AI recipes: Two ingredients every association already has
You don't need a data project before you start with AI
Every association already has the two ingredients AI needs: structured data and unstructured knowledge. Most only ever use one.
If you've held off on AI because you assume there's a data project standing between you and a working assistant, there probably isn't. The two ingredients are almost certainly already sitting in your CRM and your filing cabinet, you just haven't put them to work together yet.
In short: structured data is anything organised in rows and fields, your CRM, membership records, event and engagement history. Unstructured knowledge is everything else, your reports, recordings, transcripts and documents. Most associations already have both. Most AI tools only ever use one.
What's the difference between structured data and unstructured knowledge?
Structured data lives in rows and fields; unstructured knowledge is everything else. Most associations have plenty of both, they just usually only build with one of them.
Structured data is anything that already lives in rows and fields: your CRM, membership records, event attendance, engagement history. It's organised by definition, which is exactly why it's usually the first thing people think of when someone says “AI needs data.”
Unstructured knowledge is everything else: your reports, webinar recordings, transcripts, emails, PDFs, FAQs, website content. It doesn't sit in neat fields, so it's easy to overlook, but it's often where the more valuable, harder-to-replace knowledge actually lives: the judgement calls, the context, the “why,” not just the “what.”
Most AI implementations only ever use one of the two. A chatbot built purely on structured data can tell a member their renewal date but can't explain your CPD policy. A tool built purely on documents can answer policy questions but can't tell a board member what next year's renewals are likely to look like. The gains that actually change how an association runs tend to sit in the overlap, where both ingredients are working together.
Do we have enough data to start using AI?
Almost certainly, yes. Structured and unstructured knowledge exist in some form in every association we've come across. The more useful question isn't “do we have data,” it's “where does ours actually live, and in what state?”
That's worth answering properly before you start, because it changes what you build first. Three questions are worth sitting down with:
- Where does your knowledge actually live, and in what state? Years of reports and recordings are only useful if something can find and read them.
- What structured data do you have, and how clean is it? A CRM full of duplicate or stale records will feed an AI assistant duplicate or stale answers.
- What can't be used yet, and why? Sometimes it's a permissions issue, sometimes it's format, sometimes it's simply that nobody's looked at it in years.
A short audit against those three questions saves a lot of wasted effort later. It's also usually much quicker than people expect, because the answer is rarely “we have nothing.” It's “we have more than we realised, and some of it needs tidying up.”
Do we need to clean up our data before we start?
Not as a prerequisite, but it does matter. This is where the National Association of Local Councils (NALC) is a useful example, because they'd already done the harder version of this work before AI came into the picture at all.
When NALC launched their ReadyMembership platform, they ran a full content audit and started from fresh, clean data rather than migrating legacy records they weren't confident in. At the time, that decision was about building a better membership platform. In hindsight, it turned out to be exactly the right foundation for AI. Clean, authoritative content and reliable member data are what make an AI assistant trustworthy, without them, even a well-built tool will produce shaky answers.
We needed ReadyIntelligence because our members are using AI tools but without reliable source information. Members and users don’t always know the exact terminology they need to use when searching. With ReadyIntelligence they can ask a natural question and they won’t only get the answer to that question but they get links to source information.
The lesson generalises well beyond NALC's specific platform decision: the work you do to tidy up your content and data isn't a side project separate from AI. It's the groundwork that determines how well any AI investment actually performs. And critically, AI itself can help with the tidying: tagging, summarising, and organising unstructured material is one of the things it's genuinely good at, so this isn't a multi-month project you have to finish before you're allowed to start.
Where this leaves you
Two ingredients, most of which you already have: structured data that's organised but often used alone, and unstructured knowledge that's valuable but often ignored entirely. The associations getting real value from AI aren't the ones with more data. They're the ones using both ingredients together, with a bit of care taken over quality first.
Every approach that follows in this series starts from this same pair of ingredients: making your knowledge findable, reading the charts and video you already have, proactive alerts for your team, and workflow automation. We call each one a recipe: a specific, repeatable way to put AI to work, not just a good idea. Knowing what you've got is the first step. The next one is picking which recipe to try first.
Quick recap
- Every association has both ingredients already: structured data and unstructured knowledge.
- Most AI tools only use one. The real gains sit in the overlap.
- A short audit (where it lives, how clean it is, what's unusable and why) beats a long data project.
- Clean, trustworthy data isn't a prerequisite, but it is what makes an AI assistant worth trusting, and AI can help tidy it up rather than needing it finished first.
Ready to see what's next? Download the recipe card deck: four practical recipes for putting these two ingredients to work in your own organisation, not just examples of how other people have.