--- title: "How canvases make agentic workflows visible, steerable, and cost-efficient" url: https://daily.dev/posts/how-canvases-make-agentic-workflows-visible-steerable-and-cost-efficient-krjkyqawf source_url: https://github.blog/ai-and-ml/github-copilot/how-canvases-make-agentic-workflows-visible-steerable-and-cost-efficient type: article source: "GitHub Blog" published: 2026-08-17T16:02:53.356Z updated: 2026-08-17T16:03:20.189Z tags: ["github", "ai-agents", "productivity", "code-review"] reading_time: 5 upvotes: 1 comments: 0 language: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # How canvases make agentic workflows visible, steerable, and cost-efficient **[GitHub Blog](https://daily.dev/sources/ghblog)** · 5 min read · 1 upvotes · 0 comments ## Summary A GitHub Developer Advocate describes using 'canvases' in the GitHub Copilot app as a durable, shared surface for agentic workflows, contrasting them with chat interfaces that bury important context in a long scroll. He shares two examples he built: Java Modernization Studio (for assessing, planning, and validating Java migrations) and Site Studio (for managing personal site content), both available in the awesome-copilot repo. He outlines a blueprint of defining workflow states, surfacing key decisions, persisting progress, and keeping human approval points explicit. He notes the cost of building these canvases (about 2,000-3,000 AI credits each) but argues the investment pays off for repeated workflows by reducing rework and coordination overhead. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is a canvas in the GitHub Copilot app and how does it differ from chat? A canvas is a durable, shared surface where developers and Copilot agents interact on persistent workflow state, unlike chat which becomes a long scroll of instructions and logs. Canvases make the plan, decisions, validations, and approval moments explicit and inspectable instead of buried in chat history, letting humans steer and approve as agent work unfolds. _daily.dev surfaces practical writeups like this for developers shaping their own agentic coding workflows._ ### How much did it cost in AI credits to build a Copilot canvas like Java Modernization Studio? Building a Java Modernization Studio canvas cost about 3,000 AI credits, while a content-focused Site Studio canvas cost about 2,000 AI credits. These are described as upfront investments that pay back over repeated workflows by reducing repeated prompting, context loss, and rework. _Developers weighing AI credit spend against workflow gains can track real-world cost breakdowns on daily.dev._ ## Similar posts on daily.dev - [Expanded technical preview availability for the GitHub Copilot app](https://daily.dev/posts/expanded-technical-preview-availability-for-the-github-copilot-app-w758b2dzo) · GitHub Changelog · 0 upvotes · 0 comments --- Tags: [#github](https://daily.dev/tags/github), [#ai-agents](https://daily.dev/tags/ai-agents), [#productivity](https://daily.dev/tags/productivity), [#code-review](https://daily.dev/tags/code-review) [View this post on daily.dev](https://daily.dev/posts/how-canvases-make-agentic-workflows-visible-steerable-and-cost-efficient-krjkyqawf)