English-language podcasts

The Enterprise AI Show

Build the Machine That Builds the Machine with Adam Jacob

51 min

Listen through the publisher, then discuss the episode here or arrange an in-person Roundtable. Your city determines where to meet.

Roundtables are designed for groups of four, with a shared subject and room for everyone to speak. Come for discussion, with no dating expectation.

Brandon interviews Adam Jacob, CEO and Co-Founder at Swamp Club, about building “the machine that builds the machine” and why AI should change how software gets made, not just speed up old workflows. Jacob, who created Chef and served as its CTO, revisits his Config Management Camp 2026 talk, AI Is the Future of Software Development. He says it won over a loyal core of fans and drew plenty of skeptics in a room full of operations engineers. He argues that dropping AI into an existing SDLC leaves engineers stuck reviewing a flood of generated code. The real shift, he says, is encoding your point of view, your architecture, and requirements like accessibility and security into adversarial agents and black box tests that run every time. When the output is wrong, he says, the instinct should flip from fixing the code to fixing the machine that produced it. He describes a “trough of awful” between the old way of working and the payoff on the other side, and insists that nobody is actually behind. Jacob explains that Swamp grew out of System Initiative, his previous company. The team won a proof of concept at one of the world’s largest banks and realized the agent driving the system had impressed the customer, not the six years of technology underneath it. Five former colleagues regrouped as “AI maximalists.” Their code name, Swamp, built from their initials, stuck after Config Management Camp attendees loved it. Swamp moves an agent’s control flow back into deterministic code. It breaks a request into reusable behaviors with typed inputs and outputs, then stitches them into workflows that call frontier models only when a step truly needs intelligence. Jacob says this cut one code review workflow’s token cost by roughly 10x, and that every run is stored as versioned, queryable data, which one user relies on for reproducible science. He positions Swamp beneath MCP, often replacing it. He also describes extensions, which he says number in the thousands, that cover the AWS, Google, and Azure APIs and let teams move the data store to S3 or Postgres. Jacob also shares how his team works. He says the three engineers working full time on the product average 35 to 45 pull requests a day, a pace new hires reach within a week. He credits that to the system handling architecture and review, which leaves people time to talk with each other and with users. He says that freedom produced Swamp Club’s gamified leaderboard and double-points accelerators, a feature he admits he would have rejected as a product person. To get started, he recommends running swamp start or adding “with Swamp” to any prompt in your coding agent, joining the Swamp Discord, and taking the change in small steps. He closes by plugging the Swamp Club podcast, a Monday news show with co-founders Paul Stack and Nick Stinemates. He argues that after 12 to 15 years of stasis in infrastructure, this is a genuinely fun time to build. SHOW: 1081 SHOW VIDEO: https://youtu.be/hfRo_dOIN3E SHOW SPONSORS: SHOW LINKS Swamp Club (https://swamp-club.com/) AI Is the Future of Software Development (Config Management Camp 2026) (https://www.youtube.com/watch?v=g1R71Wbxlkk) Swamp Club Discord (https://discord.gg/swamp-club) AI Frontiers Podcast (https://swamp-club.com/podcast) Swamp Ushers in the Second AI Frontier with Behavior-Centric Design (https://swamp-club.com/press/swamp-ushers-in-the-second-ai-frontier) GUEST BIO Adam Jacob is the CEO and Co-Founder of Swamp Club, which helps teams build AI systems out of reusable, deterministic behaviors instead of costly autonomous agents. Swamp Club raised a $7 million seed round led by Amplify Partners in October 2026. Jacob previously co-founded Chef Software, where he wrote the original Chef and served as CTO, and later co-founded System Initiative, which he led as CEO. Contact Adam Jacob: LinkedIn (https://www.linkedin.com/in/adamjacob/) Website (https://www.adamhjk.com/) FEEDBACK? Email: show @ the enterprise ai show dot com Twitter/X: @TheEntAIShow (https://x.com/TheEntAIShow) Instagram: @TheEntAIShow (http://instagram.com/TheEntAIShow)

Episode information is provided by the publisher. Listening takes place on the publisher’s service.

