Product manager. Prev: @stripe, @Airbnb @Shopify @Verizon. Father, Husband, Christian. I built a few things. All opinions are my own, especially the bad ones.
On Product Judgment
As AI coding tools have gotten more capable and more expected in the product development process, there’s been a lot of conversation about what humans still bring to the table. Two questions keep coming up: how do the three core functions of product, design, and engineering operate in this new world, and where is human involvement still actually needed?
Most of the time, that conversation lands on taste. Humans have taste. AI doesn’t.
I think taste gets brought up so often because it’s vague enough to avoid scrutiny. It can mean entirely different things to different people. And when I push on what people usually mean by it, the answer is roughly: the ability to look at a user’s problem and land on the right solution more often than not.
That’s not a bad description. But it obscures more than it explains. It makes product quality feel intrinsic. You either have taste or you don’t. It treats judgment as something you’re born with rather than something you build.
What people are actually pointing at when they say taste is product judgment. And product judgment is a lot more specific.
Product judgment means having a repeatable process to understand a user’s problems and root causes so deeply that you can reliably identify solutions that help them reach their desired outcome. It means being able to validate those solutions through experimentation, collaboration, and yes, intuition built from experience. And it means doing that well enough, often enough, that you start to recognize patterns.
Those patterns matter. They’re what let you distill a messy set of user problems into the key insight about what’s actually causing them. They’re what let you connect solving that root cause to a clear user outcome. And they’re what let you describe a solution at enough resolution that design and engineering can actually work from it.
That last part is worth dwelling on. A product decision doesn’t just have to be right. It has to be communicable. The problem needs to be defined clearly, not just to you but to your collaborators. The solution needs to be articulated in a way the user can understand. The feature needs to be described at a level of detail where design can work through the interaction model and engineering can assess what exists today versus what needs to exist tomorrow, including the edge cases that make delivery reliable over time.
All of that design and engineering work is downstream of product judgment. Understanding what the problem is. Understanding what the user is trying to accomplish and what’s getting in their way.
Understanding what a good solution looks like, and how it fits within the broader product experience the business already delivers.
None of that is taste. Taste implies you just perceive the right answer. Product judgment means you’ve done the work to earn it.
That work is at bats. Repeat exposure to user problems, enough times and across enough different contexts, that you build real pattern recognition. The product builders with strong judgment aren’t just perceptive. They’ve been wrong enough times to understand why, and right enough times to know what that process looks like when it’s working.
Taste obscures the need for at bats. Product judgment requires them.
The reason this distinction matters now is that AI tools are genuinely changing what the job looks like. They can write code. They can generate designs. They can prototype ideas quickly. The parts of product development that used to require a large team and a long timeline can now move much faster.
But none of that changes what judgment is, or who has it. The AI doesn’t know which problem to solve. It doesn’t know whether the solution fits the user’s mental model. It doesn’t know whether the feature coheres with the rest of the product, or whether the business can feasibly deliver it on a timeline that makes sense.
Those are human questions. And the ability to answer them well, consistently, is product judgment means.
I built Calvin to make life easier for busy parents, but I think Calvin is genuinely useful for anyone.
The scheduling and email reply features are 💯
Sign up for access: https://t.co/btemb0UP22
@yrechtman System of context is actually what these companies actually mean, but to your point, SaaS = # of actions/mo per $$.
Starting from what outcomes does my user need and then what context do I need to have to achieve those outcomes is the better positioning.
As I wrote this, I saw X go into meltdown over tokens.
You've seen the headlines: “Uber blows yearly AI budget in just one quarter.” “Meta employee burns 281 billion tokens in April.”
But, the problem isn't spending. Spending works. Since 2023, the top quartile of our AI spenders doubled their revenue. The bottom quartile? Flat.
It's blind spending. We don’t know which spend worked.
A sales team has qualified leads. A support team has resolved conversations. These are units you can measure against. All a token tells you is the meter ran, not whether the work was worth it or not.
Finance says, “half the budget,” engineering says, “double it” and you don’t know who’s right because there is no shared language of value. There’s no attribution, and no attribution means no allocation.
For example, right now, all work, no matter the size or shape, defaults to frontier models. But meeting summaries and calendar updates don’t require GPT-5.5 Pro.
In isolation this seems trivial, but re-route just 10% of a $10M AI bill from frontier to GPT-4 level intelligence you’ve saved nearly one million dollars. This sounds like a made-up stat — it’s not. It truly is that much cheaper.
