Every time you switch to a new phone or a new smartwatch, your entire running history gets wiped out. The "Run with Full Clarity" solution starts with screenshots, bypasses the closed interfaces of different platforms to implement cross-platform archiving.
Running is a matter that requires persistence, but sports data is not. Every time a runner changes to a new mobile phone, a new watch, or a different running app, the records accumulated over years cannot be transferred; if they stay on the original platform, it means they have to continue tolerating the inconvenient parts that do not fit their usage habits. In the mutually closed ecosystems of Huawei Health, Apple Health, Xiaomi, Keep and Garmin, this is a dead end — cross-platform aggregation of data can never be realized by relying on the cooperation of these platforms.
The WeChat mini-program "R2Know" adopts a completely different solution: instead of asking platforms to open their interfaces, it starts with the running screenshots taken by runners. 36Kr has learned that the official version of the product has been approved and launched, the ICP filing has been completed, and the core process of "upload screenshot - automatic identification - archiving - ranking entry" has been fully verified. The product is developed by an independent developer, and has not received any financing so far.
Every time you change a device, all your historical records get reset to zero
Ray, the founder, graduated from South China University of Technology with a master's degree. He now works in human resources at a listed company, and is also an ordinary runner. Over the years of running, he has used sports devices from Xiaomi, Apple, Huawei and Garmin, and tried several different running apps — every time he switched to a new device or app, his records were stuck on the old platform and could not be moved.
What really pushed him to develop this product was an injury. After he got injured accidentally and stopped running for a month, what he wanted most during that period was someone to tell him two things: whether his running posture and plan were correct, and whether the risk of injury was high. When he tried to review the relationship between his total running volume, pace and injuries over the years, he found that the data could not be pieced together at all.
This is not a situation unique to him. Runners' data are scattered on various platforms, and all platforms regard such data as assets to lock in users — handing over user data to competitors does not conform to the business stance of any platform. So it is not that no one wants to do this, but no one is willing to do it. "Inaccessible data" has become a structural problem that no one has touched for a long time.
A more hidden problem is that records do not equal guidance. What he wants is not just a curve chart, but a judgment before going out every day: under the current temperature, humidity and the physical state of his legs, should he go running today, and how long or how far should he run.
The underlying market of this business is quite large. According to Fastdata's *White Paper on China Road Running Events 2025*, the number of runners in China reached 163 million in 2024, and the scale of running-related industries exceeded 400 billion yuan. In April 2025, the monthly active users of running and fitness platforms reached nearly 60 million. The more devices and running apps a runner uses, the more fragmented their personal running records will be.
Ray's solution is to bypass the requirement for open interfaces. After a runner finishes a run on any sports app, they only need to upload the screenshot to the mini-program to complete a cross-platform archiving. Manual entry is also supported if no screenshot is available.
Technically, the product does not use traditional OCR to extract characters, but lets large language models conduct visual understanding: it automatically extracts core fields such as date, distance, duration, pace, heart rate and cadence. The recognition results can be checked and corrected by users, and the server will then perform rationality verification, so that the same screenshot and the same set of data will not be repeatedly stored and counted. All records are unified into a set of core fields, which are stored in the user's personal running file for long-term retention.
What is archived into the file is not only running data. After submitting the record, the temperature and humidity of the day as well as the physical state after running can be saved together by clicking a few options in a few seconds. It will not make judgments for you for now, but next time you run slower with the same weather and the same distance, you will at least have a reference coordinate to compare with.
The value of this path is that it does not rely on the open interfaces of any platform. Platforms cannot block it, nor do they have reason to do so — giving user data to competitors does not conform to the stance of any platform. "All platforms are guarding their exclusive interfaces, and we want to be the USB-C for runners — one port that connects all devices," Ray said.
The runners with the largest running volume should not always top the ranking list
He does not intend to make the product another list that only compares who runs more. A common problem in the industry is that people who blindly pursue high running volume dominate all rankings: the runner who runs the most will always top the list, and new users will give up as soon as they open the app for the first time. To solve this problem, he has designed two reverse mechanisms.
The first one is stratification by ability. With years of experience in compensation and performance management, Ray introduced the logic of hierarchical and classified performance appraisal in enterprises: the system automatically classifies users into entry, intermediate and hardcore levels according to their pace and cumulative mileage. Users cannot choose their own level, and the ranking list is only sorted within the same level, so new runners can also have their own position in their corresponding level.
The second one is the point coefficient that decreases with consecutive running days. The more consecutive days a runner keeps running, the lower the point coefficient per unit running volume: 1.0 for the 1st to 2nd day, 0.7 for the 3rd day, 0.5 for the 4th day, 0.3 for the 5th day, 0.2 from the 6th day onwards, and returns to 1.0 after one rest day. The running volume itself will not decrease, this mechanism only punishes the behavior of "continuously piling up running volume".
Points cannot be exchanged for goods, and there is no point mall. Points are only used as one dimension of the weekly ranking as well as cumulative records. Ray's explanation is that once points can be exchanged for benefits, there will be motivation to brush points, and this anti-involution mechanism will collapse first.
The product does not target a specific group of people. Runners who have switched between different brands of sports devices want their historical records not to be lost; intermediate runners want opponents at the same level; new runners are afraid of getting injured and want to see their own progress; people who run alone for a long time only need a reference system — these four types of demands all point to the same thing: compete with yourself first, then compete with people at the same level.
Completed the launch all by one person, AI complements the other half of the work
Up to now, all functions of the product can be verified item by item: the official version of the WeChat mini-program has been launched, and users can directly search for "R2Know" on WeChat to use it; the back end is deployed on Tencent Cloud, and all functions including login, record, statistics, ranking, points and achievement system have been fully tested; the ICP filing has been completed, and the software copyright application is in progress.
Before the launch, Ray talked to 7 runners who had experience in switching between different brands of sports devices. 5 of them said they would use the product, and 2 of them directly asked when it would be available. The two most frequently asked questions are: After changing to a new device, what should I do with my historical data? After finishing a run, is there any other output besides the curve chart?
The second question is exactly what he will work on next. The product has reserved a position for AI analysis: by combining training load, recovery status and environmental factors, it can answer the question of "how should I run today". He wants to make it more like a personal coach — it will not provide a universal training plan for all users on the whole network, but only provide the set of training suggestions generated from the user's own years of running records. This function is temporarily restricted by the rules for individual subject mini-programs and is not open to the public, and will be launched after the business subject is upgraded.
There is only one person in the team. Ray completed product definition, front-end and back-end development, cloud deployment and filing by himself, and AI complements the other three parts: coding, visual recognition and knowledge base construction. The direct result of this combination is low cost: with a single-machine architecture, the monthly service cost is only at the level of hundreds of yuan, and there is no need to bear the team burden during the verification period.
In terms of business model, all core functions are completely free at the current stage. His judgment is that it is not difficult to get runners to say "I am willing to use it", but the hard part is to make them keep using it for data archiving. He will leave the monetization direction for later, and wait for the accumulated user data to guide the business. In the medium term, the product may expand to team scenarios such as running groups, competitions and corporate health. The project has not received any financing for now, and has not started financing yet. If he wants to seek investment in the future, he will value the capital that can bring industrial resources — including running and sports health ecosystems, corporate health, content and coaching institutions, rather than just pure funds.
"This project is born from the real pain point of runners themselves," Ray said. Only when the data is connected first, can runners truly understand their own running status. Next, he will fully verify the demands, and then move forward step by step.