One of the leading crypto exchanges is now in the news for its bold AI-first plan, which has matured into something with its own ecosystem, and provides a good example for companies looking to innovate.

It’s called Forge now, but apparently, before that, it was Cloudbot, and in the beginning, it was Claudebot. That first one is confusing, since the technology doesn’t belong to Anthropic, but to be fair, in those days, Claude was still relatively nascent.

Anyway, Forge has become part of a greater community of tools in an open source framework called OpenSWE. OpenSWE is an open-source coding agent framework . It provides the architecture for an AI software engineer: agent orchestration, cloud sandboxes, tool use, Slack/Linear/GitHub integration, subagents, and automatic pull request creation. But it’s based on Coinbase Forge, as well as systems pioneered by payment handler Stripe and financial operations platform Ramp.

One of the most interesting pieces of the Forge platform is something called Mux, a tool with around 600 users at Coinbase. It’s described as a part of the stack that lets individual engineers become orchestrators, running multiple AI agents in continuous workflows.

I got the following input from software engineer Ry Walker, who has put together a briefing of the technology on his web site :

“AI coding agents made individual engineers faster, but the workflow around them stayed sequential — coding had a concurrency problem. With Mux, one engineer runs three or four agents in parallel (one implementing an API, one writing integration tests, one fixing a bug, one refactoring a legacy module), reviewing each as it finishes.”

That’s a redefined workflow for the twenty-first century, the agentic age that we are now in.

Other functions include Slack routing and a “bug-to-fix pipeline” for automating what humans used to do themselves.

Here’s a helpful little bit of context, involving Chintan Turakhia , Head of Engineering at Coinbase.

It turns out that in January, Turakhia tried a bit of an experiment, asking engineers (reportedly nearly 1000 of them) to stop using their IDEs (Integrated Development Environments) for two weeks. This was ostensibly to see what workflows could be automated and what life is like at a company like Coinbase during this kind of coding freeze.

Let’s face it, AI is doing a lot of the coding everywhere. But Chintan Turakhia emerges as a front-runner in this regard.

What People Are Saying – Or Not Saying

It’s also interesting that when I searched for ground-level input, I didn’t find much, and Ry Walker’s post backed that up, articulating it this way:

“Practitioner discussion of Forge and Mux specifically is thin as of June 11, 2026. The Coinbase Mux engineering post was submitted to Hacker News in May 2026 but drew no comments, and searches of HN and X surfaced no verbatim engineer testimonials about Forge under any of its three names — the substantive HN threads about Coinbase and AI date to the 2025 adoption-mandate controversy, before Forge was publicly named. All detailed accounts of the system come from Coinbase leadership (Chintan Turakhia) via the Linear case study and podcast appearances, not from rank-and-file engineers. The absence of independent practitioner voices is worth noting when weighing the disclosed metrics.”

The feedback is certainly “thin,” and if you don’t work at Coinbase, you don’t know very much about how all of this was received, although Turakhia has made some comments to reporters.

“There was a time where we didn't have, really, sort of any limits on AI spend, and yes, the spend went up,” Turakhia said, according to reporting by Laura Bratton for The Information. “Now, our spend has gone down while our token usage continues to go up.”

Bratton sums up some of the context of the phenomenon this way:

“Coinbase and some other tech-savvy firms such as Shopify and Ramp have built their own AI coding agents for internal use, giving their employees an alternative to pricey options from firms like Anthropic and OpenAI.”

We can see from examples like this that agentic AI is a “real thing,” and it’s here to stay.

Forge is quite a case study in agentic AI. This type of system is likely to become a blueprint for others, as companies scramble to get on the AI bandwagon, to stay ahead of the competition, and to leverage technologies more powerful than anything that we have seen before.