Using the 10-20-70 heuristic without treating it as a budget rule

Use the heuristic as a question

A 10-20-70 framing can draw attention to algorithms, technology and data, and people and process. Treat the ratio as a discussion prompt. It is a planning heuristic, not an allocation that has been established for every company. Do not infer your staffing, spend or return from those percentages.

Build a workload estimate from evidence

Map the current workflow, then estimate integration, source cleanup, evaluation, authorization, reviewer time, training and operational support. Ask each owner to identify the hidden work in their area. If data access dominates the critical path, assigning a fixed technology share will not remove the constraint.

• Record role changes and exception handling before introducing the tool.
• Budget time for fresh-task assessment and retraining after material releases.
• Include client review, maintenance and backup practice in full cost.

A practical review prompt

At each gate, ask: what must a person do differently, what system must permit it, and what evidence would show the change helps? A people-heavy narrative can still overlook broken authorization or missing data. Treat the operating system as a connected set of responsibilities and tests.

Sources and limits

These are Otomat’s proposed operating methods and original decision artifacts. External references inform risk management; they do not certify this method or establish client results. Linked sources were selected for this review; their live revision dates may change.

• NIST AI Risk Management Framework (2023)

For anonymous documentation and proposal patterns, see delivery evidence and its boundaries.

Related decisions

• AI training that proves independent task performance
• A 100-day AI plan with evidence gates

Bring a workflow to the discussion

Share the business goal and constraints through the mandate process. Do not send sensitive production data in the initial brief.