Where you are with AI, and what to do about it.
A team assessment that ends with your first moves and the kind of partner who helps.
Developed by
Funded by
Schwab Foundation
If AI worries you
You can be skeptical about AI.You can’t be passive.
Many social impact organizations are wary of AI. The people they serve have the most to lose when it goes wrong.
Waiting does not keep AI out. Once people use it in their own work, the organization is already at Exploring.
What a result looks like
A result reads where you are, then names your first moves.
Northgate Food Bank is an invented example. Its AI ranks the families that apply for parcels.
Where you are
Systematizing, conditional
- Aspiration
- Hybrid organization. Systematizing or Innovating is healthy.
- How far AI goes
- The ranking already goes as far as Innovating.
- In place
- 19 of the 24 things Systematizing asks for.
What to do about it
Agree first on how the team shares what works.
- A way for families to question the ranking.Who helps: governance design with a pro-bono team
- An owner for the data the ranking uses.Who helps: a data-readiness sprint with a data partner
- Criteria for which workflows are ready for normal operations.Who helps: responsible-AI advisory with a nonprofit intermediary
Next step of the Adoption Loop: deploy with proportional governance.
The five stages
Five stages, each with conditions you can check.
Deciding conditions place a stage. From Systematizing up, a readiness checklist confirms it.
Systematizing
Reached when
The organization runs AI pilots on selected workflows, or has run the Adoption Loop on them more than once.
Required
- A named AI owner, champion or adoption group with time and authority.
- People share what works and what fails.
- AI use is expected in selected areas. One team or function where AI is part of how the work is normally done is enough.
Then confirmed by
Up to 24 readiness conditions. Seven of them apply where AI affects the people you serve.
From Systematizing up, a stage has a status
- Confirmed
All deciding and all readiness conditions that apply are met.
- Conditional
All deciding conditions are met, and more than half of the readiness ones that apply.
- Not reached
A deciding condition is missing, half or fewer of the readiness ones are met, or a stage below is not confirmed.
What the Navigator checks
Seven domains, and what is looked at in each.
Your result shows what is in place and what is missing, domain by domain. There are no colors and no score.
Leadership and ownership
Up to 10 conditions
- An owner with time and authority
- Accountable workflow ownership
- Workflow priorities and a regular cadence
- A successor for the owner
- Agreement among those who answered
People, workflows and change
Up to 6 conditions
- A team learning rhythm
- A map of changing tasks and roles
- Staff shaping how AI is used
- Redesigned workflows with explicit review
- Staff and user acceptability
Data and organizational context
Up to 2 conditions
- A data and context inventory
- Ownership, upkeep and quality checks
Technology, implementation and resilience
Up to 10 conditions
- An approved tool stack
- Technical support and maintenance
- Security and privacy
- Vendor, model and partner dependency
- Portability, rollback or migration
Governance, risk and accountability
Up to 15 conditions
- Safe-use and data-handling norms
- A risk and consequence assessment
- Review, escalation and exception rules
- Criteria for going into operation
- Governance and failure handling in practice
Seven protections for the people you serve
They apply when AI affects those people. A gap in any of them comes first in your recommendations.
- Affected people or their representatives take part
- Written limits on automated decisions
- Qualified, accountable human review
- A way to question, correct or appeal
- Testing for exclusion, bias and harm
- A plain explanation of AI’s role
- Safeguards for the political and surveillance context
Evidence and learning
Up to 10 conditions
- Sharing what works and what fails
- Baselines and intended outcomes
- Evidence that changes the workflow
- Monitoring and failure reviews
- Outcomes over output volume
Resources and sustainability
Up to 5 conditions
- Budget, credits and protected time
- Cost and consumption controls
- A plan beyond temporary funding
- Sustainable funding and maintenance
- Distribution partnerships
How far AI goes in your work
6 conditions, outside the domains
When AI goes further than the organization can carry, the result says so in one sentence.
Who is behind it
A safe path into AI is Tech To The Rescue’s mission.
Safe for your organization, your volunteers and the people you serve.
How the Navigator holds to it
The people you serve come first
Where AI affects them, a gap in the seven protections leads your recommendations.
Your team before the tools
The Adoption Loop starts with literacy, trust and permission.
No pressure to climb
A higher stage is not a better result. Your aim fits your kind of organization.