Introduction: The AI Pressure Cooker

Across industries, teams are feeling the same pressure: “Do something with AI—and do it fast.”

This vague top-down directive has become the default leadership mandate. It rolls down from boardrooms to product teams, data teams, operations, HR, and customer experience. Everyone is expected to have an “AI strategy,” even when no one can articulate the actual problem worth solving.

And in 2025, leadership expectations became even bolder.

Executives no longer want experiments. They want transformation—scaled, responsible, agentic AI embedded into core workflows.

They want efficiency, reinvention, and competitive differentiation.

They want measurable impact, not another demo.

But that mandate creates stress.

Teams ask:

• ‍Where do we begin?
• Which use cases matter?
• How do we avoid wasting time, talent, and credibility?

The natural impulse is to start building immediately. But this rush produces a sea of sameness—faster email writers, prettier slide generators, slightly improved workflow tools that don’t fundamentally change anything.

The real risk today isn’t moving too slowly.

It’s moving fast in the wrong direction.

AI Problem Framing gives teams a way to stop, think, and align—before they build. It creates the focus that speed actually requires.

This guide breaks down the process into five actionable phases, giving facilitators a step-by-step method to turn vague AI pressure into clear, fundable opportunities.

In this guide, you’ll learn:

• What AI Problem Framing is
• Why teams need it in 2025
• When to use (and not use) it
• Who should be involved in the workshop
• The 5-phase facilitation process
• What happens after the workshop

‍

What Is AI Problem Framing?

AI Problem Framing is a practical, repeatable method for cross-functional AI teams to get aligned before they start building.

It evolved from Design Sprint Academy’s original Problem Framing method, launched at Google in 2018 and later adopted by teams at SAP, RGA, HSBC, and many others. The goal was always the same: Help teams stop chasing ideas and start prioritizing real problems worth solving.

In 2025, we redesigned the method to fit today’s AI reality.

Over the past year, we worked closely with Turner Construction Company —one of the largest construction organizations in the U.S.—helping teams refine how they find AI use cases and validate them through design sprints. We took those lessons and built a new formula specifically for the gap between what senior executives expect and what teams on the ground can deliver.

The result is an intensive, one-day workshop that forces collaboration. It brings together people with different levels of AI literacy, different functions, different priorities, and different assumptions—and aligns them quickly.

Ultimately, AI Problem Framing shifts the question from:

“What can AI do?”

to

“What should AI do to create real value?”

‍

Why AI Problem Framing Matters

AI today is where electricity was in its early days.

When electricity first arrived, people used it to replicate old tools: candles → electric bulbs, charcoal irons → electric irons. Useful, but unimaginative.

Much of today’s AI effort is still stuck in that replication stage.

We use AI to write emails faster, create slides faster, summarize documents faster.

Helpful—but not transformative.

True differentiation comes from reinvention:

• new workflows
• new service models
• new products
• new experiences

AI Problem Framing is the mechanism that helps teams make this leap.

It connects:

customer insight → business priorities → what’s newly possible with AI.

It prevents teams from wasting time, money, and talent building “the wrong thing—beautifully.”

‍

When to Use AI Problem Framing

✅ Use AI Problem Framing when:

• You’ve been given a vague AI mandate (“use AI somewhere in the business”)
• You have too many competing AI ideas
• You’re choosing which AI solution to fund first
• You need alignment across product, data, business, and operations
• You want to avoid spinning up pilots that never scale
• You want to frame AI opportunities before a design sprint or build cycle

❌ Do NOT use it when:

• The problem is already validated and well-defined
• You’re optimizing an existing model, workflow, or feature
• You’re doing a narrow technical proof-of-concept
• You only need UX, UI, or content improvements

‍

Who Should Be in the Room

An AI Problem Framing workshop only works when the right people are in the room.

This method isn’t powered by brainstorming or creativity alone — it’s powered by expertise. Everything the team needs to make decisions must come from them: data, constraints, customer insights, operational realities, research, technical limitations, and even existing AI capabilities.

A facilitator can guide the process, but the quality of the workshop depends entirely on the quality of the people in the room. Without strong, experienced experts, you won’t get strong, actionable outcomes.

Here’s who you need:

1. DECIDER

Product Manager, VP, Director

Owns outcomes and sets direction.

Accountable for decisions, priorities, and what moves into the roadmap.

2. CORE TEAM (Cross-functional, 6–8 people)

Subject-matter experts and doers who bring the facts — not opinions.

