A few weeks ago we published a piece called We Called It Context. Turns Out, It's Ontology. It resonated more than we expected — which tells us the underlying problem is hitting a nerve.
So we wanted to go deeper. Not on the theory, but on the practical question: what does an ontology-first AI strategy actually look like?
The problem with data-first
Most AI strategies follow a familiar sequence:
- Audit the data
- Clean the data
- Centralise the data
- Then — finally — do something with AI
Steps 1 through 3 take 12-18 months. Step 4 never quite arrives. Or when it does, the landscape has shifted and the audit needs redoing.
The assumption is that AI needs clean data to work. That's partly true for certain use cases — machine learning models trained on structured datasets, for example. But for the majority of enterprise AI applications — search, knowledge management, workflow automation, agent-based work — what AI needs is structured relationships, not clean spreadsheets.
It needs to know that "marketing" in your CRM refers to the same team as "brand and comms" in your project management tool. That a "proposal" in Salesforce and a "pitch deck" in Google Drive are related artefacts of the same sales process. That when someone says "the Sydney project," they mean the three workstreams, the seven team members, and the 40 documents associated with it.
That's ontology. And no amount of data cleaning will create it.
What an ontology-first strategy looks like
Instead of starting with data quality, start with three questions:
1. How does work actually flow in this organisation?
Not the process maps from three years ago. The actual flow. Who requests what from whom? What gets approved and by whom? Where do handoffs happen? Where do things get stuck?
This isn't a data question. It's a people and process question. And it's the foundation that makes AI useful.
2. Where are the relationships between systems?
Map the connections. A customer in your CRM is a contact in your email, a signatory on your contracts, and a name in your project management tool. Those are four representations of the same entity. AI needs to understand that.
Platforms like Glean do this automatically through their knowledge graph. By connecting to your existing systems and mapping the relationships between people, documents, and projects, Glean creates the ontological layer that AI needs — without requiring you to build it manually.
3. What does your organisation know that isn't written down?
This is the hardest question and the most valuable. Every organisation has institutional knowledge — the "ask Sarah, she knows" kind — that exists in people's heads but not in any system.
An ontology-first strategy surfaces this knowledge by making it easier to capture and connect. When AI can understand relationships, it can also identify gaps — areas where knowledge exists informally but hasn't been documented.
The Glean connection
We implement Glean because it's the most practical tool we've found for building organisational ontology without a multi-year project.
Glean doesn't ask you to model your ontology in a formal schema. It doesn't require a data engineering team. It connects to your systems — all of them — and constructs the knowledge graph from what already exists. The permissions, the authorship, the project associations, the team structures — it's all already there in your tools. Glean makes it visible and navigable.
This is what we mean by practical ontology. Not an academic exercise. A working model of how your organisation operates, built from the systems your people already use every day.
Strategy before tools
The shift we're asking for is simple but important: before you buy another AI tool, before you launch another data clean-up project, describe how your organisation actually works.
That description — that ontology — is the foundation everything else builds on. Search, agents, automation, knowledge management — they all depend on structural understanding.
Our AI Readiness Blueprint is designed to create exactly this understanding. It maps your organisation's people, processes, and technology, identifies where AI can create the most value, and produces a strategy that starts with structure rather than data.
Ready to build your AI strategy on the right foundation? Talk to the JOURN3Y team.