I’d like to provide an update on the AI ITAM tools I’ve been looking at over the last four or five months. I’ve grouped these into what I see as four core categories: tools where AI is arguably table stakes, what all technology should have these days; a new category of startups doing data aggregation; real disruptors in the ITAM space using AI; and a fourth I’m calling the smart bet, a call on where expertise and AI really sit in future.
This is based on briefings I’ve done with technology vendors and technology I’ve been watching in the space. Caveat: This is not based on deep-dive technical review or proof of concepts, and some of these technologies are very new to the market.
Table stakes
First of all you’ve got the likes of EZO AssetSonar. What they’re doing is putting AI as a copilot within the technology, so you can ask questions rather like a chatbot. “How do I do this?” type questions, sourced from a knowledge base. Chatbots of this sort have been in technology for a number of years, but what they’re moving towards is action. Not just “how do I”, but “can you do this”: push this asset through this process, for example.
They’re also doing things that have been table stakes in the ITAM space for a little while, pre-2023 and the AI explosion, such as ingestion of forms. Upload a purchase order or a spec sheet and it populates a form based on that data. And they’re doing AI-driven insights: tell me interesting things about this asset that I should know, or what I should do about certain types of assets.
This is a good example of table stakes. Any IT management technology, not just in the ITAM space, should be offering this sort of thing. AssetSonar in particular is addressing the mid-market, and their customers have most likely never used an asset tool before; they’re most likely replacing a spreadsheet. So although I say this is table stakes, it’s going to be very advantageous for their customers.
Removing the need for external reporting
In a similar vein is Calero, who have a long history in technology expense management (TEM). One of the failure points of a lot of tools in the ITAM space is that they do all sorts of fancy reporting, but you end up exporting it to Excel to really get the data you want. What I liked about what Calero are doing is you can have a conversation with the report to develop it into what you wanted. I think this is the direction all tools will take. It’s one less export of data to a spreadsheet; you don’t have to push it out to Tableau or Power BI. You can manipulate the data to get the answer you want and present it to your manager without leaving the platform.
What these tools are doing is not a traditional LLM scouring the internet. They’re using your data as the resource. My analogy here is NotebookLM versus ChatGPT: NotebookLM is AI which only looks at the documents you upload to it, as opposed to an LLM trained on a massive corpus.
Both these companies are doing cool things, but I don’t think it’s particularly disruptive or innovative. As I mentioned, these are table stakes. Their roadmaps point to much more exciting things: AssetSonar looking at detecting shadow AI usage, and Calero looking at token consumption. Both are exciting for the future.
The data aggregators
The next type of tool I’m seeing in market is startups that do data aggregation. I looked at Samplify in a previous roundup, and this time I’d like to point you towards a new startup called Archways. What they’re doing is aggregating market intelligence about vendors and helping you make smarter decisions.
The point of friction Samplify looks at is the incoming request: you’ve got a request for a piece of software, and Samplify helps you avoid that purchase by saying, actually, you’ve already got something in house that does that, here’s the technology, and here’s how to word the email back to the requester. You thwart the request and stop sprawl.
SaaS Renewal Intelligence
Archways is looking at a similar sort of dataset, your existing procurement data from systems like SAP, and surfacing when renewals are coming so you can prepare. The major SAM tools already help customers do that for the major vendors. Where I think this is particularly useful is the long tail of renewals that creep up on you and have auto-renewed before you even realised they were due. It allows you to do the spadework of preparation: do I need this, at what point do I need it, and how do I prepare? A lot of that data is actually at your fingertips, if only you knew how, and this is what Archways is promising. Interesting startup, one to watch.
My only question here is, given the speed and intelligence of AI tools, I don’t think the competitive patch for these tools is particularly defensible. I don’t see a strong moat here. This looks very straightforward for existing incumbents to replicate quite quickly. I’d love to be corrected and to see them succeed, but I think they’ll either specialise in a particular area or get swallowed up.
The disruptors
Then we look at the disruptors. I’ve picked a couple of examples here, they are not the only ones challenging, but examples of biting at the traditional SAM tools, the Flexera, USU, ServiceNow type market, in a pincer movement.
First there is Raynet. They have built a codex of licensing intel from working with Deloitte. They say they’ve packaged up the collective intelligence of 800 Deloitte consultants and are doing AI-based reconciliation and querying against that codex. You can have a natural conversation with this intelligence.
Fear of the Black Box Solution
The fear here, and we saw this in the early days of the ITAM market, is the black box. Okay, you’ve come to an answer, but how did you come to that answer? Show me your working, as your maths teacher used to say. This is especially true of AI. AIMultiple’s HALC-Bench benchmark of 37 LLMs found even the latest models hallucinate in more than 15% of answers when analysing provided material, with rates across the models tested ranging from 15% to 52%, and, as I mentioned on LinkedIn recently, AI speaks with particular conviction about topics it knows nothing about.
