11 Dec
AI knowledge management is the invisible crisis behind every AI project. While enterprises rush to deploy AI tools, they overlook the foundation that determines success or failure. Your teams waste hours searching for information. Expertise disappears when employees leave. New hires take months to become productive. And AI amplifies everything—your strengths and your weaknesses equally.
Consider SPIE ICS‘s service desk: 3,300 employees, 140 support agents facing high turnover and operational pressure. By implementing AI knowledge management, they halved turnover whilst slashing information search time by 73%. Their approach? Build proper knowledge foundations before deploying any AI.
At the ADIRA 2025 Convention in France, they shared exactly how they did it. Here’s how you transform scattered tribal knowledge into strategic advantage – the four core transformations, the measurable wins, and the methodology for building a knowledge engine that makes your organisation antifragile.
Table of contents
Why AI knowledge management is the strategic move enterprises skip
Everyone’s talking about AI. Fair enough – it’s powerful stuff. But here’s what gets glossed over: behind every successful AI deployment sits robust knowledge management. Data might be “AI’s fuel”, but there’s a world of difference between raw data and knowledge that actually creates value.
Data is just… stuff. Raw, often messy, sometimes actively misleading. Information happens when you structure that data within your business context. But knowledge? That’s where you bring in the human and organisational factors that create meaning. Real, applicable meaning that people can use.
It’s this knowledge – your true intangible asset – that deserves strategic attention.
During workshops at the ADIRA Convention, Elium and SPIE ICS dug into this challenge. Beyond the theory, they showed concrete daily transformations with measurable results. More than that, they demonstrated how to build organisational robustness – creating organisations that don’t just survive crises, but come out stronger. Nassim Nicholas Taleb, author of the Black Swan theory, calls this antifragility.
Knowledge is what lets you solve a problem you’ve never seen or have forgotten, using the experience of people who’ve already cracked it. Without it, every crisis starts from scratch.
AI knowledge management in action: transforming Service Desk operations
SPIE ICS, a major French IT services firm, partnered with Elium – we’ve specialised in knowledge management for 15 years. Perfect match of field reality and tech expertise.
The challenge: fragmented and obsolete
SPIE ICS faced what most organisations face: knowledge scattered across Teams, OneNote, file servers, wherever. Finding anything quickly became impossible. Team collaboration suffered.
Pascal Coquerel, head of managed services development, puts it simply: “When knowledge lives only on a file server, you’re not working in the same context. That’s just content storage – no real sharing, no evolution.”
The knock-on effects were brutal. Information goes stale. Updates don’t happen. Quality decisions become impossible. Operational efficiency tanks.
The fix: from silos to collective intelligence
SPIE ICS deployed a solution integrating their ITSM tool with Elium’s platform to transform their service desk. Worth noting: Elium plays nicely with any ITSM – EasyVista, GLPI, JIRA, ServiceNow, Zendesk, the lot.
Their Toulouse-based service desk handles over 540,000 requests annually with 140 people. The measurable results come later, but the philosophy shift happened immediately. “The individual agent works for their client, yes – but now they work just as easily for the collective,” says David Matrat, consultant at SPIE ICS. “Everyone contributes to a shared knowledge pool.”
That’s balanced performance: operational gains that don’t burn out your people. They improve everyone’s daily work instead. Sustainable because it’s built on wellbeing and employee value in a genuinely tough profession.
4 core transformations of AI knowledge management
Transformation 1: from hunting documents to getting solutions
Traditional knowledge management is like a well-organised library. You’ll find documents, procedures, guides. But interpreting them, synthesising them, adapting them to your specific context? That’s all on you.
Augmented KM is the expert librarian who’s always on. Knows your needs. Anticipates your questions. Gets better continuously. Crucially, doesn’t hand you documents – gives you contextualised answers and solutions.
Real example: “How do I handle an angry customer after a delayed delivery?” or “What’s our network incident procedure including our specific gotchas?” or “Deployment checklist for banking clients with GDPR requirements?” The virtual assistant pulls from the entire knowledge base, identifies relevant solutions for your precise context, suggests pre-written responses. You validate, adjust if needed, save considerable time.
Transformation 2: from dreaded task to natural habit
The biggest barrier to knowledge sharing has always been friction. Where does this document go? How should I structure it? Which sub-folder in our maze of directories?
Generative AI changes everything by helping throughout the contribution chain: integrates existing documents, auto-summarises, improves writing, adds illustrations, suggests classification. What felt like admin hell becomes a quick, natural action. Everyone enriches the collective knowledge without excessive effort.
Transformation 3: from hidden decay to active governance
This might be the most critical shift. Traditionally, content obsolescence is a blind spot. You might stumble on an outdated procedure, but the damage stays limited to one incident.
With generative AI, the stakes change completely. “Quality becomes fundamental in this new era,” says Gregory Culpin. “When I base responses on an assistant’s answer, or trigger automated flows responding to dozens or hundreds of clients, one error can propagate everywhere.”
This reality demands rigorous, proactive governance. AI detects obsolete content, flags inconsistencies, suggests updates. It becomes your quality guardian for a responsibility that humans often neglect through sheer lack of time.
