doXmind - The AI editor Notion should have built
byβ’
Since our February launch, doXmind has evolved dramatically:
Database Blocks β Notion-style databases with table, board, gallery & list views, custom properties, and CSV export.
π¬ Inline Comments β Highlight text to leave comments with resolve/unresolve tracking.
Multi-Column Layouts β Arrange content in 2-4 flexible columns.
AI Thinking Mode β Deeper reasoning for complex requests.
Pro & Max Plans β Premium themes, animated frames, and expanded AI credits.
Replies
OpenYak
@mentionsΒ @wangzhang_wuΒ Bloody congrats on the launch to you and your team! Just one doubt; how does doXmind handle context retention across long docs or multiple revisions, especially when blending uploaded research with deeper reasoning tasks?
OpenYak
@mentionsΒ @swati_paliwalΒ Thanks Swati! Great question.
We handle this in a few layers:
Long docs β Documents are parsed into a hierarchical section outline. The AI navigates by structure first, then dives into specific sections on demand, so even very long docs don't blow up the context window.
Multi-turn revisions β We use a "3-1-3" compression strategy: first messages (original intent) and last messages (recent context) are kept intact, middle turns get compressed into key notes (files touched, decisions made). Long sessions stay coherent.
Uploaded research + reasoning β KB documents are first-class. The AI can search and cross-reference them alongside your active doc through a unified tool system. Thinking mode streams reasoning steps separately so you see the full chain of thought.
In short: smart compression + lazy section loading + unified KB access. The AI "remembers" what matters without needing infinite context.
Happy to go deeper on any of these!
So proud to see this go live! π The team has been shipping non-stop since February β Database Blocks, Inline Comments, Multi-Column Layouts, AI Thinking Mode β every feature came from real user feedback. Excited to see what the PH community thinks. Happy to answer any questions! π
OpenYak
@pei_lin1Β Thank you Pei!! π€ Couldn't have shipped this fast without the whole team grinding together. v0.1 to v1.1 in 6 weeks β and honestly, the best features came from listening to what real users kept asking for. Let's keep building!
Tried a bunch of "AI writing tools" this year and they all feel the same -chat window on the left, doc on the right, hope for the best. The skill system here is what's actually different. Great job on the UI!
OpenYak
@spunchevΒ Thanks so much, Serge β really appreciate this.
Thatβs exactly the problem we wanted to solve. Most AI writing tools feel like the same wrapper, so we built Doxmind around skills to make the workflow actually different. Glad the UI stood out too.
β8 updates in 6 weeksβ is the kind of shipping cadence that tells you everything about a teamβs conviction. Congrats on this @wangzhang_wu Are you planning a Notion import feature so users can migrate their existing workspace?
OpenYak
@jerrybydayΒ Thanks Jeremiah! That means a lot π
Notion import is definitely on our radar. We know migration friction is a real barrier β nobody wants to start from scratch. We're planning to support Markdown and CSV import first (CSV is already live in v1.1), and Notion export-to-Markdown is a natural bridge.
A dedicated Notion importer that preserves database structures and page hierarchy is something we want to build properly, not just partially. It's on the roadmap for a future release.
Hey @wangzhang_wu Congrats on this! With thinking Mode now handling complex reasoning, how are you approaching context windows when the AI is referencing multiple database blocks and documents simultaneously?
OpenYak
@jacklyn_iΒ Great question Jacklyn! Context management is one of the hardest problems we've tackled.
Our approach: we don't dump everything into one prompt. When the AI references database blocks and documents simultaneously, we use a prioritization pipeline β the system first identifies which blocks and KB files are most relevant to the current query (using embeddings + relevance scoring), then selectively loads only the high-signal content into context.
For Thinking Mode specifically, the AI gets a structured summary of available data sources first, then "pulls in" specific blocks/documents as needed during its reasoning chain β similar to how a researcher would consult references while writing, rather than reading every paper upfront.
It's still an evolving system β we're constantly tuning the retrieval to balance completeness vs. context efficiency.
The multi-agent AI integration with Database Blocks is really smart - instead of treating databases as separate from AI, you're weaving them together. The AI Thinking Mode for deeper reasoning is particularly interesting.
I'm curious about the collaborative use case: how does the inline comment system work when multiple AI reasoning threads are happening? Can you see when someone's AI is 'thinking' through a complex decision in a shared document?
Really appreciate the detailed breakdown! The Writing Agent, Global Agent, and KB Agent architecture is fascinating - having specialized agents that can reason across Database Blocks rather than treating them as dumb storage is a meaningful distinction. The KB Agent especially resonates; we're exploring similar knowledge-graph thinking with Kinetic Mingle's matching engine, where user context needs to flow through multiple reasoning layers. Excited to see how the collaborative AI threads evolve!