How One Researcher Builds a Knowledge Base for the Entire Team

The Limits of Keeping Up with Research

Every morning: check arXiv, scroll through X for trending papers, scan conference award lists. For anyone on the front lines of research or investment decisions, information collection is part of the job.

The problem is that this work is inherently individual.

• Only certain people know which sources to follow
• Insights from papers stay in personal notes, never reaching the team
• When shared, they get buried in chat history and become impossible to find again

PE/VC investment teams, research labs, R&D departments. All need a system where one person's research efforts become available to everyone.

The Three Layers of a Research Knowledge System

I track award-winning papers in AI robotics: ICRA, IROS, RSS, CoRL, and other major international conferences. Collecting, summarizing, and sharing them with my team.

Built on Kylon, this system has three layers:

• App – a paper database. Titles, authors, PDFs, AI-generated summaries, and tags – all structured and searchable.
• Workflow – automated collection. Regularly pulls new papers from arXiv, X, RSS feeds, and conference sites.
• Channel – team collaboration. Members discuss collected papers, share insights, and connect findings to their own work.

Step 1: Build a Paper Database App

Using Kylon's App feature, create a structured database to store and organize papers.

Example fields:

• Title – paper title (original language)
• Conference / Journal: ICRA, IROS, RSS, CoRL, T-RO, etc.
• Award Category – Best Paper, Best Student Paper, etc.
• Year – publication year
• Authors – author list
• PDF – link to the paper
• AI Summary – an AI-generated overview in plain language
• Tags – manipulation, SLAM, Soft Robotics, etc.

The finished app displays papers as cards, filterable by conference and year. Unlike a spreadsheet, each paper record has its own discussion thread , making it a searchable, discussion-ready knowledge base.

Step 2: Automate Collection with Workflows

No need to manually check every day. Kylon workflows pull new content from your specified sources automatically.

Example configuration:

• arXiv – daily scan of the robotics category for new papers
• X (Twitter) – monitor posts from researchers you follow, extract paper links
• Conference sites – capture award paper lists when published
• RSS / News – track new articles from specified media outlets

The AI agent parses incoming items, extracting titles, authors, and summaries, then adding them to the app. Deduplication is automatic; the same paper from multiple sources won't create duplicate entries.

When new papers are added, a notification arrives in your team channel: "12 new papers from arXiv this week. Here are the 3 most notable ones." Everyone stays current without anyone doing extra work.

Step 3: Discuss and Build on Findings as a Team

Collecting information is only half the value. The real knowledge emerges when the team thinks about, discusses, and applies it to their work.

In Kylon, every paper record in the app has its own discussion thread.

How teams use this in practice:

• An investment team member notices a paper's methodology could apply to a portfolio company's technology, and shares the insight in the thread
• A lab researcher posts a question about an award-winning paper's experimental approach, and a colleague answers
• A manager picks "3 papers worth reading this week" and posts them to the channel

One person's curation becomes the team's decision-making fuel. Conversations like "Have you read this one?" happen in structured threads that persist and can be referenced later.

One Person's Work Becomes the Team's Knowledge Asset

With traditional approaches, information collection depends on individual effort. Tracked papers stay in personal bookmarks, insights get dropped in chat and buried, and retrieval is nearly impossible.

With Kylon, workflows collect information automatically, structured apps centralize storage, and record threads preserve team discussions. Search and filter give instant access to any paper.

When one person builds the system, the entire team benefits. When a new member joins, every past paper and every discussion is already there, waiting. What was once siloed information becomes the team's shared knowledge asset.