They hold the knowledge that makes the workshop work:

• customer behavior and insights
• operational workflows
• data availability and limitations
• compliance and risk constraints
• market and business context
• technical feasibility and AI capabilities
  

These people are essential. Without their input, the team can’t align on a real opportunity.

3. FACILITATOR

The facilitator guides the group through the process.

They:

• prepare the room and agenda
• explain activities clearly
• enforce timeboxes
• manage energy and participation
• document outputs

The facilitator’s job is simple: Guide the process so the team can focus on the decisions.

Who usually facilitates?

This role is often taken on by:

• Agile coaches
• Designers or service designers
• Innovation consultants or managers
• Scrum masters or team leads
• Design Thinking coaches

But one important truth remains:

The facilitator does not provide the answers — the team does.

This is what turns the workshop from a creative session into a strategic decision-making engine.

‍

The 5 Phases of AI Problem Framing: Your Facilitation Guide

The following five phases provide a step-by-step playbook for facilitating an AI Problem Framing workshop. This process isn't just a series of creative activities; it's a structured system for making critical choices. Each phase builds on the last and concludes with a decision gate—a clear point of alignment that builds momentum and ensures the team is ready to advance.

‍

1. Phase One: Ideas - Make the Mandate Visible

This first phase reframes the vague, top-down mandate from a blocker into the starting line. The objective is to take that mandate and the scattered ideas floating around the team and make them tangible and visible for everyone.

The core activity is simple: The team captures all known AI ideas, leadership requests, backlogs and existing initiatives on sticky notes. These notes are then shared aloud and clustered into themes on a wall or flipchart. This process creates a shared baseline of "what's on the table," turning a messy directive into raw material the team can work with.

• Facilitator's Note: Your role here is not to judge the ideas but to ensure every voice is heard. The goal is creating a shared visual baseline, turning abstract pressure into tangible raw material.

‍

2. Phase Two: Business - Connect Ideas to Business Goals

AI initiatives fail when they don’t tie back to what the business cares about: growth, efficiency, or risk reduction. The goal of this phase is to filter every idea through the lens of strategic business goals.

First, the team reviews and agrees upon the organization's key business goals, often sourced from OKRs or strategy decks. Next, they map the AI ideas from Phase One to the specific goals or business blockers they could help address. Any idea that doesn't clearly connect to a defined business goal is set aside. The phase ends once the team hits its first decision gate: the strategic business goals are selected, and any off-strategy ideas are parked.

• Facilitator's Note: This is the first major filter. The act of mapping ideas to goals forces a crucial conversation about priorities. If an idea doesn't connect, it doesn't mean it's bad—just that it's not a priority right now.

‍

3. Phase Three: Customers - Find the Human Problem

This phase shifts the perspective from the business to the people who actually experience the problems. The goal is to define precisely who the solution is for and which of their problems are most important to solve. The team works through a clear sequence: first, they identify their MVS, then build a Proto-Persona to bring that segment to life, and finally use the P.A.L.T. Framework to prioritize that persona's most critical problems.

• Minimum Viable Segment (MVS): The smallest, clearly defined group of customers who share a common problem and are most likely to adopt your solution first. The key is to be both Minimum (small enough to dominate) and Viable (big enough to be sustainable).
• Proto-Persona: A quick, assumption-based persona that captures the team's shared understanding of the MVS. It outlines the segment's key facts, goals, behaviors, and challenges, bringing the target user to life.
• P.A.L.T. Framework: Like salt, P.A.L.T. helps surface the real flavor of what matters to users. It's a 2x2 matrix that prioritizes problems based on whether they are Painful vs. Aspirational and Latent vs. Top of Mind.

At the end of this phase, the team has hit its second decision gate: the target customer segment and their prioritized problems are defined and agreed upon.

• Facilitator's Note: Encourage the team to rely on their collective knowledge here. A Proto-Persona isn't about deep research; it's about making assumptions visible so they can be tested later. The goal is alignment, not perfection.

‍

4. Phase Four: Context - Map the Experience

Context is what turns technology from an interesting demo into a life-changing solution. The goal of this phase is to find where an AI solution can integrate into a customer's life to solve a problem at the moment it occurs by mapping their experience.

