Where ideas
branch, not scroll.
Your research becomes a map. Every node is an AI agent. Ask a question, branch into sub-topics, set instructions that cascade down.
Challenge assumptions. Cite counter-evidence.
Every branch sees the full picture
Child nodes inherit the entire conversation history from every ancestor. No copy-paste. No "as I mentioned earlier." Branch deep: the AI still knows what came before.
Usage-based pricing growing 35% YoY across mid-market...
- 3 direct competitors in segment
- Avg. ARR growth: 127%
Set the rules once. They cascade.
Attach an instruction to any node and every descendant follows it. Change the instruction and the whole sub-tree adapts.
Challenge assumptions. Ask "what could go wrong?" Cite counter-evidence.
Warning: organic-only GTM rarely scales past $2M ARR...
Counter-argument: freemium conversion at 2.3%...
Hours of research, one brief
Compress a sprawling conversation into a concise memory node. Descendants see the distilled insight. Your context window stays sharp.
Long conversation compressed...
Building on the GTM framework: Q1 focus on ICP validation...
Charts, diagrams, and code in the node
You are not limited to a scrolling transcript. Every node can search the web, render charts and diagrams, embed interactive web pages, write code, and call any MCP you connect.
SELECT dept,
SUM(cost) total
FROM expenses
GROUP BY 1- Base: $4.2M ARR by Q4
- Optimistic: $6.1M ARR
Usage-based pricing growing 35% YoY across mid-market...
- 3 direct competitors in segment
- Avg. ARR growth: 127%
Challenge assumptions. Ask "what could go wrong?" Cite counter-evidence.
Warning: organic-only GTM rarely scales past $2M ARR...
Counter-argument: freemium conversion at 2.3%...
Long conversation compressed...
Building on the GTM framework: Q1 focus on ICP validation...
SELECT dept,
SUM(cost) total
FROM expenses
GROUP BY 1- Base: $4.2M ARR by Q4
- Optimistic: $6.1M ARR
Start with a question. End with a map.
Real prompts, real branches. Here's how people use it.