What began as a casual revelation quickly went viral after Google engineer Jaana Dogan shared an eye-opening comparison: a task that took Google teams nearly a year of meetings, planning, and debate was replicated in just one hour by a rival AI tool. The offhand comment struck a nerve online, fuelling fresh conversations about how fast AI coding tools are advancing and what that speed could mean for traditional software development.<\/p>\n
\nI’m not joking and this isn’t funny. We have been trying to build distributed agent orchestrators at Google since last year. There are various options, not everyone is aligned… I gave Claude Code a description of the problem, it generated what we built last year in an hour.
\n\u2014 Jaana Dogan \u30e4\u30ca \u30c9\u30ac\u30f3 (@rakyll) January 2, 2026<\/a><\/p>\n<\/blockquote>\nJaana Dogan, a principal engineer working on Google\u2019s Gemini API, recently shared a moment that caught the tech world\u2019s attention. While testing Anthropic\u2019s Claude Code, an AI tool designed to help write and organise code, she fed it a problem her team had been working on for nearly a year. To her surprise, Claude produced something strikingly similar to Google\u2019s solution in just 60 minutes. \u201cI\u2019m not joking and this isn\u2019t funny. We have been trying to build this since last year. I gave Claude a description, and it generated what we built last year in an hour,\u201d she wrote on X.<\/p>\n
ALSO SEE: 7 ChatGPT Prompts That Helped a Man Lose 27 Kg in Just 3 Months Without Gym<\/a><\/strong><\/p>\n
The task, she explained, involved designing systems that manage multiple AI agents working together, similar to traffic control, but for AI bots. Google\u2019s team had explored several approaches but hadn\u2019t settled on a final design. To fairly test Claude, Dogan avoided using any internal data and instead framed a simplified version of the problem using public ideas. Her prompt was just three short paragraphs, with no proprietary information or hidden tricks.<\/p>\n
\nIt wasn’t a very detailed prompt and it contained no real details given I cannot share anything propriety. I was building a toy version on top of some of the existing ideas to evaluate Claude Code. It was a three paragraph description.
\n\u2014 Jaana Dogan \u30e4\u30ca \u30c9\u30ac\u30f3 (@rakyll) January 3, 2026<\/a><\/p>\n<\/blockquote>\nDogan was clear that Claude\u2019s output wasn\u2019t perfect and would still need refinement. However, she urged skeptics of AI coding tools to test them on problems they understand deeply, where the contrast is most striking. When asked if Google uses Claude Code internally, she said it\u2019s only permitted for open-source work, not internal projects. On whether Gemini would catch up, she replied, \u201cWe are working hard right now. The models and the harness,\u201d signalling active but challenging progress.<\/p>\n
Rather than framing the moment as a threat, Dogan praised the pace of competition and innovation. \u201cClaude Code is impressive work. I\u2019m excited and more motivated to push us all forward,\u201d she wrote. Reflecting on how quickly AI coding has evolved, from single lines in 2022 to rebuilding entire codebases in 2025, she admitted she never expected this speed. Her post crossed 4 million views, with developers pointing out how AI bypasses corporate bureaucracy and accelerates creativity. The takeaway, as many saw it: AI may not replace whiteboards or snacks just yet, but it can certainly replace a year of meetings.<\/p>\n
ALSO SEE: Apple to Launch iPhone 18 Pro Soon: Expected Price, Specifications<\/a><\/strong><\/p>\n