Draft for Tamir's review. Not published.
Shape a commitment you can keep
Ask what the ministry needs by the date and what could follow. Commit to a dated first stage with the resources, approvals, procurement and security steps it needs, and make the rest conditional. Agree the commitment in writing.
For a citizen-facing AI assistant, define the questions it may answer and exclude decisions about individual rights. Test it on real citizen questions with correct answers before launch, and staff a human path for what it cannot answer.
Ownership, funding and channels
Keep a service only with an accountable owner and a workable operating model. Name the department that completes each request it accepts, and who updates the content when rules change. For a shared platform, separate shared capabilities from department requests and agree how common costs are allocated.
Close a staffed channel only when the digital service can meet the affected needs. Test complete outcomes through the new channel and check where unresolved requests end up. Judge new feature requests against the original service outcome and the changed cost of completing it.
Depending on your seat
If you're on a government company's board, approve a staged commitment with the first stage dated. Ask management for its delivery capacity and the risks of the timeline before the company commits.
If you run IT and must choose between the central shared platform and your own system, get the central option in writing. Ask for scope, dates and service levels for your functions, and which of them it runs in production for another unit. Check the policy on procuring alone.
What to check before you decide
- Ask what the ministry needs by the date and what can follow later.
- Commit to a dated first stage in writing, and make the rest conditional.
- Get the privacy and legal positions, and the procurement and security timelines, before the date is public.
- Test the service on real citizen requests and measure the results before launch.
- Keep a human path for what the service cannot handle, with staff for it.
- Name the owner who completes each request and updates the content when rules change.
- Close staffed channels only after complete outcomes are shown through the digital one.
Questions people ask
The ministry wants our government company to launch a digital service fast, what should the board require before committing?
Require a scope the company can deliver on the timeline, the resources and approvals it needs, the risks of the timeline, and a staged plan with a first version that works. Commit to the first stage with dates and make the rest conditional. It depends on the company's delivery capacity and on what the ministry actually needs by the date.
The minister wants a citizen AI assistant live in months, what should I insist on before we commit publicly?
Insist on a defined scope of questions it may answer, a tested answer set on real citizen questions, a visible path to a human, and clear accountability for wrong answers. A narrow assistant that is right beats a broad one that is sometimes wrong. It depends on which services the assistant covers and on what your legal and privacy advisers require for automated answers.
Who should fund a shared municipal platform when departments receive unequal benefits?
Agree the shared purpose and allocation principles before expanding scope. Continued investment depends on which costs are common, which benefits are local, and who owns the combined decision.
Should we keep a resident app that nobody owns operationally?
Keep a service only with accountable ownership and a workable operating model. The choice depends on resident use, unresolved work, and whether a simpler channel can provide the same outcome.
Should we close staffed service channels after launching a resident portal?
Close a channel only when the remaining service can meet the affected needs. The decision depends on completion, accessibility, assisted alternatives, and obligations checked by responsible owners.
Government unit choosing between the central shared platform and procuring our own system, how do we decide?
Judge the central platform on written commitments for your functions and dates, and judge your own tender on what central policy allows and what you can operate. Ask the central body for a service agreement with your unit as the customer. It depends on how specific your needs are and on whether central policy lets you procure alone.
Should we approve more features when a public digital program keeps expanding?
Decide additions against the original service outcome and the changed cost of completing it. It depends on genuine new obligations, expected public value, dependencies, and the authority to approve scope changes.
How I can help with this decision
- Ask or talk (Free)
- I give my view on which conditions make the launch survivable, and how to frame them for the ministry.
- Review (Pay if it was worth it)
- I write an independent assessment of the request, the scope, the readiness evidence and the ownership design. I recommend what to commit to and on what conditions.
- Retain (When it makes sense)
- I stay close through launch to review test results and progress against the commitment.