Decide whether to build or wait for AI capabilities

The problem

You've identified an AI use case that could help your charity, but you're not sure whether to invest in building it now or wait. AI is moving fast - what's cutting-edge today might be a free feature in your existing software next year. You don't want to waste limited resources building something that becomes obsolete, but you also don't want to fall behind while waiting for perfect solutions.

The solution

Use a structured decision framework that evaluates four factors: urgency of the need, technology maturity, your organisation's readiness, and the commoditisation trajectory. Score each factor and use the results to decide: build now, wait 6 months, wait 12 months, or never build (wait for it to become a product feature). This prevents both premature building and perpetual waiting.

What you get

A documented decision with clear reasoning you can share with trustees or funders. The output includes: your scores on each factor, the recommended action (build now/wait/never build), specific conditions that would change the decision, and a review date to reassess. This creates accountability and makes it easy to revisit the decision later.

Before you start

• A specific AI use case you are considering (not just "we should use AI")
• Basic understanding of what the use case would require
• Knowledge of your current systems and team capabilities
• Awareness of your budget constraints
• Consider data privacy and ethical risks as part of your evaluation - if the use case involves beneficiary data, factor in DPIA requirements and bias risks

When to use this

• Before committing budget or staff time to an AI project
• When evaluating vendor pitches for AI solutions
• During annual planning when prioritising AI initiatives
• When a trustee or funder asks "why aren't we using AI for X?"
• When you're excited about a new AI capability but want a reality check

When not to use this

• The decision is trivial (e.g., using free ChatGPT for drafting)
• You've already committed and can't change course
• The use case is legally required with a fixed deadline
• You have no budget regardless of the outcome

Steps

1. 1

   Define the use case clearly

   Write down exactly what you want AI to do, who would use it, and what success looks like. Be specific: "Automatically categorise 500 incoming enquiries per month and route to the right team within 2 hours" is better than "use AI for enquiries". If you can't be specific, you're not ready to evaluate.

2. 2

   Score urgency (1-5)

   How pressing is this need? Score 5 if the problem is causing significant harm now (staff burnout, service users not getting help, major inefficiency). Score 1 if it's a nice-to-have with no deadline. Consider: What happens if we do nothing for another year?

3. 3

   Score technology maturity (1-5)

   How proven is the AI approach? Score 5 if many similar organisations have done this successfully with off-the-shelf tools. Score 1 if it requires cutting-edge research or custom development. Check: Are there case studies? Established vendors? Open source tools?

4. 4

   Score organisational readiness (1-5)

   Can your organisation actually implement this? Score 5 if you have clean data, technical capacity, staff buy-in, and budget. Score 1 if you'd need to fix multiple prerequisites first. Be honest about what "ready" means.

5. 5

   Score commoditisation trajectory (1-5)

   How likely is this to become a standard feature soon? Score 5 if major vendors (Microsoft, Google, Salesforce, your CRM) are actively building this - it'll be a checkbox feature within 12 months. Score 1 if it's highly specific to your context and unlikely to be productised.

6. 6

   Calculate and interpret

   Add up your scores (minimum 4, maximum 20). 17-20: Build now - the need is urgent, tech is ready, you're capable, and it won't be commoditised soon. 12-16: Wait 6 months and reassess - something isn't quite ready. 8-11: Wait 12 months - multiple factors suggest waiting. 4-7: Never build - wait for it to become a product feature or reconsider the use case entirely.

7. 7

   Document your reasoning

   Write a one-page summary: the use case, your scores with brief justifications, the decision, and what would change it. Set a calendar reminder to review. Share with stakeholders so the reasoning is transparent.

8. 8

   Use an LLM as a sparring partner

   Paste your assessment into Claude or ChatGPT and ask: "Challenge my reasoning. What am I missing? Are my scores realistic?" This catches blind spots and strengthens your case.

Tools

Resources