The encouraging flip side, per Vectara’s hallucination leaderboard, is that when top models are grounded in supplied source material, hallucination rates fall to around 1%, which is exactly the architectural argument for grounding the AI in your own data and a curated codex rather than letting it freewheel. So it’s important to show your working, and Raynet are keen to make sure this is not a black box experience. They explain their calculations.
AI Driven Catalogue Augmentation
What they’re also doing, and this is similar for Licenseware, is replacing the traditional approach of ITAM tools: a squadron of 40 people collecting evidence on assets around the world, looking at market reports, looking at new releases, looking at the data of existing customers, and bringing that intel into the catalogue. That was certainly the spine of Snow’s offering. We identified the Software Recognition Service as Snow License Manager’s key competitive differentiator in our reviews here on the ITAM Review going back to 2014. Now all of that can be driven by AI, a lot quicker than a room full of people can, and the people shift to being the overseers rather than the collectors.
Pepsi Challenge Proof Point Required
My take on this is that for these disruptors to really accelerate, we need some form of Pepsi challenge. I remember in the early days of ITAM, people used to say, why do I need an ITAM tool when I’ve got SCCM, or Intune as it is now, and you had to go and prove the difference. I think the industry would benefit from a Pepsi-type blind test of a conventional ITAM tool versus one of these AI-driven ones.
You can do this via proof of concept, but PoCs are a lengthy, expensive process for customers, and it would be a lot quicker to prove it in the market. I spoke to Onyx last year, a Microsoft licensing AI with a human in the loop to verify its decisions. Raynet says its codex prevents the need for that, but I think some further proof would really help their cause. I’m not doubting the quality of what they’ve built; I’m saying their marketing reach could accelerate with a Pepsi challenge proof point.
Licenseware are doing a very similar thing, but at a very accessible level. There’s a free tier, and pay-as-you-go starts at a dollar per device per month, the same sort of technology Raynet are offering at the enterprise level. So where Raynet might be argued to be hitting the incumbents at the top, Licenseware is biting them on the bottom. A real pincer movement on the conventional tools.
The smart bet
Then we get to the fourth type, and I’m calling this the smart bet, because it’s not table stakes and it goes beyond disruption. It’s looking at the future of where AI is going.
I really liked what Zylo are doing. They’re assuming that their own interface is not the be all and end all, and that in the future, or even now, the workflow is likely to be the customer’s own AI infrastructure pointing at Zylo as an expert. This needs MCP access.
For the layman: MCP (Model Context Protocol) is best thought of as a universal plug, a USB-C for AI, that lets your AI assistant connect directly to another system’s data and actions, so your assistant can ask Zylo questions and instruct it to do things, rather than you logging into Zylo yourself.
Rather than looking at another window, or building an integration, you’re connecting Zylo as a team member into your existing AI infrastructure, which might be Claude or ChatGPT, for example. All of your automations and processes happen there, and they use Zylo as an expert, perhaps an operator, doing some of the execution as a team member, rather than a separate pane you have to log into. It goes way beyond simply an integration. This is really pointing at where ITAM technology should be heading.
MCP freeing up the stranglehold of ServiceNow?
There was a conversation on LinkedIn around ITAM tooling recently, started by George Arezina (link to post), and my comment there sums up what I keep hearing from customers:
“The most common thing I hear from customers on your top 3 tooling is that ServiceNow is ‘good enough’. It seems Orgs are willing to sacrifice better quality ITAM in favour of ITAM being in the same IT management ecosystem and single stack as everything else. At that point, ServiceNow doesn’t have to be better, or even be pushed to be better, because it owns the rest of the stack. That’s a real shame and a key missing competitive element in the ITAM tools market.”
Now, if you’ve got an AI integration layer whereby ServiceNow is your plumbing, but another SAM tool such as Zylo is integrated at the AI level, that changes the game. We don’t need to settle for good enough; we could use something else. I would imagine ServiceNow will want to defend against that sort of AI-level integration, in the same way SAP defended against integration to its own data, which is where the Diageo fault line was established, because it encroaches on IP generated by their platform. So it might be that in future you have ServiceNow as the process plumbing, but a lot of the intelligence is done by different team members, such as a Zylo MCP, with humans as the overseers.
Where this leaves us
Interesting things going on here.
Four camps, then: table stakes being raised, aggregators racing for a moat, disruptors squeezing the incumbents from both ends, and one vendor betting its own interface won’t matter. It’ll be interesting to see how things pan out.
If you’re doing AI innovation in this space, whether you’re a technology vendor building it or a customer implementing it, I’m keen to talk to you. Give me a shout, happy to chat.
Further reading:
