Team UP

Every AI initiative that fails late fails here first. Team UP is where TERRAIN puts the two hardest questions before any code is written: who is building this, and why does the business need it?

What happens in this phase

• Assemble a cross-functional team. AI delivery needs data science, engineering, domain knowledge, and project management at the same table from day one — not data scientists working in isolation with a hand-off at the end.
• Define the vision and goals. What problem is AI solving? What value does it deliver, and to whom? Goals are written down, measurable, and linked to business objectives — this becomes the anchor for every later trade-off.
• Engage stakeholders early. Domain experts, business users, and end-users are in the room before development starts, so objectives are aligned and concerns surface while they’re still cheap to address.
• Anticipate the terrain ahead. The team names its risks — data availability, ethics, integration — and plans early experiments to test the scariest assumptions first.

Watch out for

• A “team” that’s actually one discipline. If the room is all data scientists or all IT, the phase isn’t done.
• Vision statements that can’t be measured. “Use AI to improve customer experience” is a wish; a goal has a number and a deadline.
• Skipping stakeholders because “we’ll demo when it’s ready.” That demo is where misalignment becomes expensive.

Method spotlight

Team UP pairs with Design Thinking: empathize with users, define the problem, ideate, and develop the business case before committing serious budget. See the method diagram on the framework page.

Go deeper

Chapters 7–8 of the book (“Assembling the Avengers”, “Breaking Down Silos”) cover team assembly and cross-functional collaboration in depth — free on this site from September 2026, or in the full edition on Amazon today.