"Yes, and..." Product Development: Improv Lessons for Building Products at AI Speed
Now showing: Yes, and… Product Development
In the 1990s, before I transitioned to a career in tech, I studied, practiced, and taught theater. Part of that included becoming a trained improvisational actor and improv coach. While it may seem like improv and creating software have nothing in common, core improvisational principles like working collaboratively in fast-paced and ambiguous environments, staying calm under pressure, and the concept of "yes, and..." all helped prepare me for a career in designing and developing software products.
Today, as the industry transitions into faster-paced, AI-augmented teams, these improv principles are more critical than ever. When AI enables teams to ship in hours instead of weeks, product development starts to look less like an elaborate Broadway production and more like an improv show.
Product Development From Broadway to Improv
In scripted theater productions, everything is rehearsed—lines memorized, blocking perfected, sets and costumes designed and ready by opening night. Traditional Agile/Scrum teams operate similarly: comprehensive documentation, polished designs, technical specs, and carefully staged releases.
But when AI can generate working prototypes in less than a day, all that rehearsal becomes a liability. The bottleneck isn't development capacity anymore, it's how quickly the team can align, adapt, and learn from users in real time.
In product development, which matters more: perfectly executing last month's documentation & assumptions, or responding effectively to today's user feedback?
Like performers trained in improvisation, AI-accelerated teams don't need to follow detailed scripts. They rely on their training, built on trust, collaboration, and rapid response skills. Preparation shifts from generating stacks of deliverables to cultivating:
Deep user understanding
Rapid feedback loops
Strong prototyping and delivery habits
Psychological safety
The Cardinal Rule of Improv: "Yes, and..."
During an improvised scene, when your scene partner says "What a beautiful sunset," you don't respond with "What are you talking about? It's midnight, and we're in a submarine!" Instead, you might say, "Yes, and the way the sun hits the water reminds me of our first date."
You accept what's been established and build on it collaboratively.
Applying this principle to product teams: don't kill ideas prematurely. Explore them, prototype them, learn from them. "Yes, and..." fosters experimentation and turns resistance into momentum.
AI Speed Is Already Reshaping Teams
Shopify now gives PM candidates coding assessments—not because they’ll replace engineers, but because in an AI world, the most effective PMs can spin up prototypes themselves. That’s not a gimmick, it’s a recognition that product roles are shifting.
Shopify is hiring product managers who can use AI to generate code
This is another example of the evolution toward Product Architects: professionals who move fluidly between vision, strategy, and hands-on execution with AI as their partner.
The Rules of "Yes, and..." Product Development
1. Accept and Build, Don't Debate and Block
Old way: "I’m not convinced. We should spend a few more weeks analyzing the problem."
New way: "Yes, and we can prototype it this afternoon."
2. Start with Prompts, Then Tailor for AI
The foundation of any improv scene is a prompt—often provided by an audience member. In the context of "Yes, and..." Product Development prompts align the product team on what and why, e.g., "How might we reduce users’ frustration when they’re searching for products by price?"
Then these prompts are refined into user stories with clear instructions for AI: "As a user, I want to filter by price range and location with real-time updates."
3. Embrace Ongoing Feedback
Daily deploys, not quarterly releases
Consistent feedback loops
Begin each day by reviewing yesterday's user data
4. Clearly Define Roles
Small, high-trust ensembles with autonomy and clear roles move fastest.
Product Architects: product vision, feature prioritization, user feedback & data, prototyping
Dev Leads: architecture, security, integrations, code production
Fractional Specialists (content, visual design, deep research, marketing) consulted as needed
5. Fail Cheap, Learn Fast
AI speed significantly reduces the cost of delivery. Experiments and risk taking become fuel for learning, not cause for punishment because teams won’t burn weeks or months delivering something users didn’t need.
Move fast and learn things, not break things. Build for deliberate and targeted feedback loops and kill switches. Respect your users while designing your experiments.
A Week in "Yes, and..." Product Development
Monday – Plan Experiments
Review weekend data from live experiments
Verify any potential experiments connect to our product goals
Design this week's experiments based on feedback and data
Define success metrics and feedback collection methods
Tuesday-Thursday – Build & Deploy
Rapid prototyping with AI partners
Deploy experiments via feature flags and canary releases
Continuous user feedback collection
Quick iterations if early data suggests changes
Friday – Learn & Decide
Review all experiment results from the week
Kill/continue/pivot decisions based on data
Prioritize prompts for next week's experiments
Team retrospective: what’s working? what’s not? what surprised us?
Daily – Data & Direction (15 min)
Yesterday's user feedback highlights
Today's experiment status
Any blockers to getting user feedback quickly
The focus shifts from "what will we build?" to "what will we learn?"
The Essential Foundation: Psychological Safety
The best improvisers are generous performers & collaborators. They’re not scene stealers. They trust each other and don’t blame their scene partners if something doesn’t work on stage. Likewise, a Product Pair will only find success if they build a solid foundation of psychological safety.
Every improv scene is an experiment: Will the actors successfully collaborate to produce an entertaining scene? Will it tell a compelling story? Sometimes experiments don't work, but great troupes learn from their mistakes.
Every AI-accelerated product iteration is an experiment: How many different ways can we try to solve this user problem? Sometimes experiments fail, but great teams learn what not to do or how to do it better next time.
Psychological safety unlocks:
Willingness to be wrong without finger pointing
Fast pivots when experiments don't work
Candid feedback about what's working and what’s not
Learning velocity over feature velocity
It’s Agile at its Core… Accelerated
Agile's principles still apply—just much faster:
"Early, continuous delivery" = daily experiments with continuous user feedback
"Deliver frequently" = ship daily, not bi-weekly
"Respond to change" = the natural output of rapid experimentation
When cycles compress from weeks to hours, agility isn't optional, it's survival.
The Mindset Shift
Think you’re ready to drop the script and fully trust your partners? The biggest shift to “Yes, and…” Product Development isn’t technological—it’s cultural.
From finger-pointing → fearless iteration
From exhaustive planning → rapid learning
From individual productivity → ensemble creativity
AI speed means everyone can build fast. The teams that win are the ones that learn fast—the ones who turn every experiment into momentum and every failure into insight.
Teams that embrace “Yes, and…” won’t just ship more, they’ll discover more, adapt faster, and consistently build what users truly value. That’s the magic of improv. That’s the magic of great product teams.
