25:12
DoorDash gave every employee Claude Code
Claude
Overview
This video discusses DoorDash's initiative to equip all employees with access to Claude Code, an AI coding assistant, to enhance AI fluency across the company. It explores the journey from initial adoption to widespread use, highlighting the impact on engineering throughput, the challenges of integrating AI into existing workflows, and the strategic importance of fostering experimentation while maintaining security. The discussion emphasizes how AI is reshaping software development, knowledge work, and the overall pace of innovation, with a focus on practical implementation, return on investment, and advice for both leaders and new graduates.
How was this?
Save this permanently with flashcards, quizzes, and AI chat
Chapters
- DoorDash provided all employees with access to Claude Code to increase company-wide AI literacy.
- While some employees, including executives, have embraced AI tools, many still view AI primarily as a chat interface.
- Connecting AI tools to everyday applications like Gmail or Slack can significantly boost efficiency for knowledge workers.
- The initiative aims to raise the baseline AI proficiency across the organization, not just within engineering.
Understanding the broad adoption strategy helps contextualize the company's commitment to AI and its potential impact on all roles, not just technical ones.
Many employees' eyes light up when they realize they can connect AI tools to Gmail or Google Calendar to become more efficient knowledge workers.
- The speaker's coding journey began at age nine, and they later stopped coding as DoorDash scaled.
- Using Claude Code allowed the speaker to return to shipping production code after years away from direct coding.
- The speaker set a personal goal to have AI agents write almost all code, using terminal or desktop apps.
- Early attempts to use AI for production code failed due to environmental configuration issues, but later attempts succeeded due to advancements in AI models.
This personal narrative illustrates the transformative power of AI coding assistants, showing how they can re-engage experienced developers and enable new workflows.
The speaker recounts a distinct memory of trying to code actual features in production, aiming not to ask for help, but failing initially due to the AI's inability to configure the local environment correctly.
- Recent AI model advancements have created an inflection point, enabling AI agents to understand and execute complex coding tasks.
- It's now possible to ship production code in multiple programming languages using AI assistance.
- The rapid evolution of AI models means that ideas that previously failed might now work, necessitating continuous re-evaluation of approaches.
- Teams are encouraged to unlearn rigid attachments to existing solutions and reimagine workflows based on new AI capabilities.
This chapter highlights the dynamic nature of AI development and the importance of adaptability for individuals and organizations to leverage emerging capabilities.
An idea that previously didn't work with an older AI model might now function effectively with a newer one, suggesting that re-testing old concepts with updated technology is a valid strategy.
- Leaders should actively use AI tools themselves to build empathy and understand the challenges engineers face.
- Setting goals for managers to ship production code using AI, not just prototypes, provides valuable insights.
- DoorDash invested in providing token budgets and tools, emphasizing that AI exploration is a supported initiative.
- Encouraging leaders to play with AI tools helps them understand capabilities and identify engineer blockers.
Leadership buy-in and active participation are crucial for driving successful AI adoption and fostering a culture of experimentation and learning.
Engineering managers are encouraged to set a goal to ship production code using AI, going through all the necessary hoops at DoorDash, to gain firsthand understanding of the process and capabilities.
- Increased AI-driven coding leads to new challenges, such as longer code merge times and the need to reimagine CI/CD processes.
- Security concerns are paramount, requiring AI to help catch security issues before deployment.
- DoorDash introduced Cloud Code in 2025 and worked with IT security to establish rapid procurement and review processes for AI tools.
- An internal platform called 'Flux' was built to provide secure, cloud-based environments for AI coding sessions and agents.
Addressing the practical challenges and building robust infrastructure are essential for scaling AI adoption safely and efficiently.
The internal 'Flux' platform provides secure VMs in the DoorDash cloud, allowing users to spin up AI coding sessions and powering AI code review agents.
- Encouraging employees to share both successes and failures is vital for collective learning.
- Identifying and empowering early adopters and success stories as internal advocates helps drive adoption.
- Written artifacts are valuable for distributing learnings across the company and can be read by AI agents.
- Creating safe spaces and forums for teams to share experiences, both positive and negative, is crucial.
A culture that supports open sharing of knowledge and experiences, including setbacks, accelerates learning and innovation.
Sharing written artifacts about successful workflows or experiments, like a workflow that didn't work or wasted tokens, helps others learn without having to repeat the same mistakes.
- AI is enabling significant increases in development throughput, with projects being completed 3-5x faster.
- Smaller, self-sufficient teams can move faster by reducing coordination overhead and process friction.
- Upfront investment in making codebases 'agent-friendly' and standardizing skills accelerates AI adoption.
- Measuring ROI involves tracking code throughput and delivery speed, with the ultimate goal of delivering customer value faster.
Quantifying the impact of AI and adapting organizational structures and processes are key to realizing its full potential for accelerating value delivery.
One engineer used AI to complete a massive code migration in 3 weeks, a task that would have historically taken four engineers a quarter.
- Traditional processes like design docs and extensive reviews can become bottlenecks with AI-driven speed.
- Cross-functional buy-in is necessary to adapt processes and unlock AI velocity across the entire company.
- Encouraging engineers to become more full-stack and generalists makes them more fluid across the codebase.
- Empowering teams with executive sponsorship and token budgets allows them to experiment and identify non-coding blockers.
Shifting mindsets and adapting organizational processes are as critical as adopting AI tools to achieve significant gains in speed and efficiency.
Instead of requiring engineers to achieve pixel-perfect designs, AI allows them to reach a workable state, which designers can then refine, enabling faster iteration.
Key takeaways
- AI coding assistants like Claude Code can significantly boost engineering throughput and re-engage developers who may have stepped away from coding.
- Continuous adaptation is essential, as advancements in AI models mean previously unsuccessful approaches may now be viable.
- Leadership must actively engage with AI tools to foster empathy, understand challenges, and drive adoption across the organization.
- Building secure infrastructure and reimagining CI/CD processes are critical for safely scaling AI-driven development.
- A culture of open sharing, including both successes and failures, accelerates learning and innovation in AI adoption.
- Empowering smaller, self-sufficient teams and investing in agent-friendly codebases can dramatically speed up project delivery.
- Achieving AI velocity requires not just technological adoption but also a fundamental rethinking of cross-functional processes and team mindsets.
- The ultimate measure of success for AI adoption is the faster delivery of customer value, not just increased code throughput.
Key terms
Claude CodeAI FluencyKnowledge WorkerProduction CodeAI AgentInflection PointCI/CDFlux PlatformAgent-Friendly CodebaseCross-Functional Buy-in
Test your understanding
- How did DoorDash aim to increase AI fluency across its entire workforce, beyond just the engineering department?
- What personal experience did the speaker share that illustrates the impact of AI coding assistants on an individual developer's workflow?
- Why is it important for leaders to actively use AI tools themselves, according to the video?
- What are some of the key challenges DoorDash faced when integrating AI-driven coding into its development processes, and how did they address them?
- How does DoorDash measure the return on investment (ROI) of its AI initiatives, particularly in engineering and knowledge work?
- What advice is given to engineering leaders and new graduates regarding the adoption and use of AI tools?