experiment analysis without bias. AI can dissect experiment logs, calculate lift, and most importantly, catch you when you’re cherry-picking metrics because you’re emotionally attached to a project you spent weeks on. Ask it: “What am I missing? Which segment responded best? Is this actually statistically significant?”
Non-obvious customer segmentation. Stop relying only on demographic personas. Ask AI to cluster users by behaviour instead. You might discover power users in an unexpected segment or identify long-tail opportunities you’d never consider otherwise.
Voice of customer at scale. Thousands of support tickets, reviews, and surveys? AI can extract themes, sentiment, and urgency far faster than any human. This lets you prioritise the top pain points without drowning in Zendesk.
Parallel “what-if” scenarios. AI can generate multiple scenarios simultaneously and force you to consider options you’d normally skip. This makes your decisions more well-rounded and less emotional.
The underlying principle is making the invisible visible. AI’s job isn’t just to work faster, it’s to help you see patterns and opportunities you’d otherwise miss entirely.
What This Actually Means
The software development cycle hasn’t changed. You still go from idea to execution to learning. What’s changed is the speed of that cycle and the tools at each phase.
As Keeley puts it: “Your job as a PM really is to close that gap between what’s possible today in your daily workflow and what’s actually been adopted.”
For startups, this isn’t optional. When you’re resource-constrained and competing against better-funded companies, these AI tools are the closest thing to a force multiplier you’re going to get.
This week, try this. Next time you’re pressure-testing an idea, create a board of 2-3 dissenters instead of asking ChatGPT directly. Spend 30 minutes learning to vibe code, start with “I’ve never coded before” and see where it takes you. Take your last experiment results and ask AI: “What patterns am I missing? What would challenge my conclusion?”
The teams that figure this out now will be moving 2-3x faster than their competitors by mid-2026. The teams that don’t will wonder why they’re getting left behind.
Watch the video for yourself here
An update from me:
I’ve been thinking about the direction of this newsletter and what the future might be for it. I could keep banging on about interviews and process and all that stuff. But deep down I don’t think it’s the thing I want to write about forever.
Ever since hiring for product teams at Zendesk and then MessageMedia and beyond, I’ve been heavily engaged in the Product space and I think it’s where a lot of my future work is going to be focused.
Which makes me think that the future of this newsletter is going to be heavily skewed in that direction. There’ll still be recruitingy topics to talk about but for the most part it’s going to be Product focussed, somewhat along the lines of this one.
Just thought you should know.
Before you go, can you help me out?
Nearly 1300 people have now subscribed to this newsletter in ~15 weeks, which makes me feel that it’s resonating. So thank you to all of you 🙏🏻
If you found this issue useful, here are three ways you could support me:
- Share it – Forward this to a founder, hiring manager, or job seeker who you think might find it useful and/or interesting.
- Follow StartUpHiring on LinkedIn if you’re not already – https://www.linkedin.com/company/startuphiringaustralia
- Send feedback – What topics would help you most? What’s working? What isn’t? Just hit reply.
I’m building this into the go-to resource for startup hiring. Whether you’re hiring your first employee or looking for your next role, I want this to be genuinely useful for you.
These small actions make a massive difference in helping me reach more people who need this content.
Thanks for being here. Have a brilliant week ahead.
Weekly newsletter drop: Wednesday 1:30pm
Need hiring help?
If you’re struggling with a specific hiring challenge, I still offer 1:1 strategy calls through my consultancy, Crew. Book a 15-minute clarity call here.
Simon
