Overview of Pioneers of AI: How fast can you upskill in AI? We did a sprint to find out.
This episode follows the media company WaitWhat as it runs a three-day, company-wide “AI sprint” to rapidly upskill its team and identify practical ways to use AI across podcast production, event planning, operations, and internal workflows. Hosted by Rachel Ishikawa with Rana El Kaliouby, the episode explores what happens when a small but busy company pauses regular work, hands everyone AI tools, and asks them to rethink how they work—while also confronting the fear that the same tools could eventually replace some jobs.
Why WaitWhat Ran the Sprint
WaitWhat is a small media company making podcasts like Pioneers of AI, Masters of Scale, and Rapid Response, plus newsletters, video, social content, and live events like the annual Masters of Scale Summit.
The company’s leadership felt that:
- Their AI usage had been inconsistent and ad hoc
- AI would likely reshape media workflows whether they liked it or not
- Employees needed hands-on experience, not just theory
- The team should be prepared for an AI-driven future rather than react to it later
To help guide the experiment, they brought in AI engineer Parth Patil, who urged the company to become “AI native” and to start thinking in terms of collaboration with AI rather than occasional experimentation.
How the AI Sprint Worked
The company paused operations for three days and split into small teams focused on real business problems. Each team used tools like Claude and Replit to prototype solutions, with the goal of presenting their results at the end of the sprint.
Teams explored questions such as:
- How can AI speed up video production?
- How can AI surface better guest ideas?
- How can AI improve Summit planning and logistics?
- How can AI reduce repetitive admin work?
The sprint was highly collaborative, but also messy—just like real adoption tends to be. People ran into tool limitations, technical bugs, and uncertainty about where AI should stop and human judgment should begin.
The Three Main Lessons
1. Treat AI like a colleague
The best results came from back-and-forth conversation, not one-shot prompting.
Key practices included:
- Asking AI to interview the user before building
- Refining prompts through dialogue
- Using natural language to explain goals and context
- Treating AI as a collaborator rather than a command line
The guest-speaker team demonstrated this well by asking Claude clarifying questions before generating a database of potential guests.
2. Target repetitive, click-heavy work
AI was most useful when it reduced tedious manual tasks.
Strong use cases included:
- Consolidating scattered travel and hotel data
- Building dashboards for event operations
- Sorting ticketing and application workflows
- Pulling information from email, Slack, and spreadsheets into one place
D’Angela Napier’s Summit hotel operations dashboard stood out as a strong example of AI helping with “invisible work” that is essential but time-consuming.
3. Decide carefully what to delegate, build, or buy
The team emphasized that AI should not replace the most human, creative parts of the work.
They had to think through:
- What should remain under human judgment
- Which tasks AI can accelerate but not own
- Whether to build custom tools or buy off-the-shelf products
- How to measure cost, speed, and quality tradeoffs
For creative areas like video editing and audio production, the goal was not full automation. Instead, the aim was to get to rough cuts and first passes faster while preserving editorial and artistic control.
Notable Projects and Outcomes
Several prototypes emerged from the sprint:
-
Guest Speaker Engine
A standalone web tool for surfacing guest ideas and providing context, tags, and suggestions for podcasts and live events. -
Summit Hotel Operations Dashboard
A real-time system for tracking hotel preferences, travel changes, and related logistics for the Summit. -
Summit Application / Ticket Monitoring Tools
Tools to monitor applications and ticket sales more efficiently. -
Video Workflow Experiments
Attempts to use AI to speed up rough cuts, file organization, and other production steps, though no perfect solution was found.
The sprint did not produce a single “winner,” but it did generate many promising ideas and showed that a small team can uncover meaningful workflow improvements quickly when given time and permission to experiment.
Tensions, Risks, and Real-World Limits
The episode also wrestles with the uncomfortable side of AI adoption.
Job displacement anxiety
Some team members were understandably skeptical, especially creative staff who worried about:
- AI scraping artists’ work without consent
- Environmental costs of large-scale model usage
- AI gradually replacing roles rather than supporting them
Leadership was transparent that AI could change jobs significantly, and that some aspects of current roles may look very different in the future.
Cost and ROI uncertainty
The sprint made clear that AI adoption has hidden costs:
- Tool subscriptions
- Token usage
- Time spent learning instead of producing
- Ongoing maintenance and security needs
It’s not yet easy to know whether a project should remain manual or become automated, because the economics are still evolving.
Security and data protection
Because WaitWhat handles sensitive information such as emails, financial details, and event attendee data, security became a major concern.
The team began building guardrails, including a security-focused AI agent nicknamed “Warden”, to help protect against issues like prompt injection and misuse of private data.
What Happens Next
The sprint was only the beginning. WaitWhat formed a task force to decide which prototypes to keep, refine, or merge into larger internal tools.
Next steps include:
- Rolling out selected AI projects across the company
- Training staff during weekly company-wide meetings
- Evaluating security and token costs more carefully
- Measuring actual ROI over time
- Combining related prototypes into broader systems
The episode frames this as a shift from experimentation to integration: the sprint was the start of the race, but the real work is turning AI ideas into durable workflows.
Core Takeaway
The biggest lesson is that AI upskilling works best when it is:
- Hands-on
- Collaborative
- Grounded in real work
- Focused on augmentation, not blind automation
The episode’s closing message is optimistic: thriving in the age of AI means staying curious, experimenting continuously, and using AI to remove drudgery so people can spend more time on judgment, creativity, and meaningful work.