GitHub today announced that Copilot for Business, the company’s $19/month enterprise version of its AI-powered code completion tool, is now generally available, after a short beta phase that started last December. Copilot for Business adds features like license management, organization-wide policy management and additional privacy features. Until now, you had to work with GitHub’s sales organization to sign up for the business version, but now there is a self-serve option as well.

“[This] effectively completes our v1 Copilot story,” GitHub CEO Thomas Dohmke told me. “We announced the preview in June 2021 — which feels like ages ago — and then had the general availability last summer. Now we are ready to roll it out to organizations, companies, teams, enterprises — really everybody. In fact, we already have more than 400 organizations that are on Copilot for Business at launch and we see tremendous interest.”

Image Credits: GitHub

GitHub also today announced that Copilot now supports connections over proxy, including those with self-signed certificates, and that its AI-powered Copilot code completion tool is now powered by an improved (OpenAI-powered) model. The team is constantly refining the models and adding new features as they become available in Azure’s OpenAI Service. He likened this process to spec bumps in the hardware business, where the model gets a little bit better and the team can then take that and bring it to Copilot.

As the model gets better, the team is also adding new features, including things like “fill-in-the-middle,” where the model can’t just complete a line but also start adding words in the middle because it knows what sits before and after the current cursor position, for example. Dohmke noted that to do this, the model also looks at related files that you work on and then uses that info to craft its model queries, too. “It not only uses what you type in your own open file but also leverages adjacent files and adjacent information that is available to craft the prompt that’s sent to the model for inference.”

With these latest updates, the team also enabled Copilot — with the help of another model — to recognize common security vulnerabilities in code that comes back from the model. If it finds those, it’ll automatically jump to a suggestion that is more secure.

But it’s not just about getting better, but also about getting faster. Dohmke noted that the team is constantly working to improve latency. GitHub’s data shows that developers quickly get restless when it takes too long for Copilot to generate its code.

Dohmke expects that soon, Copilot will be able to generate 80% of a developer’s code. Today, that number is about 46% across programming languages — and 61% for Java. Today, ChatGPT and Bing can obviously generate entire applications (or at least parts of them) with just a simple prompt. That’s enabled by the same models as Copilot, so the obvious question, which I asked Dohmke, is when these capabilities will also come to Copilot.

“We think it’s a really exciting feature that the Bing team launched and we have nothing to announce today — wink, wink, nudge, nudge — but it’s very exciting.” He did add that for a product like Copilot, the team always wants to help developers do their work more efficiently — and the current models may simply produce code that isn’t correct.