Key Points
- Ed Sim explains the battle is between headless agents that read and write systems of record and the agent platforms those same vendors are building.
- Nobody rips out Salesforce, Sim says, but every renewal becomes a moment to consolidate systems or hand the work to agents.
The most valuable real estate in enterprise software has long been the screen an employee stares at all day. On CXOTalk episode 928, Ed Sim, founder and general partner of boldstart ventures, argued that AI agents are starting to contest it. Sim funds companies at inception, which forces him to look past the current market: "if something is already a problem, it's probably too late. We need to see the problems of tomorrow."
The question underneath SaaS
An audience question during the live show asked how agentic workflow companies would disrupt the SaaS model. Sim split the category before answering. Application companies such as Salesforce, HubSpot, and ServiceNow are one group. Subscription infrastructure such as Datadog, Snowflake, and Databricks is another. The disruption question lands on the first group, and he stated its stakes as a question: "will we have an interface two or three years from now?" His aside about ServiceNow and Salesforce arrived with the apology attached: "That software is kind of clunky and ugly to use. Sorry to say."
Headless agents versus the platforms
What replaces the screen is not a better screen. In his telling, "the real battle is going to be around headless systems." On one side sits automation that treats today's applications as databases: "I'm going to have agents kind of interacting and pulling the data that I need and writing the data into some of these systems of record". On the other side, Salesforce is trying to build Agentforce, and ServiceNow has its Control Tower.
The shape of enterprise work is changing too. Sim said the founders coming to him are asking "what happens when the world has long-running agents that not only run for a minute or an hour, but for days or weeks or months?" Software designed around a person at a keyboard was never built for that.
The new toll: paying for access
The incumbents are not waiting to find out who wins. Sim pointed to access itself becoming a product: "there are some companies like SAP, I think, that if you're not approved, you're going to have to pay." He said Salesforce changed its business model about six months earlier, and he summarized the new posture bluntly: "hey, have at it, agents, have at my platform. You're just going to have to pay per access."
That turns agent traffic from a threat into a revenue line, and in his view the repricing of machine access has only begun.
Why nobody rips Salesforce out
Host Michael Krigsman observed that it now sounds odd to describe Salesforce as legacy software, given how much both the technology and user expectations have changed. Sim declined the easy conclusion: "I don't think anyone's going to rip it out because I don't think people understand kind of that layer between the user and Salesforce. There's so much customization built into it, so many rules built into it."
He allowed for erosion at the margins, saying "Not to say that AI can't write some of that up or people can eat around at the wedges". The real pressure arrives somewhere else, at contract time. He described the moment: "every time there's a new renewal, people are going to say, hey, I can take these, you know, one, two, or three different systems together or agents doing work". That squeeze on the vendor keeps tightening, he argued, "unless they can find this new revenue curve from agents accessing it versus humans."
The founder signal
Sim's deal flow gives him an early signal. On new systems-of-record businesses, the pipeline has gone quiet: "no one's coming into us saying, I'm going to start one of these businesses." He named one exception: "unless they want to just destroy it completely from scratch and reimagine and reinvent it from an agent-first perspective."
The clearest picture he gave of agent-first came from acquisition diligence: buyers want teams building for the future, where perhaps "80% or 90% of the pull requests are happening through agents". The wider AI market is a different story, crowded to the point that in his estimate "every category that exists today probably has 30 or 40 companies kind of trying to do something". It cuts both ways: "starting a company has never been easier, but I think building a durable company has never been harder."
The ten-year question
Sim's largest frame is a question he refuses to answer with confidence: "do these systems of record continue to exist 10 years from now or not?" He sketched the alternative: agents that do the read-write "and maybe write a different database on the backend" until, over time, "you have your own system of record". Then he admitted he does not know. For buyers, the practical version of his question arrives sooner, at the next renewal.
Watch the full conversation with Ed Sim and read the complete transcript on the episode page.
CXOTalk prepared this article with AI assistance from the verbatim transcript of episode 928. Quotations are unedited from the broadcast.
