Strava has, over several years, become increasingly protective of how its users use the data collected by the platform. It\u2019s now quite hard for any third party app, or rider looking ot use a third party app, to easily use data collected by Strava. Unless, as of June 1, you\u2019re planning to consult Claude.<\/p>\n
Claude is, of course, Anthropic\u2019s massive AI platform. After Strava trialed its own AI interface within the app, with, uh, limited success. Now the social training platform is turning to the big guns to satisfy user\u2019s AI demands.<\/p>\n
User driven?<\/p>\n
What exactly is new? Strava\u2019s new\u00a0\u201cModel Context Protocol (MCP) connector lets subscribers query their own Strava data through Claude and get in-depth insights about their training.\u201d The optional connector is available to subscribers only. It\u2019s started rolling out, so it could already be available in your area.<\/p>\n
Strava claims<\/a> this new partnership is driven by user demand. That could very well be true.<\/p>\n \u201cAthletes have been telling us, in increasingly creative ways, that they want more ways to analyze their own training data,\u201d said Ryan Dixon, VP, Partnerships & Developer Relations at Strava. \u201cThey\u2019ve been doing it with spreadsheets, exports, and third-party scripts for years. The MCP connector gives them a far more efficient, safer tool while keeping the athlete in control, representing a step-change in our subscription offering.\u201d<\/p>\n Strava users gain ease of AI assistance<\/p>\n As Dixon hints at, using any AI to do whatever you want to do with your Strava history, whether that\u2019s develop a training program or asses what training is working or, well, pretty much anything, was quite labour intensive. Basically, you had to slowly export your data from Strava, then upload it you the LLM of your choice. And then do all of this again the next time you wanted to ask another question.<\/p>\n \u201cThe new MCP makes that process much easier by providing a seamless, secure experience connecting Claude to a breadth of activity data, including: access to full stream data \u2013 per second heart rate and apce, GPS data for geographic analysis of activities, power data for cycling, and access to your club and event data.\u201d And probably some other things, depending what questions you want answered.<\/p>\n Strava: API vs AI<\/p>\n This Claude MCP is a significant swing away from a recent push to \u201cprotect users data.\u201d For those of us around Strava for long enough, will remember that the platform was initially quite open. In recent years, the social fitness platform slowly made it more and more difficult for third party applications and platforms to access and use data, and for Strava users to take their data to another platform in an at all streamlined way. That escalated when Strava overhauled it\u2019s application programming interface (API)<\/a> back in 2024. The update made it much more difficult for any third party app to work off of Strava data, even when users sought to do so.<\/p>\n It\u2019s the last part that makes Strava\u2019s partnership with Claude a little surprising. This is not quite a swing back towards the initial era of openess. It\u2019s more of a walled openess, where only Claude is granted access. For now, at least. In the FAQ section of this announcement, Strava says \u201cWe\u2019re launching with Anthropic (Claude) and looking to support other AI clients in the future.\u201d<\/p>\n What will Claude have access to?<\/p>\n If you\u2019re interested, but wondering about privacy, here\u2019s Strava\u2019s word on how Claude\u2019s access is limited.<\/p>\n \u201cThe Strava MCP is built by Strava and was designed with security and quality guardrails. As always, your data stays yours. The MCP connector provides read-only access, scoped to your account, and is revocable anytime from Strava settings\u201d<\/p>\n