Lenny's Podcast

Aparna Chennapragada: Why prompt sets are killing the PRD

Through Microsoft's Frontier program, prompt sets replace PRDs; live demos beat documents and tastemaking trumps title for AI-era product leaders.

Aparna ChennapragadaguestLenny Rachitskyhost

May 17, 20251h 1mWatch on YouTube ↗

WHAT IT’S REALLY ABOUT

Microsoft CPO explains why AI prototyping and NLX will reshape products

Aparna Chennapragada, Microsoft’s Chief Product Officer, shares how AI is transforming product development, enterprise software, and the future of work. She argues that every product builder should be rapidly prototyping with AI and treating prompt sets as the new PRDs, while distinguishing between “solve” and “scale” phases of product building.
She introduces concepts like agents (autonomous, multi-step, naturally interactive tools) and NLX—natural language interfaces—as the new UX, with their own emerging design primitives such as prompts, plans, and showing work. Aparna also explains why PMs and coding are far from dead; in a world flooded with ideas and prototypes, tastemaking, editing, and computer science-thinking become more valuable.
Drawing on experiences from Google, Robinhood, and now Microsoft, she discusses enterprise constraints (governance vs. usability), how she operationalizes “living one year in the future” via Microsoft’s Frontier program, and lessons from past failures like Google Now. She closes by imagining co-working environments where humans and AI agents collaborate in real time to produce far more than either could alone.

IDEAS WORTH REMEMBERING

5 ideas

If you’re not prototyping with AI, you’re building too slowly.

Aparna insists that new features and products should come with live demos and prompt sets; AI drastically shortens the time to first demo and improves communication within teams, even if deployment and scaling still take time.

NLX—natural language interfaces—are becoming the new UX layer.

Conversational products aren’t just “the model leading the product”; they require deliberate design of prompts, plans, follow-ups, and how much internal reasoning to expose, just like buttons and menus in GUIs once did.

Agents are defined by autonomy, complexity handling, and natural interaction.

Good agents can take delegated goals, execute multi-step, non-trivial tasks asynchronously, and interact through rich, natural modalities (not just chat), for example, preparing a persuasion strategy for a high-stakes meeting.

Separate “solve” mode from “scale” mode in zero-to-one work.

Early on, teams must tolerate chaos, wide lurches in direction, and avoid premature metrics and over-optimization; only after finding a real ‘set timer and play music’ use case should you lock into scaling and hard metrics.

AI increases the supply of ideas, so editorial taste matters more.

As engineers, designers, and others can prototype rapidly with AI, PMs and leaders must lean into tastemaking and editing, earning influence by judgment rather than title, or risk shipping Frankenstein products.

WORDS WORTH SAVING

5 quotes

In this day and age, if you’re not prototyping and building to see what you want to build, I think you’re doing it wrong.

— Aparna Chennapragada

NLX is the new UX.

— Aparna Chennapragada

Prompt sets are the new PRDs.

— Aparna Chennapragada

Being early is the same as being wrong.

— Aparna Chennapragada

Excel is proof that non-coders also have to program.

— Aparna Chennapragada (relaying an Excel PM’s insight)

AI-driven product development and rapid prototyping (“prompt sets as the new PRDs”)Agents: definition, capabilities, and how they change software and workNLX (natural language interfaces) as the new UX and its design principlesDifferences between consumer vs. enterprise product building, especially around governanceThe evolving roles of product managers, engineers, and “software operators” in an AI worldFrameworks for zero-to-one products (tech, behavior, and business model shifts)Career lessons, failure (Google Now), and “living one year in the future”

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.