This is the future of finance: not blindly rubber-stamping or rejecting AI spend, but allocating it with the same rigor companies apply to headcount.
No more missed appointments. No more scheduling conflicts. Parents, your week just got easier.
Sign up for access to Calvin here: https://t.co/btemb0UP22
New and improved Calvin, including Agent Actions such as calendar management and email replies in you voice and tone.
Understanding your family context, Calvin proactively suggests and takes personalized actions on your behalf. You just review and confirm the action.
This is the way! Love the focus on an AI assistant that understands your context then works on your behalf.
This is the reason why I’m building CalvinAI, an AI assistant for families that understands their context and works on their behalf https://t.co/JI7cuNvXhB cc @jgreze
Today, we’re launching @TownAI: the AI assistant that learns you.
We’re coming out of beta with a $55M Series A led by @ARampell at @a16z, with participation from @KirstenGreen at @forerunnervc and continued support from @firstround, @altcap, and @conviction.
Right now, getting real value from AI means prompting, configuring, building workflows, managing agents.
We think that’s backwards.
The future of AI is a companion that already knows you and how you work. Town connects across your inbox, calendar, Slack, docs, messages, and workflows to understand what you need, then starts doing the work with you.
Drafting. Scheduling. Project tracking. Follow-ups. Context gathering. Multi-step tasks. And it only acts when you say so.
All adapting to your voice, priorities, routines, and relationships over time.
Your Townie is the AI assistant you actually need.
One person maybe doing all three roles, but outcomes of each role have not changed. AI tools just lower the foundational understanding needed to go the outcome. Still doesn’t mean it’s the right outcome.
"Engineering, product, and design are all merging into a 'builder' role"
Yeah... I'm not so sure. This feels like an oversimplification and podcast talking point. Reality is a lot more complex.
Even with 1000 "Member of Technical Staff" titles, someone still has to wake up and care 100x more about Product or Design than anyone else. It is their Main Thing™
That's not to say MTS titles are universally bad, but I think they're an example of this 'builder' talking point that's become bastardized.
AI and coding agents have made generating code easy and yet... you're in for a world of pain if non-engineers ship a bunch of slop and don't have great engineers to tame the complexity.
The SF hivemind has a tendency to overfit what works at startups for every company. And to be fair, sometimes this is true! Startups can be a leading indicator for how the industry is changing and often cause disruption.
However, it is going to be incredibly hard to disrupt the extremely human parts of corporate jobs. You really think there's going to be a PM who also does some engineering and design on the side at JPMorgan Chase?
This is true for the simple parts of most jobs, like people wanting to have ownership over something and do good work, move up a career ladder, support their family, get paid well, make an honest living...
And also the hard parts: internal politics, some critical business system that has a bus factor of 1 which has been running for 15 years and isn't documented anywhere because it's that guy's job security. The real world has a lot of this stuff.
It's easy to pontificate about all roles collapsing but it's actually really nice to have a specific person or team who is an expert in one thing that you can work with. I don't expect that to change. Further, I think AI disruption to knowledge work will take decades to play out because it is more fundamental to the human condition (e.g. sociological/organizational) than pure intelligence.
On-call and reactive? Present and proactive? I think it’s the latter and its what I’m trying to validate with Calvin.
Calvin is a proactive AI assistant for busy parents. If you’re a parent, I’d love for you to try it and provide honest feedback https://t.co/btemb0UP22
I built a product, Calvin, which is admittedly an very early solution and not gonna be final final product, but I think it gets at something fundamental about what comes next for humans and AI. Especially, non-business AI.
What role do we want AI to play in our lives?
Warm take: There really won’t be a “winner take all” dynamic in consumer AI, but the biggest opportunity is the being the context wallet used by all apps including the AI labs.
@DKThomp I think we need to separate demand for compute and demand for product and services built using that compute.
The Uber and Microsoft stories are management not seeing measurable business impact (aka increased customer demand) from their increased token spend (compute demand)
That is a real time savings for the prompter, but that average-level of writing means repeated patterns (for consistency) and writing that never hits for power (to use a baseball metaphor)
This is because the act of writing consistently well for your intended audience is incredibly hard.
When I wrote 2K-3K word @Coin_Labs articles every week for a year, I realized writing well was at least a part-time.
AI can write consistently average work product and…
People are realizing that promising 100x productivity gains don’t matter if there isn’t a market for what you are building.
It’s now more important than ever to know what to build and how to communicate it to customer effectively.
These are skills that AI is just not great at