Home / Responsible AI
How the Stradem team uses AI
AI reads across large amounts of customer data, runs more analysis than a team could by hand and drafts the briefs and results summaries. The Stradem team checks each output against your business before it informs a decision. Calculations use an agreed method, and your team approves each decision.
Where AI works
What AI does and what people do
Statistics run as repeatable models that give the same answer on the same data. AI reads across the data, drafts and explains. The Stradem team reviews each output before it informs a decision, and the decision log keeps the review.
| Work | What AI does | What people do |
|---|---|---|
| Finding segments and audiences | Reads across the agreed aggregate data for patterns and explains what sets each customer group apart. | The Stradem team tests whether each group makes business sense. Your marketing lead picks the audiences. |
| Studying journeys | Summarizes the paths customers take and drafts test ideas for each step. | The Stradem team judges which ideas your teams can deliver. Your marketing and platform leads choose which tests to run. |
| Test plans | Drafts the plan from rules set in advance. Drafts stay marked as drafts until approved. | The Stradem team designs the measurement with your analyst and sponsor. Together they agree the metric, the comparison and the result that triggers action before launch. |
| Measurement | The measurement engine runs the calculation. | The Stradem team reviews the method and the assumptions with your analyst. |
| Results summaries | Drafts the narrative around the calculated result. | A Stradem lead and your team review it before it is shared. |
| Scale, refine, stop or hold | The calculation shows where the result sits against the target. | Your sponsor decides. |
Explanations tied to evidence
AI explains the numbers the models produce
Each explanation draws on the model results for the decision. It names the assumption that drives the pick, shows what holds an option back from starting and says whether a result supports scaling. A reviewer checks every figure against the calculation before the explanation is shared.
Example explanation
Why does the second-product offer start before the onboarding fix?
The onboarding fix has the highest forecast value, but the product team has no capacity this quarter, and no delivery owner is assigned. The second-product offer returns the margin the team needs and can start now.
Before starting the onboarding fix, the team needs to confirm capacity, name a delivery owner and update the cost and timing.
Drawn from the model results and the evidence listed for each option.
Data handling and traceability
Aggregate data, protected and traceable
Customer-level records stay in your systems under your data team. Stradem works with agreed aggregate data on segments, costs and outcomes.
- Stradem works with aggregate data only, so customer datasets hold no customer-level personal data
- The workspace holds account information for the people who use it, such as names, work email addresses and access records
- Cloud services and third-party language model providers process data under contract
- Customer data stays out of training for shared or third-party models
- The decision log keeps the data used, the version of the analysis, who approved it and when
- The current list of providers is available on request
Questions about AI at Stradem
Why does Stradem use AI?
AI lets Stradem compare far more segments, journeys and channels than people could by hand and draft the briefs and results summaries. The Stradem team checks the work against your business, and your sponsor makes each decision.
Does AI make the investment decision?
Your sponsor makes the decision. AI and the Stradem team inform it with explanations, drafts and a recommendation, and your sponsor decides whether to scale, refine, stop or hold.
Is our data used to train models?
Customer data stays out of training for shared or third-party models. Your agreement states the terms.
What personal data does Stradem hold?
Stradem works with aggregate data only: counts, rates and totals by group and period. Customer datasets hold no customer-level personal data. The workspace holds account information for its users, such as names, work email addresses and access records.
How is AI output checked?
Numbers come from calculation, and a reviewer checks every narrative against the calculation before it is shared. Assumptions sit next to each result, and your team can challenge any output. The challenge goes into the record.