# How to run a day one Design Sprint workshop and why AI facilitation helps _A practical 68-minute design sprint workshop for framing a challenge, generating several responses and choosing one direction to explore._ Teams often start solution work before they agree what problem they are solving. A customer complaint becomes a feature request, a founder’s hunch becomes a roadmap item, or the first plausible idea becomes the plan because everybody is relieved to have something concrete to discuss. A focused design sprint creates a better sequence. Define the challenge, turn it into questions, make room for several answers, then choose a direction against a shared criterion. This guide explains how to run that sequence with 2 to 8 people in 68 minutes. ## What this Day 1 workshop covers The original Design Sprint is a longer, multi-day process with more research, mapping, expert input, prototyping and testing. This 68-minute workshop does not pretend to replace all of that work. It is a focused first session for a team that needs to move from a broad challenge to a selected direction for further discovery. The session combines four activities: Problem Statement, How Might We, Crazy Eights and Blind Vote. Each output becomes the input to the next activity. That connection matters more than any individual template. Without it, a sprint can feel like four energetic exercises that happen to share a calendar invitation. ## Before the design sprint Choose a challenge at the right altitude. “Improve our product” is too broad. “Increase completion of setup for new team leads” gives the group a user, a moment and an outcome to examine. Avoid writing the preferred solution into the challenge. “Design an onboarding checklist” has already skipped the work the sprint is meant to do. Invite people who hold different parts of the evidence: somebody close to users, somebody who understands delivery constraints, and somebody able to make or sponsor the next decision. Send the challenge in advance and ask participants to bring relevant facts. A sprint is a poor place to discover that the only person who has spoken to a customer is not in the room. - Name the user or group affected by the challenge. - Bring observations, research or operating evidence. - State any genuine constraints before ideas are generated. - Agree what decision this workshop can influence. - Keep the group between 2 and 8 people. ## 1. Write a useful Problem Statement Allow 25 minutes. Ask who experiences the problem, what they are trying to do, what gets in the way, what effect that has, and what evidence makes the problem worth attention now. Separate direct observations from assumptions. Both can be useful, but the team should know which is which. > New team leads cannot tell which setup steps matter for their role, so they abandon setup or ask an administrator to complete it. Support conversations and incomplete workspace data suggest this is happening repeatedly. > — Example problem statement A useful statement is specific without claiming more certainty than the evidence supports. If participants disagree, capture the uncertainty rather than polishing it away. “We have not confirmed whether the issue is relevance or language” is valuable input for later research. ## 2. Turn problems into How Might We questions Use the next 25 minutes to reframe important parts of the problem as opportunities. A good How Might We question is open enough to permit several responses but narrow enough to guide thought. “How might we fix onboarding?” is vague. “How might we help a team lead recognise the setup steps that matter for their role?” holds onto the user and the obstacle. Have people write independently before sharing. This reduces anchoring on the first suggestion and gives quieter participants time to form a view. Group questions that are genuinely similar, then choose one or two that the team wants to carry into idea generation. Do not spend ten minutes debating the perfect label for every cluster. - Does the question keep the relevant user visible? - Could several meaningfully different ideas answer it? - Does it avoid assuming a particular interface or feature? - Can the team act on what it learns? ## 3. Generate range with Crazy Eights Crazy Eights uses eight minutes for eight rough ideas. Fold a sheet of paper into eight panels or divide a digital canvas into eight spaces. Choose a How Might We question, start the timer, and sketch one response in each panel. Words, boxes and arrows are enough. This is thinking on paper, not a drawing competition. The pace discourages premature attachment. Once somebody has drawn the obvious dashboard, checklist or notification, they still have several panels to fill. Later ideas often change the unit of thought entirely: remove a step, change who acts, use an existing signal, or prevent the problem before setup begins. Keep evaluation out of these eight minutes. Questions about feasibility and risk belong in selection and follow-up. If critique begins during generation, participants quickly learn to produce safe variations of the idea that already sounds acceptable. ## 4. Choose a direction with a