What software survives AI?
Which software businesses retain pricing power when agents can build applications and perform the work?
What’s at stake
Software faces two different tests: whether customers still need the product, and whether its future profits justify the price investors paid. Agents threaten some paid tasks directly. They can also make established systems more useful by connecting their data to new workflows. The durable business may look less like an app people open and more like infrastructure other software calls.
The ideas that connect
A falling valuation is not the same as disappearing revenue
The first pressure often lands on expectations. Investors who once assumed decades of steady cash flow may accept a much shorter horizon when AI makes replacement plausible. That explains why a vendor can meet its current targets while its valuation falls. Salesforce’s scale and data give it room to adapt, but those advantages do not automatically restore the old certainty about future profits.
Supporting passages
2026-02-07
Look at this, revenue is actually stable to increasing for software companies, revenue growth. They're going down because we're discounting that future uncertainty. When something as profound as AI comes along, all of a sudden, it causes you to question whether or not there's as much certainty and durability in those future free cash flows. In the case of, take Salesforce, it's gone from 30 times free cash flow multiple to 15 times. That means somebody buying it today says, listen, I think 15 years into the future, I can count on these free cash flows. Before, they were willing to pay 30 years into the future. Well, hell, with AI today, we don't know what's going to happen seven years into the future. For people at home to understand why are these companies hitting their numbers but their stocks are going down, they're two totally different things, right?
Open the source and seek to the timestamp above.
2026-04-24
And I think they're pretty unique. It may be that of all the businesses in the scape, like the ones that have that scale, that have that multi-product platform, that have a lot of your data, there's a lot of opportunity for them to maneuver their way into an evolution of the business.
Benioff's the first one. And because if you look at it and you compare it, for example, to other companies, I think the workday response was to say, you can't have an AI interact with us without paying some kind of toll.
You're exactly right.
Whereas, Benioff's is the exact opposite, which is he's like, okay, we're going to go headless for the whole thing, which is brilliant.
Open the source and seek to the timestamp above.
Separate the analysis from the system of record
An agent that analyzes existing data can weaken the case for buying another analytical module. Replacing the authoritative record is harder: it also means rebuilding integrations, permissions and operational rules. Arora’s distinction and Salesforce’s headless strategy point to the same boundary. The interface and analysis can move elsewhere while a vendor keeps providing the underlying data and actions.
Supporting passages
2026-06-08
Somebody that says, I'm going to collect a lot of data for you and analyze it for you. I don't need you to analyze it for me. I can run models against data and analyze them myself. So if you think about, every SaaS company has a marketplace. You can buy Salesforce marketplace. What does it say? You have Salesforce data. I'm a marketplace app. Take me and I'll help you analyze the data. I don't need you.
You don't need that.
I can just go run an LLM against the data. So the entire incrementality that has been sold as incremental software modules to all of us, doesn't need to be sold to us because I'd much rather have LLMs run against that data.
Open the source and seek to the timestamp above.
2026-08-29
More importantly, I think this is really the critical thing. He is integrating Salesforce into Claude. He's willing to allow Claude to be the front end, to have the sort of the primary user relationship. And Claude is now getting able to access all of the data in Salesforce, and all of the workflows, all the actions that a user can take. So in other words, Benioff here is willing to risk disintermediation in order to ride this AI wave, because he knows it's what the customer ultimately wants. You know, you don't want to go into Salesforce or any SaaS application to write your agents. You want to create agents in the AI program that you're using. And you want those agents to be able to run across all of your SaaS platforms.
Open the source and seek to the timestamp above.
Letting agents in may be the defensive move
Vendors can charge a toll for outside agents or make their systems easy to use through them. The first protects an existing revenue model; the second accepts less control over the user interface to remain part of the workflow. Salesforce illustrates that strategic choice. The unresolved commercial test is whether serving more automated work compensates for any loss of seats, add-ons or direct customer attention.
Supporting passages
2026-04-24
And I think they're pretty unique. It may be that of all the businesses in the scape, like the ones that have that scale, that have that multi-product platform, that have a lot of your data, there's a lot of opportunity for them to maneuver their way into an evolution of the business.
Benioff's the first one. And because if you look at it and you compare it, for example, to other companies, I think the workday response was to say, you can't have an AI interact with us without paying some kind of toll.
You're exactly right.
Whereas, Benioff's is the exact opposite, which is he's like, okay, we're going to go headless for the whole thing, which is brilliant.
Open the source and seek to the timestamp above.
2026-08-29
More importantly, I think this is really the critical thing. He is integrating Salesforce into Claude. He's willing to allow Claude to be the front end, to have the sort of the primary user relationship. And Claude is now getting able to access all of the data in Salesforce, and all of the workflows, all the actions that a user can take. So in other words, Benioff here is willing to risk disintermediation in order to ride this AI wave, because he knows it's what the customer ultimately wants. You know, you don't want to go into Salesforce or any SaaS application to write your agents. You want to create agents in the AI program that you're using. And you want those agents to be able to run across all of your SaaS platforms.
Open the source and seek to the timestamp above.
What remains unresolved
Watch which products customers actually remove, what replaces their revenue and how vendors charge for agent access. A cheaper interface, a lower stock price and a completed migration are three different outcomes.
A selected guide · September 2025–September 4, 2026 · Sources and method