AI Landscape
AI Landscape is a canvas-based workshop tool for finding the places in an existing service where AI could be built in. The team lays out the service's phases, actors and front-stage and back-stage interactions, marks where the problems are, and then generates AI opportunities for those points, sorted by whether the AI would augment people, assist them or automate a task. The tool is published by the studio oblo on Service Design Tools.
Facilitation script
- 1
Frame the session: which service, which boundaries, and what the output will be used for. Explain the three kinds of AI role (augment, assist, automate) with one everyday example each.
10 min - 2
Map the current service on the canvas: phases as columns, then actors, front-stage and back-stage interactions for each phase.
30 min - 3
Add problems and improvement areas per phase on a separate note colour. Mark the three to five steps the group considers most problematic.
20 min - 4
Generate AI opportunities for the marked steps, one idea per note, written individually first and then shared. Sort each into augment, assist or automate.
30 min - 5
Review the opportunities. For each cluster, discuss impact, risk and effort, including the data needed and the consequence of wrong output. Select the most promising.
20 min - 6
Record the selected opportunities, assign an owner to each and list the questions to answer before the next step.
10 min
Tips
Do not shorten the current-state mapping to get to the AI ideas faster; opportunities are only as good as the group's shared picture of how the service works today.
Have at least one person in the room who knows what current AI systems can and cannot do reliably, and one who does the work being discussed.
When an idea lands in the automation row, ask who checks the output and who the customer talks to when it is wrong.
Common pitfalls
Starting from AI capabilities instead of from the service's problems, which produces impressive-sounding ideas attached to steps that were working fine
Mapping only the customer-facing steps: many of the workable opportunities sit in back-stage work, and they never appear on the canvas
Treating every opportunity as automation, so the group skips the cheaper and lower-risk options where AI supports a person who stays responsible
Selecting by enthusiasm without discussing data, error consequences and accountability, so the chosen ideas stall at the first feasibility review
Running the session without the people who do the work, which leads to opportunities that remove the wrong tasks and meet resistance later
Variations
Remote: build the canvas on a shared whiteboard, map the current state in one session and run the opportunity round in a second, so people can check facts in between. Short version (90 minutes): bring a finished journey map or service blueprint and start at the problem step. Larger groups: split by phase of the service, let each subgroup generate opportunities for its phase, then review across phases in plenary. A deck of AI capability cards, such as the companion card set from the same source, can prompt options when the group runs dry.
Where it fits
When to use it
Leadership has asked "where should we use AI?" and the team needs an answer grounded in how the service works, not a list of tools
Business, service and technical people each hold a different picture of the service and of what AI could do in it
Several isolated AI pilots exist and nobody has looked at how they fit along the whole service
You are at the front end of an AI strategy or discovery project and need a first opportunity backlog to assess
A known pain point keeps attracting the suggestion "let AI do it" and the group needs to compare that with other options
When not to use it
The service does not exist yet or nobody in the room knows how it runs: map it first with a Service Blueprint or interview the people who operate it
The decision to build a specific AI feature is already made: go to concept and prototype work instead of a broad opportunity scan
No one present can judge technical feasibility or data availability: the output will be a wish list, so invite those people or postpone
The real issue is that existing AI pilots are not being adopted: look at the blockers to scaling instead of generating more opportunities
Related methods
Frequently asked questions
What is AI Landscape?▾
AI Landscape is a workshop canvas for finding where AI could be integrated into an existing service. A mixed team maps the service's phases, actors and interactions, marks the problems, and generates AI opportunities for the most problematic steps. Each opportunity is classified as augmenting people, assisting them or automating a task, and the group then selects the most promising ones.
Who should take part in an AI Landscape session?▾
Aim for four to ten people who together cover three kinds of knowledge: how the service runs day to day, what the business needs from it, and what AI systems can do with the data available. Include at least one person who does the front-line or back-office work. Without that mix the session produces either unrealistic ideas or ideas nobody wants.
What is the difference between augmenting, assisting and automating?▾
The three terms describe how much of the task the AI takes over. In practice, augmenting means a person can do something better or something new with AI support, assisting means the AI handles part of a task while the person stays in charge, and automating means the task runs without a person doing it. Sorting ideas this way makes the group discuss responsibility and risk for each one.
How is AI Landscape different from a Service Blueprint?▾
A Service Blueprint documents how a service is delivered across front stage and back stage. AI Landscape uses a similar overview as its starting point and then adds an ideation and selection step focused on AI. If you already have a blueprint, you can bring it to the session and save most of the mapping time.
What happens after the session?▾
The output is a shortlist of opportunities, not a plan. Each one needs a feasibility check on data, cost and risk, and usually a small prototype or test before any commitment. Assign an owner per opportunity in the room so that the shortlist does not stay on the canvas.
Plan your next workshop with AI
Workshop Weaver helps you combine methods like AI Landscape into a complete, timed agenda in minutes.
Try it freeMethod descriptions on Workshop Weaver are original content written by our team, based on established facilitation practices. This method was inspired by work from oblo (Service Design Tools).