Your Knowledge Base Is Worthless Without This
July 1, 2026
The Quiet Shift Happening in Professional Services Right Now
Most founders I talk to are still thinking about AI as a search tool. Something you ask questions to. Something that saves you 20 minutes writing an email.
That's fine. But it's not where the real leverage is anymore.
The story that caught my attention this week was Kantata releasing their Expertise Agent — a system that doesn't just store institutional knowledge, it *routes* it. Automatically. In context. At the moment someone actually needs it.
That's a different category of tool entirely. And it points to something most service businesses are about to get wrong.
---
The Problem Isn't Information. It's Access at the Right Moment.
Here's what I see constantly in operations-heavy businesses — agencies, consultancies, professional services firms with 10 to 200 people:
They have knowledge. Lots of it. SOPs in Google Drive folders nobody opens. Slack threads with brilliant problem-solving buried three months back. Senior people who've solved the same problem six different times who are now the bottleneck because junior staff don't know what they don't know.
The average knowledge worker spends somewhere around 2.5 hours per day searching for information they already have access to. That's not a stat I'm making up for effect — it's been cited consistently across McKinsey and IDC research for years. In a 10-person firm, you're burning roughly a full-time employee's worth of hours every week on retrieval.
That's not a morale problem. That's not a training problem. That's a *system* problem.
---
What an Expertise Agent Actually Does Differently
Kantata's approach — combining a services-native knowledge base with an agent that surfaces relevant expertise automatically — isn't magic. It's a pattern that's been technically possible for a while. What's new is that it's becoming productized, accessible, and embedded directly into project workflows rather than sitting as a separate tool that people have to remember to use.
The distinction matters. A knowledge base you have to search is only as good as your people's ability to know what to search for. An agent that watches what you're working on and surfaces relevant context *without being asked* — that changes how junior staff perform, how fast onboarding happens, and how consistently your best thinking gets applied across the business.
Think about what that looks like practically:
A project manager at an 8-person consultancy opens a new client engagement. Instead of scheduling a 45-minute knowledge transfer with a senior consultant who's already stretched thin, the system surfaces the three most relevant past engagements, the approach that worked, and the one thing that went sideways last time. Before the first client call.
That's not theoretical. That's what agentic systems built on good internal data can do right now.
---
What You Should Actually Do With This Information
You don't need Kantata specifically. You need to understand the architecture they're using and start building toward it in your own stack.
Here's the practical sequence:
1. Audit where your expertise actually lives. Not where it's supposed to live — where it actually lives. For most firms, it's in the heads of 2-3 senior people, in email threads, and in Slack. That's your raw material.
2. Start capturing it in a structured, retrievable format. This doesn't mean a massive documentation project. It means building a lightweight habit: when a problem gets solved, a two-paragraph write-up goes somewhere central. Notion, Obsidian, a simple internal wiki — doesn't matter much, consistency matters more.
3. Connect that knowledge base to an agent layer. Tools like n8n, Make, or custom GPT configurations can start routing that context intelligently. This is where working with someone who understands the architecture pays for itself fast — the difference between a system that actually gets used and one that becomes another forgotten folder is mostly in how the retrieval logic is set up.
4. Measure the before and after. Pick one workflow — onboarding a new client, scoping a proposal, handling a recurring question type — and time it before and after. Real numbers. Not vibes.
---
The Opportunity Window Is Shorter Than You Think
The firms that figure out knowledge routing in the next 12-18 months are going to have a compounding advantage that's genuinely hard to replicate later. It's not about having better AI models — everyone has access to the same models. It's about having better *organizational memory* connected to those models.
That's a moat. And it's being built right now, quietly, by businesses that are thinking one level deeper than "let's use AI to write faster."
If you want to talk through what this looks like for your specific operation, I'm at degrand.ai/contact. No pitch deck, just a direct conversation about where the real leverage is in your business.