Software asset management has always been about reading contracts, comparing them with what is installed and used, and deciding what to pay for. That is document-heavy, rule-heavy work, so it is an obvious target for Gen AI. It also now has a second job that did not exist three years ago: tracking and governing the AI software and AI consumption that employees adopt faster than procurement can respond.
This article reviews what SAM and SaaS management vendors have shipped and announced as of October 2026, based on release notes, product pages and surveys. We label status carefully, flag vendor-sourced numbers, and say where we found nothing. We have already written about how Gen AI and agentic AI agents apply to software asset management in general terms. This piece is the tool-by-tool companion. For the platforms themselves, see our guide to SAM tools.
What AI can and cannot do for SAM
AI helps most where unstructured documents meet structured data: reading an agreement or an entitlement statement and turning it into fields a license model can use. It can also summarize usage, draft a renewal brief and answer questions about spend. It cannot decide what a contract means in a dispute, and it cannot make an effective license position defensible to an auditor on its own. For that you still need a person who understands the metric, the agreement and the product use rights.
We found no independent accuracy benchmark for AI contract extraction. Vendors describe the capability, and customers describe time saved, but nobody we found has published error rates on a common set of agreements. Until someone does, treat extraction as a draft that needs review.
The tools, one by one
Flexera
Flexera has the most specific shipped detail. AI entitlement ingestion became generally available in February 2026 and AI contract ingestion in April 2026. Both turn documents into records in the license model. Flexera announced AI Cost Management on 10 June 2026, which was in early access when we last checked, and an AI sustainability dashboard became available in September 2026. Flexera also publishes the annual State of ITAM survey that we cite below. It is vendor-run, which does not make it wrong, but readers should know who asked the questions.
ServiceNow
ServiceNow’s Now Assist for Software Asset Management was announced with agentic workflows in the Zurich release, and the Australia release added contract entitlement extraction. We did not find a general availability date for either, so confirm status before planning around them. ServiceNow’s AI Control Tower, which attributes AI costs, appeared in August and September 2026 release notes.
Zylo
Zylo announced Contract Assist Agent on 22 January 2026 and said Clarity AI with an MCP server would be generally available by 6 October 2026. Zylo’s focus is SaaS spend. Its 2026 index, which is vendor data, reports 36 percent of licenses unused and 78 percent of organizations hit with unexpected charges. Productiv, a competitor, shut down on 6 August 2026 according to competitor-sourced reports, which is a reminder that vendor viability belongs on your risk list.
USU, Certero, BMC, Ivanti and ManageEngine
We found AI claims from USU but no specifics we could verify. We found no SAM-specific AI features from Certero, BMC, Ivanti or ManageEngine in the sources we reviewed, though Certero and ManageEngine have assistants elsewhere in their products. Absence of evidence is not proof, so ask them directly.
The AI vendors themselves
The software you are managing now includes AI products with new billing models. Anthropic’s Claude Enterprise bills usage at API rates with spend limits, OpenAI introduced spend controls for ChatGPT Enterprise on 18 June 2026, and Microsoft sells Copilot Credits and Agent 365. None of these look like seat licenses. A SAM team that only counts seats will miss most of the cost.
The shipped features cluster around document ingestion. Agents that act on licenses, for example by reclaiming an unused seat, appear only as announced capabilities. That matches how cautious software owners are about automating a change that can trigger a true-up.
Use cases in detail
Entitlements and contracts into the license model
Most SAM teams spend a large share of their time keying in what they bought. Extraction from contracts, order forms and entitlement statements is where AI saves hours first. The effective license position still depends on interpretation: which metric, which version rights, which affiliates. Use the output as a draft and sample it against source documents before it feeds a compliance decision.
Audit readiness
Audits are where weak records cost money. Flexera’s 2026 survey of 512 respondents found that 48 percent had been audited in the past year, and 44 percent had spent more than $1 million on audits over three years. Among audited organizations, Microsoft audited 64 percent, Oracle 38 percent, up from 24 percent, and Adobe 32 percent, up from 24 percent. Oracle in particular has audit tooling caveats that one consultancy documents well, so do not assume a tool output equals an accepted position. AI can help assemble evidence faster. It does not change the standard of proof.
SaaS and shadow AI discovery
The same survey found that only 31 percent have visibility into AI software, 45 percent have identified unsanctioned AI, and 84 percent name tracking AI a top challenge. Only 32 percent track AI application licenses in their CMDB, which connects this topic to our article on AI in CMDB tools. Discovery of AI tools is partly a SaaS management problem and partly a network and identity one, so expect to combine sources.
AI spend governance
Per-user subscriptions are giving way to credits, tokens and usage-based billing. Flexera found that 59 percent say wasted AI spend has risen, and only 29 percent measure the value of AI software. Gartner said on 15 September 2026 that 60 percent of Global 500 companies will embed AI FinOps controls at inference by 2028. That is a forecast, not a measurement, but it points at the control you will be asked for: spend limits, owners and review.
