Timeline
Nov 2025 – Feb 2026
ResXiv
- Product Design
- Brand
- Marketing
- 2025
Long story short
I co-founded ResXiv: an AI-native research workspace to search, download, read and write research papers with ease.
On a two-person team, I led product strategy, UX, and frontend.
01
Research
- 97 interviews and surveys with PhD students, professors, and lab leads
- Researchers from top universities globally, CMU, Macquarie, MBZUAI, Maryland, and more
02
Product
- AI paper search and goal-driven literature discovery
- PDF AI analysis, chat, and research assistant
- AI LaTeX editor and ResXiv's design system
03
Traction
- Organic signups from Harvard, Yale, Princeton, CMU, Meta, Microsoft, Adobe, and more
- 300 MAU and 2,000+ users across 30+ countries in three months
Impact & Outcome:
On a two-person team, we shipped and grew ResXiv to:
- AI paper search, Research assistant as an AI chat, PDF AI analysis & chat, AI LateX editor, and ResXiv's design system
- Organic signups from Harvard, Yale, Princeton, CMU, Meta, Microsoft, Adobe, and more
- 300 MAU and 2,000+ users across 30+ countries in three months
Why the research stack is broken:
Academic research still runs on decade-old, fragmented tools. As AI has made research output explode, there is still no unified tool for context building and knowledge management.
Fragmented tools
Work happens across tabs and context gets lost.
Outdated writing softwares
No technical innovation since 1980s in research drafting methods
AI poisoning
AI skyrocketed research, making relevant research hard to find.
Costly ecosystem
Researchers pay at every step, trapped in a costly loop.
How Might We:
How might we help researchers build on, and implement research without losing context?
Discovery: We talked to 97 researchers and found out…
We ran interviews, surveys, and focus groups with AI/ML researchers at Carnegie Mellon, Macquarie, MBZUAI, Maryland, and labs across India.
The sessions consisted of open-ended conversations focused on researchers' workflows and the pain points across the research landscape.
Need for a system that preserves context throughout the research cycle:
No tool builds research intuition
Nothing that connects topics, ideas and relationships.
Linear workflows kill collaboration
Existing research tools are not built for iterative research team-work.
“I spend weeks reading papers, but by the end, I lose track of how they connect. There's just too much to process and I struggle to see the bigger picture.”
“The hardest part is giving structure to research. Tools help me write, not think. What I need is a system that connects the dots for me.”
As a result
Lost insights Stalled progress Inhibited intuition Information overload Closed research
Studying the competitive landscape
To create a novel solution, we mapped where SciSpace, Paperpal, Overleaf, NotebookLM, and others actually cover the cycle. While most tools own one step well; none kept context across discovery → reading → writing.
| Competitor | Full research cycle | No-code drafting | Proof-backed | Built for academia |
|---|---|---|---|---|
| ResXiv | ||||
| Scispace | ||||
| Paperpal | ||||
| Anara | ||||
| AnswerThis | ||||
| Overleaf | ||||
| NotebookLM |
Goal: Create an intuitive experience that replaces multiple tabs into a unified research workspace.
ResXiv became the single place to search papers, read with AI, annotate PDFs, and draft, without losing context at every handoff.
Iteration 1: Version 1 to Version 3 of ResXiv
Problem
We started building too early without thoroughly wireframing the product. As a result, the first version of ResXiv tried to solve every research problem - research loops, collaboration, task assignment, turning each idea into a separate tab in a cluttered fashion.
Solution
Through continuous user feedback and iterative redesigns, we narrowed the scope and removed journaling, team collaboration, and task assignment. By focusing on the core research workflow, the interface became more intuitive, easier to navigate.
Iteration 2: Rethinking the search
Problem
The landing experience was built around an AI-first chat interface. However, usage patterns showed that researchers usually arrived with a clear research problem, making literature discovery their first priority and not AI exploration.
Solution
We merged AI assistant search with paper search into a single, goal-driven experience. Researchers could search by research problem, related topics, or keywords to instantly find relevant papers and insights, making the value proposition clear from the start. The optimized search delivered AI-prioritized rankings, helping users find relevant papers up to 3× faster than the old chat.
Iteration 3: Implementing pre print analysis feature
Problem
We overlooked an immediate, high-value use case: preprint analysis. Researchers faced constant pressure before top conference deadlines to check missing citations, where top conferences reject papers even for a single missed citation.
Solution
This led to our Preprint Analyzer, which lets researchers analyze their manuscripts in one click to identify missed citations, uncover gaps in related work, surface exaggerated claims and highlight opportunities for improvement before any important submission.
Other stand out features
What researchers said when we shipped the final version
When we stopped iterating
500+ organic signups validated the problem, we paused new surfaces when growth plateaued and started to focus on marketing.
What I'd do differently:
- Define return-rate targets before building collaboration, export, and adjacent features.
- Start pricing experiments earlier; growth alone is not proof of willingness to pay.
- Own one painful moment of the user journey.
- Market the product, market the product & market the product !
Learnings: What I took forward
- Context over generation: Researchers valued traceable understanding over polished summaries.
- Advocate with research: Design decisions landed when tied directly to user study insights.
- Ship with a small team: Being designer and engineer collapsed feedback loops, but prioritization was everything.
