Are You Ready for the AI Harness Wars?

Are You Ready for the AI Harness Wars?

Claude Code's recent quality issues sparked a broader conversation about transparency, reliability, and vendor lock-in across AI-powered developer tools. Brandon Mathis, Coston Perkins, and Jonathan Fontanez discuss Anthropic's explanation for the degradation, the challenges of relying on closed agent systems, and the risks organizations face when critical development workflows depend on tools they cannot fully inspect or control. The conversation also explores the rapidly evolving coding harness landscape, including Codex, OpenCode, Pi, Goose, Gemini CLI, and Quinn Code. Topics include benchmark performance, open source versus closed source approaches, customization, context engineering, long-running agents, and why many developers are beginning to view harnesses, not models, as the next major battleground in AI-assisted software development. The discussion examines what engineering teams should consider when evaluating AI tooling and why flexibility may become increasingly important as the ecosystem continues to evolve. What You'll Learn: - Why recent Claude Code quality issues pushed many developers to reevaluate their AI tooling choices. - How coding harnesses influence agent behavior, performance, and developer workflows beyond the underlying model. - The tradeoffs between open source and closed source AI coding tools, including transparency, customization, and vendor lock-in. - How tools like Codex, OpenCode, Pi, Goose, and Gemini CLI compare as the coding harness ecosystem rapidly evolves. - Why many developers believe the next major wave of AI innovation will come from harness design rather than model improvements alone. Brandon Mathis on Linkedin: https://www.linkedin.com/in/mathisbrandon/ [https://www.linkedin.com/in/mathisbrandon/] Coston Perkins on Linkedin: https://www.linkedin.com/in/costonperkins/ [https://www.linkedin.com/in/costonperkins/] Jonathan Fontanez on Linkedin: https://www.linkedin.com/in/jonathan-fontanez-27715428/ [https://www.linkedin.com/in/jonathan-fontanez-27715428/] This Dot Labs Twitter: https://x.com/ThisDotLabs [https://x.com/ThisDotLabs] This Dot Media Twitter: https://x.com/ThisDotMedia [https://x.com/ThisDotMedia] This Dot Labs Instagram: https://www.instagram.com/thisdotlabs/ [https://www.instagram.com/thisdotlabs/] This Dot Labs Facebook: https://www.facebook.com/thisdot/ [https://www.facebook.com/thisdot/] Sponsored by This Dot: https://ai.thisdot.co/ [ https://ai.thisdot.co/] AI Workshop Series from This Dot Labs: ⁠https://ai.thisdot.co/workshops⁠ [https://ai.thisdot.co/workshops⁠] Use Code THISDOTX at Checkout for $50 tickets!

Could Markdown Become the Next Programming Language? AI Agents, Claude Code, MCP & Agentic Workflows

In this episode of The Context Window, Brandon Mathis and Jonathan Fontanez explore a provocative question emerging in AI-assisted development: could Markdown become the next programming language? Using examples from agentic workflows, Claude Code skills, MCP integrations, and evolving AI harnesses, they unpack how structured Markdown files are increasingly being used to orchestrate repeatable workflows, prototype systems, and guide AI agents with surprisingly programmatic behavior. The conversation digs into the tradeoffs of treating Markdown like executable infrastructure, including maintainability, entropy, security risks, token costs, hallucinations, and the challenges of testing non-deterministic workflows. Brandon and Jonathan debate where Markdown-based “programming” fits compared to traditional software engineering, especially for short-lived automations, prototyping, internal tooling, and rapidly evolving startup environments. Along the way, they discuss concepts like skill.md files, harness maturity, vibe coding, structured prompting, evals, repeatable workflows, and how AI agents are reshaping the boundary between natural language and software development itself. In This Episode, You’ll Learn: - How AI agents are turning Markdown into a lightweight way to define repeatable development workflows - Why short-lived automations and fast-moving projects may benefit from “programming” with Markdown instead of traditional code - The risks involved with agentic workflows, including hallucinations, security concerns, token costs, and workflow entropy - How skills, MCPs, and AI harnesses work together to connect agents to external systems and automate real engineering tasks - A practical framework for deciding when to use prompts, Markdown workflows, or full software systems in AI-assisted development Brandon Mathis on Linkedin: https://www.linkedin.com/in/mathisbrandon/ [https://www.linkedin.com/in/mathisbrandon/] Jonathan Fontanez on Linkedin: https://www.linkedin.com/in/jonathan-fontanez-27715428/ [https://www.linkedin.com/in/jonathan-fontanez-27715428/] This Dot Labs Twitter: https://x.com/ThisDotLabs [https://x.com/ThisDotLabs] This Dot Media Twitter: https://x.com/ThisDotMedia [https://x.com/ThisDotMedia] This Dot Labs Instagram: https://www.instagram.com/thisdotlabs/ [https://www.instagram.com/thisdotlabs/] This Dot Labs Facebook: https://www.facebook.com/thisdot/ [ https://www.facebook.com/thisdot/] Sponsored by This Dot: https://ai.thisdot.co/ [https://ai.thisdot.co/] AI Workshop Series from This Dot Labs: https://ai.thisdot.co/workshops [https://ai.thisdot.co/workshops]Use Code THISDOTX at Checkout for $50 tickets!

