About:

Joseph E. Gonzalez is a leader in AI, co-founding RunLLM and researching at UC Berkeley.

Website:

Specializations:

Interests:

AI technology Open-source LLMs Technical support
Subscribe to RSS:
The post argues that the 'hard-hard' quadrant of AI applications, characterized by complexity and friction, offers the most sustainable competitive advantages through accumulated experience and context-specific solutions.
Selecting AI agents should prioritize their adaptability to specific workflows over traditional benchmarking scores, which often fail to reflect real-world performance.
Cursor's success stems from its user-friendly UX that emphasizes small, manageable tasks, enabling effective feedback and trust-building in AI applications.
Establishing a data moat is crucial for AI agents, requiring effective data gathering strategies to enhance agent quality and maintain a competitive edge.
AI agents have made uneven progress across domains, with ease of adoption and data acquisition being crucial for rapid advancements, particularly in coding applications.
The post forecasts significant trends in AI for 2026, focusing on data center investments, acquisitions, and the shift towards solving complex application-layer problems.
The post discusses the changing dynamics of token demand in the inference economy, particularly for application builders using LLMs (Large Language Models). It highlights the increase in token consumption per request due to the ne...
The post discusses the advantages of domain-specific AI products over general-purpose platforms, emphasizing that increasing specificity in AI development leads to greater value and return on investment. It highlights the importan...
The post discusses predictions about the AI market over the next 1-2 years, emphasizing the need for enterprises to derive value from AI applications, primarily through cost-cutting and automation. It outlines a theory that the AI...
The post discusses two archetypes in AI application development: AI Artists, who give LLMs full creative control, and AI Engineers, who operate under constraints to optimize for quality. It highlights the strengths and weaknesses ...
Embracing an abundance mindset in UX design encourages rapid experimentation and iteration, enhancing product development and user experience.
Effective AI applications must prioritize user experience by integrating into existing workflows and fostering trust through transparency and control over their operations.
The post discusses the current trends in data center buildouts and their implications for application-level businesses, particularly in the context of inference economics. It highlights the decreasing costs of intelligence over th...
The post discusses the concept of knowledge debt, which arises when new best practices in software development are not applied to existing codebases, leading to technical debt. It explores how this idea can extend beyond software ...
The post discusses the challenges of differentiating AI applications, emphasizing that the key to success lies in the quality of data used rather than solely relying on LLMs (Large Language Models). It argues that as LLMs become c...
User expectations differ significantly between coding agents and AI SREs, highlighting the need for better product design and interaction to foster effective user engagement.
The post discusses the importance of building opinionated products and agents in the startup and AI landscape. It emphasizes that having strong opinions helps differentiate products in a crowded market and guides user experience. ...
The blog post marks the 100th entry on the AI Frontier, reflecting on the lessons learned over two years. It discusses the slow but growing enterprise adoption of AI, highlighting a Fortune 500 company's lengthy proof of concept w...
The post reflects on the blog's journey over the past two years, highlighting key lessons learned from 99 previous posts about AI. It discusses the impact of OpenAI's cost strategies, the challenges of token volume in LLM applicat...
The post reviews AI developments in 2025, grading predictions, highlighting popular content, and emphasizing a shift from model focus to application relevance.
The blog post discusses the concept of forward-deployed engineering in the context of AI startups, explaining its origins in military operations and its adoption in the tech industry, particularly by companies like Palantir. It hi...
AI has settled into a regular technology trend in 2025, creating value without drastically altering society, akin to previous platform shifts.
The post discusses OpenAI's shift from a developer-focused approach to a consumer-oriented model, particularly highlighted by the recent announcements at DevDay. It emphasizes the introduction of consumer features in ChatGPT, such...
The post discusses the detrimental effects of the 9-9-6 work culture in AI startups, advocating for realistic work expectations and the importance of work-life balance. It emphasizes that while hard work is necessary, glorifying e...