Blind Vote Give the final 10 minutes to review and choose. Let each participant present the intent of their sketches briefly, without pitching them like finished proposals. Clarify what an idea means, then state the voting criterion before votes are cast. A useful criterion might be: which direction best addresses the agreed problem and is most valuable to investigate next? Collect votes privately and reveal them together. Blind voting does not remove judgement or authority from the decision. It gives every participant an unanchored first choice and produces better information for the final conversation. If a designated decision owner can override the vote, say so before voting and record the reason if they do. Close by naming the decision state. The result might be selected, recommended or still unresolved. Record the chosen direction, the alternatives considered, the rationale, and the next piece of evidence needed. A design sprint should create a useful handoff to research, prototyping or planning rather than an unexplained cluster of dots. ## Why AI facilitation changes the economics This short session still needs active facilitation. Somebody must introduce the activities, protect silent work, keep ideas separate from evaluation, state the voting criterion, carry context forward and close with a clear decision. A canvas can display instructions, but it cannot notice that the group has skipped from a complaint to a favourite feature. ### Good process becomes available for routine work An experienced consultant can be an excellent investment for a consequential strategy sprint, a politically difficult decision or a room with serious conflict. Most teams cannot hire that person for every discovery question that appears during an ordinary month. The alternative is often a template and a hopeful volunteer. An AI facilitator makes a sound process affordable enough to use for everyday product work. Clarvelo runs the sequence, gives each phase its purpose, protects the timeboxes and carries the actual output into the next activity. The team gets process facilitation rather than a set of workshop instructions that somebody still has to operate. ### The organiser gets their seat back Internal facilitation has a hidden cost. The product manager, designer or conscientious engineer who volunteers spends the session switching between contributor, chair, timekeeper, note-taker and technical support. While others sketch, they check the clock. While others vote, they check whether the next activity is ready. That person is often the one who knows most about the challenge. When Clarvelo facilitates, they can write a problem statement, challenge an assumption, sketch eight ideas and vote on the same terms as everybody else. The team does not lose one of its most diligent members to workshop administration. ### Context survives the transitions The value is not simply automatic timing. Clarvelo keeps the problem statement available while opportunity questions are written, carries the selected questions into Crazy Eights, and presents the voting criterion with the resulting ideas. This continuity reduces the familiar drift where a team starts with a user problem and ends by voting for the most polished sketch. The team still owns the substance. AI should not invent evidence, conceal disagreement or declare a weak assumption to be a fact. High-stakes organizational choices, damaged trust and serious conflict can require a skilled human facilitator with the judgement to intervene. A routine discovery workshop still deserves a deliberate process, and every team member should be able to take part in it. ## A 68-minute Design Sprint: Day 1 agenda - 25 minutes: write and agree the Problem Statement - 25 minutes: create and select How Might We questions - 8 minutes: generate eight responses with Crazy Eights - 10 minutes: clarify ideas, vote privately and record the direction Treat the timings as constraints, not targets for filling airtime. A smaller group may finish the problem frame early and use the saved minutes to clarify assumptions. Protect the generation and selection stages. Cutting idea generation to rescue a long framing debate simply returns the team to its first solution. ## Run the sprint without running the machinery Clarvelo’s [Design Sprint: Day 1](/workshops/design-sprint-day-1) guides a team through Problem Statement, How Might We, Crazy Eights and Blind Vote in one connected session. Guests join from a link, while the AI facilitator introduces each activity, keeps context visible and records the selected direction. **Have a challenge worth exploring?** [Start a free Design Sprint: Day 1](/workshops/design-sprint-day-1). Bring the people and the evidence; Clarvelo will run the workshop process. --- Written by Joe Corcoran for [Clarvelo](https://clarvelo.com). Clarvelo helps teams run structured retros, decision jams, post-mortems, and other key decision-making workshops with an AI facilitator. Assigned outcomes are automatically exported to wherever your team works. [Read more on the Clarvelo blog](https://clarvelo.com/blog) or [try the live AI-facilitated workshops](https://clarvelo.com).