Key takeaways
- ✓A good librarian sorts and archives everything consistently, from emails to phone calls and meetings
- ✓The more categories it adds, the faster you find things and the sooner it spots trends and problems
- ✓Centralise your unstructured data and link it systematically to the data from your core systems
- ✓An agent that only gets the information it needs answers better and costs less
The librarian we deliver with our platforms
At Encima we now deliver our platforms together with an autonomous librarian. It sorts and archives everything neatly and consistently. Whether it is emails, questions over the phone or a meeting, it keeps track of all of it.
It does more than keep track. It makes sure all information can be found quickly, by adding as many categories as possible: who the author is, which genre it is, in which city the story takes place, whether it is part of a series. Its overview is so good that it spots trends and problems before they surface. It does not wait until people complain or an opportunity has gone.
From books to emails
Now replace the books with emails, documents and recordings. The library becomes a custom platform, the categories become metadata and a knowledge graph, and the librarian itself is an AI agent.
Letting AI loose on a disorganised way of working is like putting a new, expensive engine in a vintage car. It goes faster, but the risk of errors remains, and finding things out remains a large and costly job. AI on a well organised library is fast and gives the most accurate results.
A second brain for your company
More and more mid sized companies see where this is heading. Some call it a second brain, others an intelligence operating system. It means systematically centralising all your unstructured data streams: centralising all emails and making them accessible internally where possible, recording and transcribing internal and external meetings, and linking all of it systematically to the structured data from your core systems.
That is not easy, both in terms of change management and in terms of privacy. But the leverage it gives is large.
A dictionary and a memory
Companies that go a step further also keep a central dictionary of the terminology in the company, with all synonyms and the links between those words. Such an ontology prevents everyone from keeping their own definitions, and prevents double truths for the agents in the company.
Context graphs also capture the links over time, so you can always trace who decided what, and when.
Less context, better answers
If you give an AI agent only the information it needs, at the right moment, its context stays limited. You get better answers and lower token costs, because useless searches drop out.
Compare it to looking for information in a library where everything is neatly sorted, versus a pile of reports and books lying around. The second takes much more time, and the chance you find everything you need is much smaller. That was true before, and it is still true now.
Making knowledge explicit
In the systems we set up, we bring every employee and every agent as close as possible to the company's context, so everyone can reach the right information at any time. Not only on request, but proactively too, so important action points reach the right people with the right priority.
Making knowledge as explicit as possible has to be carried by the whole organisation. Companies that do so set a flywheel in motion that is hard to catch up with.
Sources
Ready for your own librarian?
Tell us how the work runs today.
