Rackhouse Founder Spotlight: Schuyler Brown and John Tokash of Guilde
The Next Era of AI-Driven Software
AI is fundamentally changing how software is built. Code that once took weeks now ships in days. Teams are deploying agents to write, refactor, and integrate faster than any human team could. But speed isn’t translating into impact. Most companies are still wrestling with the same friction points they faced before, only now they’re hitting them at ten times the velocity. The gap between what AI promises and what it delivers isn’t shrinking. It’s getting wider.
CTOs are under immense pressure to accelerate velocity or risk losing to emerging competitors. LinkedIn feeds are drowning in hyped examples of zero to one hundred million in ARR and boards are asking why they’re not seeing similar growth?
The problem isn’t effort, it’s misaligned execution. Companies are racing to adapt, investing heavily in AI tools and deploying agents across their workflows. But those efforts are undermined by deeper structural issues in how agents are integrated. Critical context still goes missing, workflows break under complexity, rules are treated as suggestions, and tests are skipped.
That’s the environment Schuyler Brown and John Tokash built Guilde for. Not to add another tool to the pile, but to solve the underlying issues that keep AI from delivering real business value.
Why AI Hasn’t Delivered ROI Yet
Everyone knows they need to invest in AI dev tools, but it’s not obvious which to buy, how to get the most out of them, and how to measure their impact. So they buy what their teams ask for: Cursor, Claude, Codex, Copilot, Gemini.
After buying these tools, Engineering teams usually experience a similar rollercoaster that starts with excitement, aggressive adoption, followed by migraine inducing frustration when the agents:
Write thousands of lines of code that at a glance looks like it should work, but doesn’t.
Rewrite functionality they has no business touching
Claim to pass all relevant tests when; in fact, they failed
That’s why studies like METR conclude perceived efficiencies aren’t translating into actual results: “even after experiencing the slowdown, [developers] still believed AI had sped them up by 20%.” Teams are left struggling to figure out what to do. They read about all these rumors of wild success, but they're not experiencing it themselves.
How Engineers Overcome AI Obstacles Today
In order to prevent these failure modes, Engineering teams usually try some or all of the following:
Lower ambition: By assigning tasks agents can confidently complete, you avoid wasted time troubleshooting failures. But the end result is incremental improvements instead of strategic wins for the business.
Greenfield projects: By exclusively working in new repos on new features, teams avoid the complexity of legacy code bases. This tends to relegate projects to prototypes instead of production-ready features.
The end result of tactics like these is that the team either accepts incrementalism or spends too much time reading Reddit in an attempt to stay up to date on the latest tactics. That’s a pretty big distraction from the daily deliverables they’re expected to achieve.
From AI Output to Business Impact
Guilde was created out of frustration with AI agents that over-promise and under-deliver. The founders lived this pain personally, which led to building tools to prevent the most common failure modes:
Too much tribal knowledge was missing from the context
Inefficient research wasted the context window
Rules were treated like suggestions
Tests were ignored
As they began sharing results, design partners started asking to adopt Guilde’s approach. This led to Guilde releasing an end-to-end agentic SDLC designed to prevent the most common failure modes at each step. Each agent is then given a focused scope, from clarifying requirements to retrieving context, writing code, and validating every rule until all tests pass. It’s a structured approach that transforms AI from a source of cleanup and rework into a reliable part of the development process.
The organic enthusiasm from their partners came through loud and clear during our diligence. Rackhouse views Guilde as foundational infrastructure for how engineering teams will build production-ready features and complex systems they can trust in an AI-driven world.
Why Rackhouse Invested
Rackhouse backs founders who take on unglamorous, high-friction problems. Problems that stay invisible until they cause a breakdown. Systems that enforce discipline in daily work rather than sitting unused in a slide deck. While other AI companies are focused on the attention grabbing stunts, Guilde is building the plumbing necessary to unleash agents’ full potential.
Developers are some of the most demanding customers you can have. As StrongDM’s co-founder and Chief Customer Officer, Schuyler Brown built a loyal customer base with dozens of repeat buyers who deployed across multiple jobs. While so many teams talk about being customer obsessed, it speaks volumes that former customers voted with their pocketbook. The CTOs of Alloy, Drata, Human Security, Troops, Abacus and more all personally invested in Guilde.
Guilde’s CTO John Tokash is also a repeat founder whose first startup, Homestead, was acquired by Intuit. He spent the past 4 years building at the bleeding edge of AI to launch Flowspace’s AI Command Center. Over too many spicy tacos, he and Brown bonded through their shared frustrations building AI agents.
Together, Brown and Tokash made a conscious decision to recruit former founders for all early hires in order to build a culture of accountability. As a result, every member of the team has a unique appreciation for both the technical and business obstacles to overcome and knows what it truly takes to find product/market fit. That's the kind of team Rackhouse wants to invest in.
The Next Test
Guilde’s beta is live, and the work ahead is proving it can deliver the same results in production at scale. Teams are under pressure to show real ROI from AI, yet most tools still fail because they amplify familiar problems: missing context, broken trust, and errors that spread faster than they can be fixed.
Guilde is built to close that gap by making AI agents a dependable part of engineering workflows. The next step is proving how well it fits into existing systems, prevents costly failures, and saves engineers meaningful time at scale.
The goal is not just to help teams ship faster. It is to prove that AI can deliver reliable, repeatable results in the systems that matter most.