Agentic Coding: The Impact Gap Is Closing
Agents are spending more tokens, but are they getting more impact in organizations?
Jellyfish Research
Insights about software engineering teams from the Jellyfish Research team
Live Last read · last published · next check
Latest posts
Tokens Are Rocket Fuel, Spending Tokens Is Rocket Science
Looking to get the most out of your software development token spend? Here's how you can overcome the headwinds to build a better engineering machine.
The State of AI: H1 2026
Exploding Spend, Diverging Returns
How building an MCP App changed the way we understand the value of MCP
It’s been over a year now since we released our Jellyfish MCP.
Product Update: The Jellyfish MCP Is Now Documentation-Aware
Grounded answers, plus real docs for your agentic coding loops
Our Journey with OpenClaw: Jellyclaw
At Jellyfish we at the research team are looking for ways to innovate and research new trends. Openclaw was one under our radar since it’s inception and we decided to start this journey to understand
The Opus 4.7 Tokenizer Tax
Anthropic quietly raised Opus prices by changing the tokenizer, not the price tag. Here's what it actually costs when measured across 17K real-world Claude Code users.
AI Agent Autonomy
Autonomous agents are becoming more ubiquitous, with releases from the industry and users experimenting with running an AI on their computer. We asked ourselves: what does autonomy look like in AI age
Teaching Your AI Agent Pays Off
Investing in skills and context files yields measurable productivity gains
Tokenmaxxing and the Agentic Barrier
Tokens are like rocket fuel, and just like a rocket, going faster and faster requires burning exponentially more