Key Takeaways

  • Jake Stauch, co-founder and CEO of Serval, calls autonomous background agents the “ultimate moonshot in enterprise software” because full autonomy triggers immediate security blocks from corporate IT.
  • Unbounded code generation creates unacceptable operational liability: allowing an agent to write and execute live code across corporate infrastructure is a risk no enterprise CISO will take.
  • Serval counters ServiceNow by ditching raw, arbitrary execution in favor of deterministic scripts wrapped in software development lifecycle (SDLC) guardrails.
  • The path to enterprise autonomy requires incremental capability escalation, where an agent explicitly asks permission to run scoped scripts before earning broader access.
  • Serval runs lean product teams by eliminating dedicated product managers entirely, embedding product-focused engineers directly inside customer environments.

Enterprise Security Stops Autonomous Code at the Gate

Every enterprise tech buyer wants automation, but almost none will sign off on unsupervised agency. On No Priors, Sarah Guo and Jake Stauch examined the fantasy of self-directed background software. The pitch sounds clean: an agent monitors your stack, notices an issue, generates code, and pushes the fix.

In practice, enterprise IT teams slam the door on that workflow. Stauch explains the problem plainly: “The agent becomes pretty competent. They could theoretically just start writing code that could take action on your systems and that is an unbounded security risk that I don't think anyone's going to be able to take anytime soon.”

When software acts autonomously inside production infrastructure, liability compounds instantly. “On the one hand, you want these to be very autonomous without somebody baby sitting them,” Stauch notes. “On the other hand, autonomy is very scary in the enterprise.” A single unexpected loop or bad database write can bring down critical business operations. Even if an AI model achieves 99 percent accuracy, the remaining one percent makes IT leaders liable for outages and security breaches.

The Solution Is Incremental Capability Escalation

Serval is targeting ServiceNow by rethinking how enterprise workflows get updated. Instead of letting an agent write arbitrary code on the fly, Stauch relies on deterministic scripts and strict software development lifecycle controls.

Real deployment demands a staged permission model rather than a blank check. Stauch calls this an incremental escalation of capabilities. As Stauch describes the loop: “You go off and you do these things and it comes back and says, 'Hey, I would like to be able to do all this stuff. This would help me do my job.' And you approve them.”

Under this model, the agent acts like a junior engineer working behind strict access gates. It discovers routine inefficiencies, prepares a concrete script to handle the task, and asks human administrators to approve that specific capability. The system never executes arbitrary live instructions across production APIs. “And then I think the permissions problem will always remain the biggest thing,” Stauch notes, “because even if I feel like you are a very, very valuable, smart agent and I you can do all these things for me, how much do I trust you to just run wild and not have to babysit your actions?”

To make this feedback loop tight, Serval stripped out traditional product management layers. Instead of routing customer requests through PMs, Serval embeds product-minded software engineers directly with customer teams. The engineers see the permission hurdles and script failures firsthand, building the exact approval guardrails enterprise buyers demand.

What to Do With This

Audit your agent roadmap today and kill any feature that lets an LLM write and execute arbitrary code directly against customer production databases. Replace that open-ended execution with a catalog of deterministic scripts, then add a manual approval workflow where the agent must request permission before running a new script type.