Building Software Factories That Scale Without Breaking Trust: A DirectorPlus Gathering by Peter Bell and Gather.dev

A quiet tension is building inside engineering orgs right now. Not the kind that shows up in earnings calls, but the kind that creeps into every sprint review. Output is up. Confidence is not. Code is moving faster than judgment can keep pace, and somewhere between commit and deploy, trust starts to thin out. Many teams can generate. Fewer can guarantee. That gap is where systems either compound or quietly collapse.

That is exactly where DirectorPlus is placing its focus on April 29 in San Francisco. “DirectorPlus (SF): How to Build Your Software Factory” is not entertaining surface-level curiosity around AI-augmented development. It is addressing the harder operational question: how do you build something that holds under pressure? The framing is deliberate. Orchestration. Context. Verification. Clean up. Compounding through learning and dreaming. This is not about stacking tools. This is about designing a system that produces reliable output at scale.

The room reflects that level of intent. SoMa, early evening, a curated group of directors, VPs, CTOs, and platform and DevEx leaders from companies with 50 or more engineers. This is not an open door crowd. It is an invite-only Gather.dev environment shaped by Peter Bell and supported by Randy Shoup, Kumar Kandaswamy, and Sam McAfee. The common thread is accountability. Everyone in the room owns a system that has to work, not just impress.

The sequence of voices builds that narrative in layers. Peter Bell opens with what actually makes a software factory effective, grounding the conversation in patterns that survive contact with real teams. Page Bailey from Google DeepMind expands the lens with the evolution of agentic coding workflows, showing how the interaction between human and system is tightening. Chetan Conikee of Modiqo introduces the Ralph loop, forcing the room to confront the balance between autonomy and control inside agent-driven systems. Tim Sehn of DoltHub pushes into persistence, where systems are expected to retain context and state, not just produce bursts of output. Then Charity Majors of Honeycomb closes the loop with the infrastructure required to support all of it, because without strong DevOps foundations, none of this scales cleanly.

Around that structure, the real value forms in motion. Conversations at arrival, again midstream, and then late into the evening. Food and drink are there, but the real exchange is operational. What is working. What is failing. What breaks at scale. Supported by Chainguard and Vapi, but not driven by sponsorship energy. The design is intentional. Keep the signal high. Keep the room honest.

What is unfolding here is a shift from experimentation to expectation. Not whether these systems can work, but whether they can be trusted to keep working. The organizations that figure out how to build and scale a true software factory will not just increase output. They will reduce friction between idea and execution, and that is where leverage compounds.

Rooms like this rarely announce their importance in real time. But the patterns shaped inside them tend to show up everywhere else later, dressed up as standard practice.