My co-pilot, until it becomes a super intelligent system.
My AI toolkit
This is what I've built with AI
Figma plugin
Vibe coded
The plugin uses “languagetool.org” api to check the spellings and grammar of the selected text layers in Figma, all for free.
Habit tracker app
xCode
Building channels to reduce information overload with gateways for efficient visual and content focused de-clutter.
Not published yet
How I work with AI
AI has changed how fast I can get to an outcome and how much complexity I can hold at once. I use AI as a thinking partner, it handles synthesis and generation, I handle judgment, reframing, and the decisions that require understanding people.
Below I lay out, how I used AI in my process with the example of Accel Atoms project, completed at Lopez.
Sprint map
Accel Atoms x Google AI Futures Fund
Week 1-2
Week 3-5
Week 6-9
Week 10-11
Week 12
AI
Map conversational flows
Cluster founder queries into intent patterns
AI
Synthesise founder pain points
Surface contradictions in research notes
Both
Pressure-test IA
AI generated edge cases and I evaluated feasibility
Me
Conversational flow review
Judgment on what actually serves founders
AI
Decision log draft
Annotate rationale behind key design choices
Me
Stakeholder alignment
Define success criteria with Accel Atoms team
Me
Reframe the problem
Reactive Q&A → proactive intelligence
AI
Copy variants
10+ tone variants for error + success states
Both
Form validation design
AI flagged logic gaps, and I curated the experience
Me
Handoff + sign-off
Strategic ownership, final call on every output
My workflow matrix
Phase
Research
Framing
Iteration
Prototyping
Handoff
AI handling
Cluster interview notes by theme
Surface contradictions, tag patterns
Argue against my problem frame
Generate alternative definitions
Generate copy + IA variants at speed
Stress-test flows against edge cases
Flag logic gaps and failure modes
Generate scenario trees quickly
Draft decision logs and annotations
Produce handoff documentation quickly
My ownership
Interrogate the clusters
Decide what's signal vs. noise
Reframe the problem
Strategic judgment on what matters most
Select, refine, and decide
Taste, empathy, and final judgment
Own the user experience
Validate with real people, not prompts
Verify intent is preserved
Strategic ownership of every output
A use-case
I wanted to create a time tracking app; that's fairly simple to use and serves its purpose. Then, I wanted to create 2 more variants; one for younger population with a gamified experience and the other one, which is extremely minimal and caters to business professionals.
The workflow created all three variants within 9 minutes.
+
My mindset towards design is slowly shifting to "outcomes". Traditional design practice focuses on creating one ideal solution that can be used by a large user group.
But, with AI, we (as designers) have the leverage to create personalised solutions for many users at once.
NN Group calls it the "Outcome-Oriented Design (OOD)".
What I'm experimenting with this month
13th Mar - 13th Apr
Context engineering
The frame shapes the action and I design the frame.
NLP (Neural Language Processing)
The intersection of linguistics and machine learning. Include things like sentiment analysis, named entity recognition, machine translation, and summarization.
Agentic AI
Independent and automated AI workflow.
My recommendations for you
© 2026 Prabhav Singh • All rights reserved