Deterministic vs Non-Deterministic AI Workflows for Developers

In this episode of The Context Window, Brandon Mathis and Coston Perkins unpack one of the biggest shifts happening in AI-assisted development: when to let agents explore freely, and when to pull workflows back into deterministic, repeatable systems. Using real engineering examples like database migrations, CI pipelines, financial reporting, and code cleanup, they break down why relying entirely on non-deterministic agents can introduce risk, hallucinations, and hidden failures into critical workflows. The conversation explores a practical mindset for modern engineering teams: agents should build the tool, not become the tool. Brandon and Coston discuss how developers can use AI to generate scripts, workflows, dashboards, and automations that are inspectable, shareable, and reliable, instead of depending on one-off prompts and unpredictable outputs. Along the way, they dive into token efficiency, deterministic validation tooling, CI/CD automation, skills vs scripts, and the growing importance of accountability in AI-driven systems. In this episode, you will learn: - Why deterministic workflows are becoming critical as AI agents take on more engineering tasks - How to use AI agents to build reliable scripts, automations, and CI pipelines instead of relying on one-off prompts - The tradeoffs between non-deterministic agent behavior and repeatable engineering systems- How hallucinations, hidden failures, and inconsistent outputs can create risk in production environments - Practical ways to reduce token usage, improve reliability, and increase accountability in AI-assisted development Brandon Mathis on Linkedin: https://www.linkedin.com/in/mathisbrandon/ [https://www.linkedin.com/in/mathisbrandon/] Coston Perkins on Linkedin: https://www.linkedin.com/in/costonperkins/ [https://www.linkedin.com/in/costonperkins/] This Dot Labs Twitter: https://x.com/ThisDotLabs [https://x.com/ThisDotLabs] This Dot Media Twitter: https://x.com/ThisDotMedia [https://x.com/ThisDotMedia] This Dot Labs Instagram: https://www.instagram.com/thisdotlabs/ [https://www.instagram.com/thisdotlabs/] This Dot Labs Facebook: https://www.facebook.com/thisdot/ [https://www.facebook.com/thisdot/] Sponsored by This Dot: https://ai.thisdot.co/ [https://ai.thisdot.co/]

