Putting the AI in ITAM

Author: Keiren Tilbrook – Head of Analysis and Design at The Mastermind Group

When I asked Bing Chat for the hottest themes in IT for 2024, it gave me a snappy summary of some Gartner content (sidenote: I asked it was I allowed to publish the info it just gave me and it said yes, so here you go). Unsurprisingly, 6 of the top 10 related to artificial intelligence, with the top one being about building trust in AI systems. This is a perfect segue to ITAM where trust is the foundation of, well everything. 

First let’s map the territory.  

AI is an umbrella term that covers a suite of technologies including decision support, computer vision, machine learning, natural language processing (NLP), and generative AI. It’s the combination of these last two that has got everyone hot and sweaty in the past 12 months; they allow us – for the first time – to have a meaningful conversation with our computers. The term ‘generative’ means that AI can create original content, for example producing a concise summary of data or a document. This talent is underpinned by large language models (LLM), which can be general (e.g. the internet) or specific to a particular task or industry. The more general the LLM, the more likely you are to get inaccurate or flat-out wrong responses, otherwise known as hallucinations. 

So, what is the current state of AI for IT asset management?  

This is a fast-moving train, and the answer will be different in 12 months, but we can frame the question in three ways;  

  • AI inside ITAM tools;  
  • ITAM data in external AI solutions; and  
  • If/how ITAM can help manage AI. 

In terms of how much AI there currently is in ITAM tools the answer is, not a lot… yet. 

Flexera’s solution has for some time had a ‘license recommendations’ feature that suggests license changes based on automated comparison with its SKU/PURL libraries. This could be considered a rudimentary form of intelligence, but I worry I may have just triggered PTSD for practitioners who’ve actually tried to implement them.  

ServiceNow SAM has for a while used machine learning to improve the normalisation rate for discovery models. The Vancouver release comes with Now Assist that embeds generative AI into various, mostly service management related functions. The only ITAM related one so far is AI powered service requests for software or hardware.  

Both companies harness some level of intelligence in their software spend detection module, where you send/upload an expense report and it get parsed to extract SaaS purchases. 

Almost none of the above uses generative AI. Doing so requires ITAM-specific LLMs, and we expect to see these from major vendors in 2024. It will be interesting to see how this shakes out but it’s likely these LLMs will become proprietary IP, much like the existing content and normalisation libraries. 

In terms of future features, Snow (under offer to Flexera) has a beta product Atlas Copilot that supports natural language queries against the ITAM dataset. It has announced plans for ML-enabled entitlement upload and an OpenAI augmented Data Intelligence Service (I don’t quite know what that means either). ServiceNow also has an ITAM specific AI roadmap, with early features likely to include AI-driven contract analysis and license positions. The latter includes the fascinating possibility of conversational analysis, being able to ask a bot to explain a particular term or drill down on a statistic.  

AI outside the tools 

The main use case is extracting data and feeding it to an AI service for analysis. At a basic level this is pretty easy; you can get OK-ish results using advanced query functions in Excel / Power BI to ask simple NLP-ish questions and automate the creation of visuals. From there it gets more complicated… I wanted to report on a wonderful experience using Copilot in Power BI, but after some research it appears – at the time writing – this combo is only available via Power BI Premium Capacity at a seriously primo price, with no trial available. Otherwise, with a paid subscription to Chat GPT – or one of a growing herd of AI data analytics services such as Luzmo – you can upload datasets for analysis, but handing over data to other companies and geographies is an issue for many enterprises. Reading and understanding T&Cs is going to be absolutely critical, which leads in nicely to the next topic… 

Can ITAM help manage the use of AI in organisations?  

This topic was discussed in a recent ITAM Forum webinar on AI, and I thought it was a reach; however, its undeniable that interpreting demand/subscription-based contracts is firmly in the ITAM wheelhouse. This was reinforced in my own experience trying to enable Copilot for our Azure tenant, which led me down a dimly-lit path of Microsoft Fabric and Power BI Premium license arrangements. The relationship between ITAM and AI is a similar Venn diagram to FinOps; it doesn’t cover the whole thing but there is definitely value ITAM can provide.  

A final thought with shades of Ouroboros (ask Bing Chat)… A strong use case for ITAM AI is to enhance the cloud cost optimisation (CCO) solutions needed to help control the massive growth in cloud infrastructure required to drive AI… 

Author: Keiren Tilbrook – Head of Analysis and Design at The Mastermind Group

If you would like to hear more information about this topic, please reach out to us via email at info@tmg100.com