Context-Mode: From a Small Idea to a Big Journey

This week, I want to step a bit outside my usual topics. It’s not that there’s a lack of buzz in the Blockchain/Web3/AI worlds, there definitely is. I mean, we could dive deep into the closure of Anthropic’s Fable 5 model, but I’ll leave that to others and instead steer towards a different path within the AI world.

I want to talk about Context Mode, which started as a simple question about making AI and token/context use more efficient while coding, and over time, with community input, evolved into a multi-layered product.

If you use AI for coding, the “you’ve reached your limit” message is probably all too familiar to you. 😊

I first came across Mert Koseoglu on LinkedIn, back when he was working on something called Seclaw, an agent product. This was around the time I started using AI for coding myself.

Initially, context wasn’t a huge issue for me, but it was the biggest complaint among my full-time coder friends. I recommended it to my BottomUP team and other developer friends; those who tried it were pleased.

When I started coding more, I tried it too and was satisfied. In fact, I even exchanged some feedback with Mert during that time.

Even though I followed both Mert and Context Mode on social media, it was only during Mert’s presentation at Clawcon that I realized how much the product had grown beyond just saving tokens or extending usage time with Claude.

Clawcon was where we finally met face-to-face, and in that short span (maybe two months), the product had expanded beyond just addressing “token” and “context” issues. It now focused on what AI needs to remember about you and your usage, and on making backend algorithms more efficient. It was available on over 10 platforms, moving far beyond just being a Claude MCP.

I won’t dive into the technical details; those interested can follow the Context Mode page.

Shortly after its release, Context Mode appeared on HackerNews, jumping from around 200 to over 2,600 stars, drawing significant interest from the Claude community.

Today, with over 16,000 stars and around 234,000 users, including employees from large global companies, and contributions from people worldwide, Context Mode is supported by a core team of eight from countries like Korea, China, and France.

New efficiency-focused branches have been added to the product family, offering a perspective that can greatly benefit large software teams using AI by tracking and enhancing their productivity.

I’m not just saying this because I’ve used it and met the founder, but Context Mode is a lovely product family. 😊

I know this week’s newsletter feels a bit like product placement, but my reason for writing this isn’t just the product’s success.

It’s also to highlight how the concept of community, which entered our lives with Web3/blockchain, reflects in the AI and software world. It’s about showing how a well-thought-out and structured product that starts from a simple idea can grow with community support.

And to demonstrate once again the immense support open-source projects can receive from people around the world whom you’ve never even met.

Additionally, the latest piece of the Context Mode product family, Insights, provides incredible “insights” for large software teams about the process, the code, and exactly where things get stuck with AI.

I truly believe that the efficiency Context Mode brings to individuals will also be highly beneficial for large companies and software teams with Insights. Like almost every AI-concept product, it will develop and progress as it’s used and as the number of users grows. So yes, I guess you could see this as a product recommendation.

Important Note: Due to some issues, when I uninstalled and reinstalled Claude, the Context Mode MCP configuration disappeared. How did I notice? My tokens ran out halfway through the first session. 😅 I had forgotten what life without Context Mode was like; it was a useful reminder. 😎