How is AI used in each phase of an impact project?
Ideation, design, implementation, measurement: in each phase AI has tasks where it excels and boundaries it must not cross. The operational map of the Simbial method.
Phase 1 · Ideation: expand before choosing
The risk of ideation is falling in love with the first idea. Here AI is worth its weight in gold as a multiplier of alternatives: describe the client's problem and ask it for ten different directions, then the weak points of each, then what a competitor would do. The boundary: the final choice of direction — and the reasoning you will need to defend in front of the client — is yours. A good test: if you cannot explain the choice without mentioning AI, you have not yet made a choice.
Phase 2 · Design: stress-testing the plan
With the chosen path, the AI helps build and stress-test the plan: breaking the work down into activities, identifying overlooked dependencies and risks, running a dry run of the measurement plan (”with these indicators, what could make them unverifiable?”). The boundary: targets and indicators are chosen together with the client and the methodology — they are not generated. A target proposed by the AI and accepted without discussion belongs to no one.
Phase 3 · Execution: accelerating deliverables, not your thinking
In the field, the AI speeds up concrete outputs: document and communication drafts, starter code, meeting summaries, operational checklists. And it helps in difficult moments: rehearsing a sensitive conversation before having it, analysing an unexpected issue from three angles. The boundary: decisions under pressure and relationships with people. A difficult message written by the AI and sent without truly owning it shows — and it chips away at the trust you were building.
Phase 4 · Measurement: structure, never invent
In the moment of truth the AI has a precise and powerful role: turning the raw account of results into organised indicators — this is what our structuring tool does, suggesting without ever writing anything on its own. And a forbidden role: generating numbers. Every value comes from the field, with method and evidence; scores are calculated by the server using public coefficients. The final confirmation is always human, because the signature on the record is yours.
The common thread: transparency at every phase
The same rule applies across all four phases: the use of AI is declared and governed. Not for bureaucracy's sake — because a project you can explain exactly how it came to be is a project you can stand behind, and it is your ability to stand behind it that counts in the market. It is also the best way to learn: those who notice what they delegate also notice what they are learning.
Where to put it into practice
This map comes to life in the simulated project of your Impact Role: the same 4 phases, with Gea accompanying and challenging you. And when you feel ready, the final step of this journey is the test: the AI Ethics Licence.
Frequently asked questions
In which phase of a project is AI most useful?
In all of them, with different tasks: alternatives during ideation, stress-testing during design, accelerating deliverables during implementation, structuring data during measurement. With one constant boundary: choices and sign-off remain human.
Can AI generate impact data?
No, never: values come from the field through method and evidence, and scores are calculated by the server using public coefficients. AI can only help structure what has actually happened into well-organised indicators.
How is the use of AI declared in a project?
By making it accountable: knowing which parts the machine proposed and which ones you decided, and being able to explain this if asked. On Simbial, this transparency is part of the method.
← The full course · Next pill: The test: the AI Ethics Licence →