GitHub Copilot moved from technical preview to general availability on June 21, 2022. After a year in preview, the product that started as an experiment in AI-assisted coding became a commercial product. The developer reaction to GA was markedly positive.
From technical preview to product
The technical preview, which ran from June 2021, gave GitHub data on how developers actually used AI code suggestions: which completions they accepted, which they rejected, and in which contexts. This data informed the GA model and the user experience. The GA product has better acceptance rates on completions for common languages and is meaningfully improved on test generation compared to the early preview.
The $10/month product decision
$10 per month for individual developers was priced to be affordable to individuals without requiring corporate procurement. The decision to price it this way rather than as enterprise-only reflects GitHub's strategy: individual adoption drives enterprise adoption. A developer who has been using Copilot personally brings the tool to their team, which becomes a purchasing conversation.
The Codex model underneath
GitHub Copilot is powered by OpenAI Codex, a derivative of GPT-3 fine-tuned on code. The model was trained on public code repositories on GitHub, which raises the questions about licence compliance that were already in legal proceedings at GA. The choice of OpenAI as the model provider rather than a model developed internally reflects the acceleration-at-the-cost-of-control trade-off that GitHub and Microsoft made.
Early enterprise response
Enterprise security and legal teams' first response to Copilot GA was often restrictive: concern about IP ownership of suggested code, concern about suggested code derived from GPL-licensed repositories, and concern about code containing company-confidential patterns being sent to an external API. These concerns were legitimate and the resolution took time. By the end of 2022, most large enterprises had either approved Copilot for development use or had explicit policies defining the conditions under which it was approved.