Renewals and platform changes
Broadcom’s changes to VMware licensing have pushed many organizations to evaluate alternatives. Gartner predicts that 55 percent of VMware users will be investigating an exit by 2029, from 25 percent in 2026. A Rimini Street survey puts the number higher, at 90 percent considering alternatives, but Rimini sells support for such moves and has an interest. We found no 2026 primary data on Broadcom audits. For any large renewal, use AI to prepare the brief and the usage evidence, and keep negotiation with people.
The benefits, and what is actually evidenced
Vendor figures such as Zylo’s 36 percent unused licenses are real data from real customers, but they come from the vendor’s own customer base, which skews toward organizations already trying to find waste. The independent evidence supports the problem more than the cure. Flexera’s State of the Cloud 2026 report put estimated cloud waste at 29 percent. Gartner said on 1 July 2026 that $234 billion of enterprise software spend is exposed to agentic arbitrage by 2030, which means software that AI agents can replace or reprice. Both numbers describe the size of the prize. We found no study that measures how much of it AI-assisted SAM tools actually capture.
How it is evolving
Documents first, then decisions. The first wave extracts data from contracts and entitlements. The next wave, visible in ServiceNow’s agentic workflows and Zylo’s MCP server, lets agents ask questions of the license position and prepare actions.
The product to manage is changing. A seat, a token and a credit are different units, and tools are adding AI cost management because the old models do not fit. Expect the line between SAM, SaaS management and FinOps to blur.
Consolidation and failure. Productiv’s closure, Freshworks’ purchase of Device42 and Atlassian’s partnerships show a market in motion. Ask about data export and continuity.
Analyst caution. Gartner said in June 2025 that over 40 percent of agentic AI projects will be canceled by the end of 2027, and on 29 September 2026 that 70 percent of vendor-built agentic AI projects would be abandoned by 2028. Plan for pilots that end, and for what you keep when they do.
Risks and prerequisites
Wrong extraction. A misread metric or end date can overstate or understate your position. Sample and review.
Data flow and confidentiality. Contracts contain pricing. Ask where documents are processed, retained and used for training.
Autonomy near a true-up. An agent that reclaims or reassigns licenses can alter the audit position. Require approval.
Status and edition. Several features are early access or announced. Get availability and edition in writing.
Usage-based AI costs. Agents consume credits too. Budget for them.
A readiness checklist before you pilot
- Do you have an inventory of your top vendors, contracts and renewal dates in one place?
- Is there a named owner for each major software publisher?
- Do installation and usage data cover cloud, SaaS and AI tools?
- Can you tell which AI tools are sanctioned and who pays for them?
- Have you decided which AI-drafted entries a person must review?
- Do you have a baseline for time spent on renewals, audits and entitlement entry?
Where a maturity assessment fits
A software asset management maturity assessment tests whether your contracts, entitlements, discovery and governance can support an AI layer. Start with the Software Asset Management maturity assessment, see all assessment categories, or read our guides to enterprise software license management, SAM challenges and fixes and the return on AI in SAM. For the hardware side, see AI in hardware asset management tools.
Frequently asked questions
Can AI read my software contracts accurately?
It can draft the fields quickly. We found no independent accuracy benchmark, so review a sample and every high-value agreement.
Which SAM tool has the most shipped AI?
Flexera has the most documented shipped features, with entitlement and contract ingestion. ServiceNow and Zylo have shipped or announced others. Several capabilities are early access.
How do we track AI tools and their spend?
Combine SaaS discovery, identity and network data, and billing exports. Assign owners and spend limits. Plan for credits and tokens, not just seats.
Does AI reduce audit risk?
It can help gather evidence faster. It does not replace correct entitlements, accurate installation data and an interpretation of the contract that you can defend.
Check your software asset foundation before you add AI
Take the Desqcon online maturity assessment for software asset management and see how ready your contracts, entitlements and discovery are. Or talk to a senior consultant about an audit-readiness review.
Take the SAM Maturity AssessmentChoose an Assessment CategoryTalk to a Senior Consultant
Sources and further reading
- Flexera, State of ITAM 2026 (512 respondents) and What’s New releases, February, April, June and September 2026
- Flexera, State of the Cloud 2026
- ServiceNow, Now Assist for Software Asset Management, Zurich and Australia release notes; AI Control Tower release notes, August and September 2026
- Zylo, Contract Assist Agent announcement, 22 January 2026; Clarity AI and MCP Server availability; Zylo 2026 SaaS Management Index
- Anthropic, OpenAI and Microsoft pricing and admin documentation for enterprise AI spend controls, June 2026
- Gartner press releases: 25 June 2025, 1 July 2026, 15 September 2026, 29 September 2026; Gartner forecast on VMware exits
- Rimini Street survey on Broadcom and VMware (sponsor has an interest)
- Consultancy analysis of Oracle LMS audit tooling
