We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
OpenAI Blog·Aug 18·5 min read
Q&A with DoorDash’s CPO, Mariana Garavaglia
TL;DR
DoorDash faced a challenge scaling AI adoption across its workforce, with employees lacking structured access to AI tools and training. The company implemented an enterprise AI enablement program, providing employees with AI literacy training, access to large language models (LLMs), and internal tools to build AI-powered solutions.
✦ Why It Matters
Learn how to structure organizational AI adoption programs that empower teams to ship features faster through training and tool access.
Key Takeaways
- Provide structured AI literacy training to all employees, not just engineers, to democratize capability.
- Grant hands-on access to LLMs and internal AI tools so teams can experiment without friction.
- Measure adoption success by tracking feature velocity and time-to-ship for AI-powered solutions.
- Combine top-down enablement programs with bottom-up innovation to sustain long-term AI culture change.
Full Summary
DoorDash recognized that while AI capabilities existed, most employees lacked practical knowledge and access to leverage them effectively in their work. Chief People Officer Mariana Garavaglia led an organizational effort to democratize AI adoption by establishing structured learning programs, providing access to large language models (LLMs—AI systems trained on vast text data), and creating internal platforms for experimentation.
The approach combined top-down enablement with bottom-up innovation, allowing engineers and non-technical staff to learn AI fundamentals and apply them to real business problems. By removing barriers to AI experimentation and providing hands-on training, DoorDash saw measurable improvements in how quickly teams could prototype, build, and deploy AI-enhanced features.
This strategy demonstrates how organizational structure and people-focused initiatives, not just technology, drive successful AI scaling.