Transformation 4: from static storage to living improvement
Traditional management often stops at publication. Knowledge stagnates. Augmented KM creates a permanent feedback loop: interactions tracked, hallucinations detected, sources corrected at the root. Every gap identified becomes an evolution request, managed like a ticket with production status tracking.
This transforms knowledge management into a living discipline. “The phone rings less, so everyone’s happier and we’ve got more time,” observes Pascal Coquerel. Fewer calls prove your base is working – users find answers autonomously and solve problems themselves. Field requests directly feed your content priorities.
That’s how you build antifragility: each resolved incident enriches knowledge, which helps resolve the next one faster, which frees time to improve the base further. Your organisation stops suffering problems and starts systematically turning them into learning opportunities.
The multimodal future of augmented KM
One more dimension emerging: multimodality. Upstream, capture gets richer – voice, visual, video, conversational. A technician films their fix, a salesperson dictates feedback from the car, AI structures it automatically.
Downstream, access diversifies: ITSM/CRM/ERP integrations, conversational assistants, APIs for autonomous agents, proactive notifications. Knowledge comes to you at the right moment in the right format. Multimodality democratises contribution and maximises use by everyone – humans and AI alike.
Measurable results: performance and wellbeing together
SPIE ICS ran rigorous audits before and after implementing their knowledge management programme with Elium. The numbers tell the story of genuinely balanced performance.
Operational wins: productivity gains measured
- 73% reduction in information search time
- 25.5% reduction in critical incident processing time
- 30–40% reduction in traditional incident time (varies by type)
- 44% reduction in time to publish new knowledge
- 65% improvement in knowledge quality
Human impact: turnover, training, engagement
- Turnover almost halved: from 29.6% post-pandemic (2023) to 17.17% end of 2024
- Training accelerated: from three weeks to two weeks to reach operational level
The balance guaranteeing sustainability
These two dimensions don’t compete – they fuel each other. Their balance guarantees lasting results. Pure operational optimisation exhausts teams. Wellbeing-only approaches lack business impact. Augmented KM creates the conditions for performance that’s both effective and sustainable.
In a sector facing brutal turnover and burnout, these figures show profound cultural transformation. “Today, employees drive knowledge enrichment and sharing within the tools. This collective work valorises the profession whilst strengthening engagement and belonging,” says Pascal Coquerel.
The virtuous circle
SPIE ICS needed to stabilise teams and guarantee user experience. Reducing turnover builds precious expertise, which feeds a richer knowledge base. That improved base accelerates new hire onboarding, strengthens team cohesion, boosts motivation. Better job satisfaction means greater talent retention. Beyond the human wins, this builds organisational robustness.
Building antifragile organisations with AI knowledge management
In our volatile, uncertain, complex, ambiguous (VUCA) world, knowledge becomes your resilience factor. “We all have backups for our information systems, but that maturity doesn’t extend to knowledge,” observes David Matrat.
Beyond simple resilience, well-designed knowledge management creates antifragility. Your organisation doesn’t just resist shocks – it profits from them to get stronger. Each resolved incident feeds the knowledge base. Each departing employee becomes a capitalisation opportunity. Each crisis improves your capacity to handle the next one.
This antifragility rests on balanced performance. Organisations optimising only costs or speed exhaust themselves and become fragile. Organisational robustness emerges when you balance operational and human performance, immediate efficiency and learning capacity, short-term wins and long-term sustainability. This balance separates organisations that suffer uncertainty from those transforming it into competitive advantage.
Your knowledge deserves the same attention as other strategic assets. You invest in talent, property, equipment. Why not intellectual capital? The stakes are real: anticipating departures and crises, operational performance, employee wellbeing, organisational agility.
Building the foundations AI demands
Augmented knowledge management isn’t optional anymore. Here are the priority pillars:
- Facilitate access. Your people need answers, not document hunts.
- Reduce contribution friction. If sharing feels like a chore, your base stays poor.
- Establish proactive governance. Content quality and freshness aren’t negotiable.
- Create continuous improvement loops. Each interaction must feed base enrichment.
- Embrace multimodality. Knowledge must adapt to everyone’s work context.
But careful: “The tool doesn’t do everything,” reminds David Matrat. “You need engagement, sponsorship, communication, processes.” An augmented KM project is fundamentally cultural transformation. Technology enables it – it’s not a miracle cure.
Start with the right questions: What’s your current maturity? Where are you losing time? What are your blind spots and risks? How do you handle key employee departures?
Invest in a “knowledge engine” : your next competitive advantage
SPIE ICS proves it works: AI-augmented knowledge management transforms organisations. 73% less search time, turnover halved, faster training. In an uncertain world, this might be your best strategic investment.
Run a 360° diagnosis to map what exists, identify friction points, build a roadmap matching your maturity and ambitions. We’re here to help with this strategic work.
Acknowledgements
Thanks to the ADIRA Convention organisers and the speakers whose presentations inspired this article:
- Gregory Culpin, Elium – CCO
- David Matrat, SPIE ICS – Strategic consultant and co-facilitator of ADIRA’s Responsible Digital group
- Pascal Coquerel, SPIE ICS – Head of managed services development DAGE
