The Future of Software Development: An Interview with Boris Cherny

The landscape of software development is undergoing a seismic shift, driven by the rapid advancements in AI. Boris Cherny, the creator of Claude Code, sat down with Lauren Reader to discuss this evolution, his personal workflow, and his predictions for the future of teams and software products.

Claude Code: From Accidental Project to Essential Tool

Cherny's journey with Claude Code began somewhat serendipitously. He joined Anthropic Labs in late 2024, an incubator team focused on innovation. During this time, his team developed Claude Code, MCP, and the desktop app. The initial motivation was to move beyond simple "type ahead" code completion, which was the state-of-the-art at the time, and enable AI agents to write entire code segments.

"The feeling was we could actually go a lot further than that," Cherny explained. "And the model was almost ready for the next big step. So, we don't have to do type ahead anymore, we can just have the agent write all of the code."

Despite initial struggles, with the tool being "barely usable" for the first six months, the breakthrough came with the release of Opus 4 in May. This marked the beginning of exponential growth for Claude Code, a trend that has continued with subsequent model releases.

Is Coding Solved?

Cherny boldly states that, for him, coding is "solved." He shared that the entire Claude Code codebase, written in TypeScript and React for optimal model distribution, is now generated by AI. He personally writes dozens of pull requests daily, with one record-breaking day seeing 150 PRs.

"For me, it's just solved," he asserted. However, he acknowledges that this isn't universally true, citing complex codebases and niche languages as areas where AI is still catching up.

Boris's Personal Workflow: Mobile-First and Autonomous Agents

Cherny's personal coding workflow has evolved dramatically. He now conducts most of his work from his phone, utilizing the Claude app's code tab. He maintains numerous active sessions, often with hundreds of agents working concurrently, and thousands engaged in deeper tasks overnight.

A key element of his current workflow is the use of "{slash}loop," a feature that allows Claude to schedule recurring jobs using cron. These loops handle tasks like monitoring PRs, fixing CI issues, and aggregating feedback from Twitter. He also highlighted "routines," a server-based version of loops that continues to run even when a laptop is closed.

"I sort of feel like loops are the future at this point," Cherny remarked, highly recommending experimentation with the feature.

The Future of Teams: Generalists and Cross-Disciplinary Skills

Looking ahead, Cherny predicts a rise in generalist professionals, but with a broader definition than currently exists. Beyond engineers who might handle multiple domains like iOS, web, and server development, he foresees an increase in individuals who are cross-disciplinary. These could be engineers with strong product, design, or data science skills, or product managers with engineering capabilities.

He noted that this trend is already visible within the Claude Code team, where everyone, from the engineering manager and product manager to designers, data scientists, and even the finance and user research personnel, writes code. "Everyone on our team codes," he stated, "so, like they're specialist in something, but now also everyone's just coding."

The SaaS Apocalypse and Evolving Business Models

Cherny addressed the potential impact of AI on the Software-as-a-Service (SaaS) market. He believes the value of software products will shift as AI makes coding significantly cheaper. Drawing on the "seven powers" framework from the Acquired podcast, he suggests that certain modes, like switching costs and process power, will become less critical as AI can easily replicate or optimize workflows.

However, he argues that other powers, such as network effects, economies of scale, and cornered resources, will remain crucial.

Furthermore, Cherny predicts a tenfold increase in disruptive startups over the next decade. He explained that small teams can now build products with the same value as large companies, bypassing the legacy systems and internal resistance that established firms face when adopting new technologies.

"If you're starting fresh, then you can kind of build with AI natively from the ground up," he said. "So, I don't know. I I think it's the best time to build. It's the best time to be a startup. It's there's so much disruption coming."

Audience Q&A: Democratizing Software and the Pace of Change

During the audience Q&A, several key themes emerged:

• Model vs. Product Decisions: Cherny attributed the success of Claude Code to a mix of both model capabilities and product decisions, emphasizing the importance of building "something people love" and paying attention to user experience details.
• Democratization of Software: He drew a parallel between the current AI revolution and the invention of the printing press in the 15th century. Just as the printing press democratized literacy, AI is poised to democratize software development, making it a skill accessible to everyone, akin to sending a text message.
• The Internal vs. External Gap: While Anthropic uses the same models as the public, Cherny believes the significant gap lies in organizational processes and how teams integrate AI into their daily workflows. He noted that Anthropic uses Claude for "literally everything," with no manually written code remaining company-wide.
• Parallelization and Agent Orchestration: Cherny explained that while prompting plays a role in enabling parallel agent execution, the models themselves are increasingly capable of identifying and initiating parallel tasks autonomously, especially with newer versions.
• Cloud vs. Local AI: He posited that the distinction between cloud and local AI might become less relevant as models become more capable of deciding which resources to leverage for optimal performance.
• Co-work and Tool Integration: For general knowledge work, Cherny highlighted the importance of MCP (Messaging and Communication Protocol) connectors to integrate AI with existing tools like Salesforce and Google Docs, enabling powerful AI assistants like Co-work.
• Future Product Development: He pointed to Claude Design and further enhancements to Claude Code, along with advancements in agent parallelization (loops, batches) and computer use, as areas where AI will continue to drive significant product evolution.

Cherny concluded by emphasizing that the future of software development is not just about the technology itself, but about how we adapt our organizations and workflows to harness its full potential.