Sandbox Mode vs Architect Mode: Bringing AI to your product team: Prerna Singh

In this episode, Prerna Singh, CPTO at Avaaz, walks us through how AI is reshaping the way we prototype, learn and build digital products. Rather than replacing teams or skipping straight to production, she argues that AI shines when used as a “thought partner” to accelerate early‑stage experimentation. 

Through her own journey building a community platform on weekends, she demonstrates how tools like ChatGPT, Lovable (and later Claude / Replet) and Figma AI enabled her to move from blank page to clickable prototype in hours — while retaining the human insight, iteration and context that underpin good product work. 

The conversation reframes common assumptions about “fast‑AI = bypass human work,” and instead proposes a balanced adoption path: start in “sandbox mode,” learn and play — before graduating to “architect mode” where the real value to business begins.

Chapters
00:00 – Introduction & AI’s impact on product cycles
01:43 – Meet Prerna Singh: her background in product and community building
03:50 – The community problem: logistics over connection
05:11 – Turning to AI to solve her own problem
06:50 – What AI can’t do: user insight and human judgment
08:08 – From waterfall to short-cycle prototyping
10:54 – Using ChatGPT as a Socratic thought partner
13:07 – Working solo vs team: where AI fits
17:17 – From prompt to prototype: using Lovable
19:06 – Iterating with Figma AI and other tools
23:00 – Real feedback from real users
25:02 – Creating a feedback knowledge base with AI
26:16 – AI vs design sprints: same principles, new tools
30:36 – Sandbox mode vs architect mode
36:48 – Bringing AI into the team culture at Avaaz
38:49 – CEO buy-in and grassroots experimentation
39:52 – Closing reflections: product joy, pain, and play

Key Takeaways
— AI accelerates, but doesn’t replace human judgment. Tools like ChatGPT can rapidly prototype flows, but they cannot yet replace user insight or reflect true behavioural outcomes.
— Use a “sandbox mode” before “architect mode.” Give teams permission to play and experiment with AI — rather than expecting immediate business ROI — to surface real potential and build confidence.
— AI can democratise early product work. Product managers can rapidly build prototypes without waiting on design or engineering hand‑offs, enabling faster discovery and validation cycles.
— Prototypes—especially interactive ones—drive better conversations. Showing a working prototype sparks more meaningful feedback and generates concrete ideas than dry requirement documents.
— Treat AI as a “thought partner,” not a magic bullet. Use AI to challenge assumptions, generate structure, document learnings — but always apply human oversight, iteration and context.
— Proving value begins with internal culture, not production deliverables. Running team hackathons or off‑sites helps shift mindset, surfaces inventive uses, and builds shared AI fluency — before scaling across the business.

Featured Links
Follow Prerna on LinkedIn | https://www.linkedin.com/in/presingh/
Avaaz | https://secure.avaaz.org/page/en/
Prerna's slides from her talk at INDUSTRY Conference 2025 | https://drive.google.com/file/d/1zCODyzED1BXgZROXfyIZOn3xWzmpJNTz/view