Have you ever had so messy closet that you ended up buying new clothes because you couldn’t find anything? For years, we’ve obsessed over collecting data, tossing everything into ever-deeper data swamps.
Juan Sequeda insightfully pointed out already in 2022, we must shift from data-first to knowledge-first, stop dumping data into swamps, and start modeling relationships. His research in 2023 and 2024 showed that knowledge graphs can dramatically improve LLM accuracy. And as he recently emphasised, clearly and convincingly: “The ontology/semantics arguably will play the definitive role of competitive differentiator in this AI world.”
Why knowledge graphs matter
According to the KPMG AI Quarterly Pulse Survey 2025, a majority of IT leaders cite data quality as a significant AI hurdle – lacking organisation, standardisation and scalability. To make data AI-ready, it needs to be:
- Contextual
- Flexible
- Standardised
Knowledge graphs and ontologies offer exactly that! Ali Khalili captured it beautifully: “Knowledge graphs are the lighthouses in the ocean of data.” They encode meaning, context, and lineage, the structure agents rely on to make sense rather than noise.
And as Kingsley Uyi Idehen cleverly wrote, “the semantic web didn’t fail, it just needed AI (the Yin to its Yang)”.
Real impact, real results
Klarna, the innovative Swedish fintech company, replaced 1,200 SaaS tools and transformed operations with AI & Knowledge Graphs, and CEO Sebastian Siemiatkowski humorously stated: “Feeding an LLM the fractioned, fragmented, and dispersed world of corporate data will result in a very confused LLM.” Thanks to Emil Eifrem and his Neo4j, they managed to avoid that mess.
At LinkedIn, implementing a knowledge graph for customer support boosted resolution accuracy by 78% and cut handling time by 29%.
At Glean, Arvind Jain demonstrated how KG-anchored agents can close IT tickets autonomously because they actually “understand” enterprise-specific entities like people, apps, and assets.
The symphony effect
Think of it this way: a knowledge graph becomes the shared musical score that every AI agent reads from. Same semantic vocabulary, same contextual cues, same harmonic rules. But here’s where the magic happens: while each agent plays its specialised part, together they create symphonies of insight that no solo performance could match. Structured knowledge transforms AI cacophony into an enterprise concerto. Without it, we risk noise and dissonance. With it, we enable the seamless harmony and clarity of a well-conducted masterpiece.
How can you start shifting your organisation’s focus from raw data to structured knowledge?
The next part of this post series is coming soon: Blending continuous creativity with discrete logic.