Backtrack

Role

Product Design Intern

Duration

2025 - Present

Project

Product Design / UX Design

1. Problem & Goal

Backtrack wanted to make meetings more actionable by letting users quickly generate follow-ups, summaries, and action items from their notes. The challenge was to design a chatbot interface that felt approachable, intuitive, and efficient for busy professionals.

2. Research & Inspiration
I looked into existing productivity tools (Notion AI, Grammarly, Jasper) and analyzed how they handle tone, context awareness, and user trust. One key insight: users value clarity with flexibility — they want a starting point they can quickly edit.

3. Workflow & Prompting

Explored Multiple Prompt Structures → Balanced specific outputs (structured follow-ups) with natural tone (casual summaries).
Tested Iterations of Prompts, such as:
- “Draft a follow-up email for today’s call.”
- “Turn my notes into an action plan.”
- “Summarize my last meeting into key takeaways.”
Used ChatGPT for Authentic Copy → Instead of placeholder lorem ipsum, I generated realistic dialogue. This helped test tone (formal vs. casual), improved user trust, and made mockups feel like a real product.

4. Iterations in Interaction Design

Initial Concept → Simple input box with generic AI responses.
Refinement → Added pre-built prompt buttons for efficiency and guidance.
Tone Adaptation → Allowed users to request edits (“make it more casual”) to reinforce control and trust.
Progression → Moved from a blank-start input to a guided prompt + editable reply system.

5. Final Experience

Balanced Speed + Flexibility → Pre-built prompts for efficiency, with custom input for flexibility.
Realistic Copy for Usability → ChatGPT-generated content revealed edge cases early and informed decisions around tone, trust, and communication style.