The team builds a Customer Journey Map (CJM), a step-by-step visualization of the proto-persona's process to accomplish a goal. The map highlights the stages, actions, challenges, and desired outcomes of their journey, revealing the precise moments where AI could remove friction or create new value. Using insights from the CJM, the team generates grounded "What if...?" questions that reframe challenges into opportunities. Examples include:

• What if AI could predict when the child will run out of money and send a warning?
• What if parents could set smart boundaries (e.g., no candy shop over €10)?

This phase concludes with the third decision gate: the customer's journey is mapped and the desired outcomes are clear.

• Facilitator's Note: The Customer Journey Map is a storytelling tool. As the team builds it, encourage them to narrate the experience from the persona's point of view. This builds empathy and helps uncover non-obvious pain points.

If your MVS is an internal customer—a specific group of employees—you won’t map a customer journey. You’ll switch to a service blueprint or another system-level visualization. This lets the team see how the workflow currently operates, where the bottlenecks are, and which steps actually matter. And here’s the key: some steps don’t need AI at all. Some need to be redesigned, simplified, or removed entirely. The team can only make those calls when they see the whole system in context.

5. Phase Five: AI Use Cases - Synthesize and Prioritize

In this final phase, everything comes together. The goal is to evaluate the most promising ideas for desirability, viability, and feasibility, and then synthesize them into a clear statement. The team uses two prioritization methods:

• Painkiller vs. Vitamin: Each participant gets two dots—one red (Painkiller) and one green (Vitamin)—and places them on the "What if...?" ideas they believe fit each category. This sparks a conversation about which ideas solve an urgent, painful need versus those that are just nice-to-have features.
• Magic Lenses: The top "Painkiller" ideas are stress-tested through multiple business perspectives to find the "sweet spot" where customer value and business viability overlap. The magic lenses (a tool borrowed from the Foundation Sprint by Jake Knapp and John Zeratsky) include:
  • Growth Lens: Which option has the best chance of scaling and attracting users?
  • Money Lens: Which option promises the highest return on investment?
  • Pragmatic Lens: Which option is easiest, cheapest, and fastest to build?
  • Data Lens: Which option has data that’s easiest to get and AI-ready to use?

The final decision gate is reached as the team turns their top 1-2 opportunities into a single story that leaders can fund by filling out AI Use-Case Cards. This card connects the customer segment, their problem, the AI approach, the desired outcome, and the business goal.

This artifact becomes the foundation for a more detailed AI Use Case Brief, providing a clear bridge from the workshop to real-world action.

• Facilitator's Note: The Magic Lenses activity often reveals hidden trade-offs. An idea that's a huge "Painkiller" might be pragmatically impossible. Your role is to guide the discussion so the team can identify the ideas that are strong across multiple dimensions.

What Happens After AI Problem Framing

The workshop ends with clarity—but the work continues.

Teams now have:

• a defined customer segment
• a prioritized customer problem
• a feasible, high-value AI use case
• strong business alignment
• and buy-in from the Decider

The next step is usually to run an AI Design Sprint, where the team:

• prototypes the concept
• validates it with real users
• and runs a technical feasibility check with engineers and data experts in the room

By the end of the sprint, the team knows:

• Will users adopt it?
• Is it strategically worth it?
• Is it technically possible with our data and systems?

Once desirability, viability, and feasibility are all validated, the team can translate the insights into a Product Requirements Document (PRD) and move the opportunity into the build roadmap.

Conclusion: Build What Matters

AI Problem Framing gives teams the structure they need to move from a chaotic list of ideas to a focused, prioritized, and fundable AI opportunity. It doesn’t slow you down. It protects you from moving fast in the wrong direction—which, in this AI era, is the real risk. Most teams don’t fall behind because they’re too slow. They fall behind because they waste time, money, and talent building the wrong thing—beautifully.

The goal is simple: de-risk innovation by shifting the team’s focus from speed for the sake of speed to building what actually matters.

‍

Resources:

‍

👉 Join our free webinar: The Comeback Framework Every AI Team Will Use in 2026 (Nov. 25)

‍

‍

‍

‍

‍

‍

‍

👉 Enroll to Build Smarter with AI – VIP Christmas Edition (Berlin, Dec 10-12, 2025) — A high-impact, in-person 3-day workshop where you’ll master both the AI Problem Framing method and the full Design Sprint process. With only 8 seats, direct coaching from experts (Dana Vețan & John Vețan), facilitator toolkits and playbooks included — suited for product leaders, innovation strategists, and transformation professionals.