Don’t Let AI Build Your Roadmap
JTBD Is the Missing Ingredient in Most AI Products
Everyone’s talking about AI. Everyone’s building with AI. And far too many are doing it backwards.
We’re seeing a wave of teams starting with what AI can do, then working backward to justify it. Features get built. Launch videos get cut. And customers…shrug.
Customer-first thinking is what builds lasting products.
The question is not, “What can we do with AI?” But rather, “What has been impossible for our users until now?"
And that requires laser precision on the user and their Job to be Done (JTBD).
Garmin: AI Misfire
I’m a recent convert from Apple Watch to the Garmin ecosystem, and I love what my Garmin watch has unlocked for me in terms of training (and of course, battery life).
Garmin has built a loyal following by being reliable, rugged, and deeply attuned to its users — especially serious athletes and outdoor adventurers.
They recently launched a subscription service, Connect+, that contains AI features.
And many long-time Garmin customers lost their minds at the concept of a subscription service (something very un-Garmin), and were full of cynical observations about the AI ingredients.
Some excepts from a few of the threads:
“Wow, their LLM can do math…”
“They used chatGPT to give you a math equation that you now have to pay for.”
“Am I missing something or is connect + just some useless ai insights.”
“AI is a solution looking for a problem. Every tech or tech-adjacent company is scrambling to get AI products out ASAP...”
“The so-called ‘Active Intelligence’ feature seems more like a gimmick than a valuable addition…”
A cautionary tale! Obviously, I have no insider info — I can certainly speculate on how and why this launch came to be though.
Garmin's stumble reminded me that even great product teams can lose the thread when tech leads the conversation. We nearly did the same as we built towards the Graph beta a few months ago.
Graph: Interrupting Our Own AI Momentum
While building Graph, we got caught in a similar current. AI was evolving fast. We saw all the things we could do — summarization, insights, predictive nudges. We spent months arm wrestling the LLM and building in possibilities.
But then we hit pause.
We went back to user interviews. Re-read transcripts. Asked ourselves:
What is our user trying to accomplish, really?
Where are they frustrated, not just functionally, but emotionally?
What moments make them feel like they’re winning?
That reset was critical.
It reminded us that AI should serve the solution — not be the solution.
Rediscovering JTBD in the AI Era
JTBD — A Refresher
Jobs to Be Done (JTBD) is the practice of understanding what customers are really trying to accomplish — not in your product, but in their lives. It brings clarity when technology muddies the waters.
The classic example is the fast food company’s curious early morning milkshake sales. My former colleagues at FullStory wrote a post from 2020 that’s just as relevant today five years later.
As Christensen put it:
When we buy a product, we essentially 'hire' it to help us do a job. If it does the job well, the next time we’re confronted with the same job, we tend to hire that product again. And if it does a crummy job, we 'fire' it and look for an alternative.
Don’t get fired. In times of disruption, JTBD becomes an even more critical compass! Stay anchored to why your customer has hired your product.
How to Leverage JTBD in the Age of AI
How might we leverage the tried and true craft of jobs to be done in the context of AI?
1. Micro-Moment Mapping
Break the experience into emotional moments:
“When I realize I’m behind on reporting…”
“When I’m unsure if I’ve missed a signal in the data…”
Use sticky notes or FigJam. Write them in first-person, emotionally charged language.
2. The Job Statement Template
“When I [situation], I want to [motivation], so I can [desired outcome].” Unbelievably powerful artifact to serve as a center of gravity for your team.
Example from Graph:
“When I’m reviewing last week’s sprint, I want to quickly see what changed and why, so I can make better calls in Monday’s meeting.”
AI doesn’t show up yet. We’re still in the human experience. And when that sentence is front and center, it makes it much easier to stay focused on the user rather than the technology.
3. Forces of Progress
Use the four forces to understand what's pushing your customer toward change, and what might be holding them back:
Push of the situation — What's frustrating or broken in their current workflow?
Pull of the new solution — What hope or promise does a better future offer them?
Anxiety of the new — What worries or risks make them hesitate to adopt something different?
Habit of the present — What routines or muscle memory keep them anchored in the current way?
Plot these forces across your key JTBD moments. Where is momentum blocked? Where is trust low? Where is friction highest?
Then: where might AI help?
Can it reduce the anxiety of trying something new?
Can it make switching feel familiar or low-risk?
Can it amplify the promise of a better outcome?
Find the high-leverage spots where AI can make the experience feel effortless to the user.
Then, when you have a deep understanding of the JTBD, AI becomes a tool for acceleration, not distraction.
The Pressure Test for Product Leaders
Before you ship that AI feature, ask yourself these six hard questions:
What job, exactly, does this solve — and what made it unsolvable before?
If you can't name the struggle it resolves, you're inventing use cases, not serving real ones.
How does this change the way users complete the job?
Is it 10x faster? More intuitive? Less stressful? Or just... more expensive?
Would we ship this without AI?
If not, why now? Are we chasing substance or spectacle?
How do we measure success in job terms, not feature metrics?
Job completion > AI engagement. Show outcomes, not usage.
Where does this create new risk, confusion, or false confidence?
AI can erode trust when it fails. What’s our mitigation plan?
Are we solving for customer delight or investor theater?
Great products win hearts. Demos win pitch meetings. Which are you building for?
Bottom Line: Stay Grounded and Lean Into AI
The best question teams can ask is, "What has been impossible for our users until now?" It puts the customer at the center, and forces focus on their JTBD.
Breakthroughs happen for teams who combine the deep discipline of user understanding and the intense power of AI. It’s only the beginning!