Agents of Dev - AI, Agent & Agentic Development

Email your questions and topics to the podcast here. 

As AI agents move from experimental prompts to autonomous production workflows, developers are facing a brand-new set of infrastructure, governance, and safety challenges.

In this episode of Agents of Dev, Mitch Ashley and Brad Shimmin break down takeaways from Microsoft FabCon in Barcelona, highlighting the rise of autonomous database management agents and the industry-wide shift toward lightweight agent sandboxing.

The duo dives deep into Google’s Argon 4 release—evaluating its bold trade-off of raw benchmark scores to slash hallucination rates down to 15%—and unpack OpenAI’s Dev Days updates like ChatGPT Space and Dots. Finally, in "The Drop," Mitch explores NVIDIA’s hardware-enforced Open Agent Safety Platform, while Brad shares his hands-on testing with Jev, a ultra-fast "System 1" decisioning model redefining task routing.

This and more on Agents of Dev, part of The Futurum Media Podcast Network.

Email your questions and topics to the podcast here. 

Are bloated frontier models already becoming obsolete for everyday enterprise workflows?

In this episode of Agents of Dev, Mitch Ashley and Brad Shimmin break down fresh takeaways straight off the plane from Salesforce Dreamforce and the Neo4J Graph Summit.

They dive into Salesforce’s new COA model and headless architecture, how graph databases provide the critical "meaning, reach, and memory" needed for agentic AI, recent security headlines surrounding AI-assisted breaches, and why ultra-focused state decision models like JEV might change the economics of AI forever.

This and more on Agents of Dev, part of The Futurum Podcast Network.

Is AI heading toward a dystopian sci-fi nightmare, or are we just watching market hype mask the real architectural shifts happening in software engineering?

In this episode of Agents of Dev, Mitch Ashley and Brad Shimmin move past the polarities of AI doomerism vs. pure hype to tackle the pragmatic reality of agentic workflows.

They analyze Cloudera’s enterprise sovereignty partnership with Mistral, examine Atlassian’s push into agentic SDLC management, and take a deep dive into the world of developer harnesses—comparing setups like Pi, Aider, and Maki. Along the way, they unpack context tokenomics, cash ownership, and why managing LLM memory often feels like living out the movie Memento.

This and more on Agents of Dev, part of The Futurum Podcast Network.

Email your questions and topics to the podcast here. 

Are enterprise AI budgets running out of control while agentic models pretend to finish tasks they haven't actually completed?

In this episode of Agents of Dev, Mitch Ashley and Brad Shimmin tackle the biggest developments across cloud infrastructure, AI behavior engineering, and enterprise data strategy.

They start with AWS’s acquisition of Duck Labs and what embedding DuckDB directly into storage tiers means for the open-source ecosystem. Next, they unpack the issue of AI "phantom completeness" and how new tools like the Unlazy extension enforce behavioral discipline on agents.

The pair also analyze Brad’s newly published Data Intelligence Platforms Signal report, break down Bill Gates’ 6,000-word call for AI guardrails and job protections, and discuss the "fund AI anyway" spending habits shaping IT budgets in 2026.

This and more on Agents of Dev, part of The Futurum Podcast Network.

Unvetted AI code generation in your software pipeline is like handing a construction apprentice the keys to a wrecking crane—it might get the job done, but it could just as easily knock down the entire building.

In this episode of Agents of Dev, Mitch Ashley and Brad Shimmin sit down with open source pioneer Tracy Ragan (CEO of DeployHub and Ortilius community leader) to dissect the state of the modern software supply chain.

Tracy breaks down why hardcoded scripts continue to paralyze DevOps evolution, how global vulnerability collection laws are reshaping cybersecurity risk, and why "reachability" matters more than static SBOM scans.

Together, the team explores the inevitable shift from bloated frontier models to lean, domain-specific small language models, proving why human collaboration still forms the ultimate zero-trust defense.

This and more on Agents of Dev, part of The Futurum Group Podcast Network.