The usual view of artificial intelligence in software is that it will automate office tasks, leading to a decrease in the number of employee licenses required, and as a result, having an impact on subscription software. Even though this idea seems simple, it may not give the full picture.
The true question isn’t whether AI will result in a reduction in the number of positions; instead, it is whether software companies can shift from making money on the basis of the number of people they have to making money on the amount of work they complete. It is this difference that will decide which of the next generation of companies will be successful and which will merely add AI features to a shrinking customer base.
A clear example of this can be seen in Workday’s second quarter, with total revenue increasing by 12.8 percent to $2.649 billion and subscription revenue going up by 13.9 percent to $2.471 billion. More than 25 percent of the new annual contract value was due to AI, and over 5,500 customers used at least one agent developed by Workday, which is more than 35% higher than the number in the previous quarter.
This is not a case of a pilot program in the theatre sector; it is evidence that customers will pay for AI when it is part of a trusted system of record. At the same time, the fact that the transition should not be regarded as a success is still clear because the total subscription backlog increased by only 8% and free cash flow fell from $588 million the previous year to $460 million, even though the non-GAAP operating margin climbed to 31.1%. Genuine adoption has taken place. The connection between adoption and steady, incremental cash flow is still being established.
The old equation
For two decades, enterprise SaaS had a beautifully simple revenue engine:
OLD MODEL Revenue ≈ licensed employees × price per seat × renewal rate
The model was successful because employment, digitization, and software density all tended to increase together; as a company grew, it hired more people, the larger number of people resulted in greater demand for access, and that increased demand for access in turn brought about more recurring revenue, while moderate price increases were amplified by the rising number of seats.
Agentic AI changes the position; for example, if a finance team can carry out its duties with fewer manual steps, or if the HR department can handle cases with fewer staff members, customers get more value even though the number of employee seats falls. A software company might improve its product but still end up with fewer seats to bill.
The fact that is hidden in most optimistic AI demonstrations is that productivity should at first be the customer’s, with the vendor being able to reap it only if the prices are altered.
The new equation
The new model makes use of a number of different approaches: it features a core platform subscription, which gives customers access to data, permissions, and controls; customers buy credits for agent activity; and premium modules enable businesses to charge their customers for high-value workflows. The revenue formula is now as outlined below:
NEW MODEL Revenue ≈ platform access + agent consumption + completed high-value workflows
That is precisely why Workday’s example is relevant not only to Workday but also to the fields of human resources and finance. In these areas, strict controls are necessary; with regard to payroll, audits, hiring, benefits, and financial planning, an AI agent should not be permitted to take action unless it has clear authorizations, a record of its actions, or some means of accountability. Since the system that maintains the official records both holds the data and is responsible for the rules governing how things are carried out, it has the responsibility.
Workday calls them “deterministic rails”. Although this is a marketing phrase, the concept is valid: an agent that can access reliable data, adhere to its roles, and leave a clear audit trail is more useful than a general AI that only provides an approximate answer. In the context of enterprise AI, the possibility of an agent acting might turn out to be more important than its intelligence.
This provides established companies with a means of coping with the move to AI. They are not required to create the most excellent AI model; instead, it is enough for them to manage the workflow, keep a record of the work carried out, and set prices in such a way that customers do not feel as though they are being penalized for becoming more productive.
Why the quarter is encouraging—but not a victory lap
Three signals are constructive.
AI is already having a real effect on bookings, since over a quarter of the new annual contract value is substantial enough to affect growth, not just the product story.
More and more customers are turning to AI, and given that the company has over 5,500 agent customers, it seems that it is successfully getting its existing users to adopt the technology.
The profit margins of the company are rising even though it is still making investments, which means that Workday can now review its pricing without needing to look for external funding.
The backlog as a whole rose by 8 percent, a rate which is slower than that of recent subscription growth. This could be due to contract timing, but it does make it more difficult for investors to confidently predict future growth.
Last year’s free cash flow was greater than it had been the year before. Even though one quarter by itself does not provide the full picture, it does show that improving profit margins cannot in any way replace a careful analysis of cash flow.
The overall number of seats is still at risk since if the AI enables customers to reduce the administrative staff faster than new revenue rises, the new product could end up harming the existing one.
After the results, the share price of Workday was about $204.72, which meant that the company had a market value of approximately $50.4 billion and a price-earnings ratio of about 41. Although this valuation does not demand perfect performance, it does require the company to achieve steady growth, not just satisfy its existing customers.
The five-metric scorecard
Investors should stop focusing on whether a SaaS company has AI since nearly all of them do. The relevant questions are the ones that can be measured.
What percentage of the new ACV can be attributed to AI? Is AI acting as a major source of extra bookings, or is it being used to help secure renewals?
Should the speed of offering customers the option to sit down in relation to seat allocation be greater than the increase in agent usage when automation is introduced?
Should the offer to customers to switch to a lower seat count be faster than the amount of agent usage increases due to automation?
Can the existing backlog produce revenue while the total backlog and the renewal period continue to be in good standing?
Will the bookings generated by AI eventually result in an improvement in free cash flow once allowances have been made for infrastructure, R&D, acquisitions, and stock compensation?
The scorecard can be used by a large number of companies; it applies to Salesforce, ServiceNow, SAP, Oracle, Adobe, and, in fact, to almost any software company that says it is aiming towards an agent-based future. Is there anything that goes against the thesis?
The optimistic situation fails if the annual contract value rises simply as a result of discounts, bundling, or being included in current agreements; if customers use the agents but at the same time reduce the number of human seats at an even faster rate; or if usage is so irregular that the procurement teams choose not to proceed with it. Moreover, the scenario collapses when general-purpose platforms take over the workflow and leave the existing systems as slowly growing databases.
The clearest sign that the initiative has failed would be for the number of adoptions to increase while the company’s overall backlog, customer retention, and cash flow fail to show any improvement. Although the product might seem better, the business could still be worsening. Investors should make sure that both of these factors improve simultaneously.
The investable conclusion
It is not likely that AI will cause enterprise software to disappear; on the contrary, it will most probably change the way in which contracts operate. Businesses whose sales are based on the number of employees will see their market decrease. Yet those firms that are able to charge for managed services while at the same time keeping trust, data control, and customer value can make use of automation as a new source of revenue.
The fact is that there is demand, as Workday has shown, but the company has not yet demonstrated that its new method of measuring value can completely replace the old one. This is why Workday is of interest at the moment: the issue is no longer whether AI will be used, but rather who will gain the most from the productivity it brings.
The seat won’t vanish completely at once, but its impact on how software is valued is already decreasing.
MoatPeak Team | moatpeak.com