Building ChatGPT and Claude Apps

In this video, Brandon Mathis and Ben Lesh break down the emerging MCP Apps protocol and what it means for the future of AI-powered applications. Drawing from hands-on experimentation with tools like ChatGPT Apps, Claude, Codex, and MCP servers, they explore how developers can build interactive applications directly inside AI chat experiences using standardized protocols, iframe-based UI rendering, and tool integrations. The conversation walks through the technical architecture behind MCP Apps, including app tools, metadata configuration, input schemas, UI resource hosting, and how models decide when and how to invoke tools. Brandon and Ben also discuss the realities of building against rapidly evolving AI standards, covering everything from TypeScript and Zod validation to local development workflows with ngrok, caching issues, integration testing challenges, and content security policies.Along the way, they unpack larger themes shaping AI development right now: the growing importance of open standards, the tradeoffs of non-deterministic systems, security concerns around MCP tooling, and how AI interfaces may reshape the future of application development itself. The episode also explores the parallels between today’s AI tooling ecosystem and the early days of the web and mobile app platforms, including the risks, opportunities, and maintenance challenges developers should expect as these protocols mature. What You Will Learn: - How the MCP Apps protocol allows developers to build interactive applications directly inside AI chat platforms like ChatGPT and Claude - Why open standards are becoming important for creating AI tools that work across multiple ecosystems and models - The practical realities of building MCP Apps, including tool registration, UI hosting, schemas, caching, and local development workflows - The security and privacy risks involved with MCP tools and why developers need to carefully manage tool permissions and data exposureWhy testing AI-powered systems is more difficult than traditional software testing due to non-deterministic model behavior and evolving protocols Chapters 00:00 Introduction to MCP apps and AI chat integrations 04:00 How MCP apps work inside ChatGPT and Claude 06:40 Building MCP apps under the hood 13:20 Local development, ngrok, and testing workflows 16:40 Ideal use cases and limitations of MCP apps 20:20 MCP app marketplaces, approvals, and discovery 21:40 Security risks, trust, and data leakage concerns 24:40 Why MCP apps require ongoing maintenance 27:10 AI tooling and plugins for building MCP apps 28:45 Testing non-deterministic AI applications 32:20 The return of iframes in modern AI apps 35:00 Final thoughts on the future of MCP apps Brandon Mathis on Linkedin: https://www.linkedin.com/in/mathisbrandon/ [https://www.linkedin.com/in/mathisbrandon/] Ben Lesh on Linkedin: https://www.linkedin.com/in/blesh/ [https://www.linkedin.com/in/blesh/] This Dot Labs Twitter: https://x.com/ThisDotLabs [https://x.com/ThisDotLabs] This Dot Media Twitter: https://x.com/ThisDotMedia [ https://x.com/ThisDotMedia] This Dot Labs Instagram: https://www.instagram.com/thisdotlabs/ [https://www.instagram.com/thisdotlabs/] This Dot Labs Facebook: https://www.facebook.com/thisdot/ [https://www.facebook.com/thisdot/] Sponsored by This Dot: https://ai.thisdot.co/ [https://ai.thisdot.co/] AI Workshop Series from This Dot Labs: https://ai.thisdot.co/workshops [https://ai.thisdot.co/workshops] Use Code THISDOTX at Checkout for $50 tickets!

Is Claude Cowork the New OpenClaw? + Surge Pricing is Here?

Tracy Lee and Brandon Mathis break down the latest wave of AI news and what it means for how we work. They talk through Anthropic’s new Claude Cowork experience, the growing trend of agentic tools that can interact with files and workflows on your computer, Perplexity’s push toward AI-driven operating environments, and the bigger question of whether keyboards and traditional interfaces are starting to feel outdated. They also react to Anthropic’s new usage-limit experiment, discuss trust and security around AI tools that touch your machine, and close with a conversation about Moltbook’s Meta acquisition and what it says about the strange new social layer forming around AI agents.In this episode, you will learn:- AI is increasing the volume of ideas and work rather than eliminating engineering roles.- Claude Cowork represents a new layer where AI can directly interact with files and workflows on your computer.- There’s a clear difference between assistive AI tools and fully autonomous agents when it comes to trust and safety.- Typing and traditional computer interfaces are becoming a bottleneck compared to faster AI interactions.- The AI ecosystem is moving so fast that new tools and trends are emerging almost daily.Tracy Lee on Linkedin: https://www.linkedin.com/in/tracyslee/ [https://www.linkedin.com/in/tracyslee/]Brandon Mathis on Linkedin: https://www.linkedin.com/in/mathisbrandon/ [https://www.linkedin.com/in/mathisbrandon/]This Dot Labs Twitter: https://x.com/ThisDotLabs [https://x.com/ThisDotLabs]This Dot Media Twitter: https://x.com/ThisDotMedia [https://x.com/ThisDotMedia]This Dot Labs Instagram: https://www.instagram.com/thisdotlabs/ [https://www.instagram.com/thisdotlabs/]This Dot Labs Facebook: https://www.facebook.com/thisdot/ [https://www.facebook.com/thisdot/]Sponsored by This Dot Labs: https://ai.thisdot.co/ [https://ai.thisdot.co/]