Listen at the source ↗

Featured places to meet in Palo Alto, CA

Explore these places and arrange a conversation through ThinkMatch.

City guide for Palo Alto, CA

More from The Enterprise AI Show

The Enterprise AI Show

AI News - Mid-November 2026

46 min

Brian and Brandon discuss the stunning resignation of NVIDIA CEO Jensen Huang, a story big enough to push election results, Anthropic’s IPO news, and a week of earnings off the agenda. They walk through what has been reported so far: relationships, including one with an executive tied to Megaspeed, a Southeast Asian cloud provider alleged to have funneled NVIDIA GPUs into China, and claims that security staff asked someone who photographed Huang to sign an NDA. Brandon cites Georgia coach Kirby Smart’s “dogs don’t beat dogs” to argue that CEOs have to follow their own company’s policies before anything else matters. To succeed interim CEO Colette Kress, Brian floats Tim Cook, Lisa Su, and Andy Jassy, while Brandon adds Lip-Bu Tan and Rene Haas and suggests NVIDIA may need a team of fixers to restructure Huang’s 60 direct reports. They debate whether extreme wealth breaks a leader’s judgment, and why success makes people afraid to call unusual practices strange. Brandon predicts NVIDIA

Discuss this episode

The Enterprise AI Show

AI’s Netflix Moment: What Early Cloud Teaches Us About the Next Phase

22 min

Brian and Brandon discuss what happens to enterprise AI if the market corrects and the era of "move fast and break things" gives way to hard questions about ROI. Brandon argues the biggest divide is between people who have only used a chatbot and those who have “touched the superpower” by using AI to finish a task that once took too long to attempt, and he says a CFO who only sees the token bills can’t see the value those tokens created. Brian traces enterprise adoption through several phases, from AI policies that mostly meant buying Microsoft Copilot, to employees pushing for Claude Code and Codex, to a token-maxing push with no clear goal. He compares the moment to early cloud, when Adrian Cockcroft’s Netflix talks split the industry into believers and companies that couldn’t imagine getting there, and predicts a bumpy year and a half in which some companies pull back, many search for balance, and a few race ahead. Brandon pushes back on the term vibe coding, arguing that nobody rea

Discuss this episode

The Enterprise AI Show

Claude, Make Me Rich: Finding Anthropic’s Next Giant Market

29 min

Brian and Brandon debate how a newly hired product leader at Anthropic would find growth big enough to match the expectations of an IPO, which Brian expects around the end of 2026. Brandon says he would start in the research group, hunting for breakthroughs that could serve markets like finance and healthcare, and raises contextual ads even though he jokes the idea could get him fired. Flipping Alex Karp’s warning about AI systems learning a company’s differentiation, Brian proposes a product that sits between employees and their data, then shows executives where to cut costs in the 70 to 80 percent of work he estimates doesn’t set them apart. Pointing to Sierra’s outcome-based pricing, Brandon suggests Anthropic take over a system, absorb the risk, and get paid only when it delivers. His AI-assisted car purchase leads to a discussion of opt-in, intent-based commerce, and Brian argues Anthropic still lacks the kind of aggregation point he sees as the biggest way to make money. They clo

Discuss this episode

Community Poll

Manufacturers should help cover harm from unsafe products even if older standards permitted them.

Share your view, then see how readers voted. No sign-in required. This poll is shared across stories.

Do you agree or disagree?

Choose one answer

Common Quandary

A Common Quandary to discuss

Another conversation to explore, separate from this news story.

A pricey AI tutor can turn B students into A students. Should schools let them use it?

An effective AI tutor adapts to each student and dramatically improves performance, but its subscription is expensive. A public school cannot provide it to everyone and worries that allowing its use will deepen existing inequality.

Should access be limited when a beneficial tool cannot yet be shared fairly?

Explore this quandary