{"id":700140,"date":"2026-07-19T15:41:26","date_gmt":"2026-07-19T15:41:26","guid":{"rendered":"https:\/\/www.newsbeep.com\/uk\/700140\/"},"modified":"2026-07-19T15:41:26","modified_gmt":"2026-07-19T15:41:26","slug":"teams-that-build-ai-programmes-will-win-races-how-artificial-intelligence-is-changing-the-tour-de-france-and-pro-cycling","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/uk\/700140\/","title":{"rendered":"‘Teams that build AI programmes will win races’ \u2013 how artificial intelligence is changing the Tour de France and pro cycling"},"content":{"rendered":"

\u201cI have a theory that\u00a0it\u2019s\u00a0all smoke and mirrors.\u00a0UAE Team Emirates-XRG\u00a0has\u00a0reportedly been\u00a0using it for years. Visma-Lease a Bike\u00a0has a link with Mistral.\u00a0Ineos has brought\u00a0Netcompany\u00a0onboard.\u00a0But apart from their press releases, we have no idea how\u00a0they\u2019re\u00a0using it, which is odd,\u00a0because people in cycling are a) bad at keeping secrets and b) have something to sell.\u201d\u00a0<\/p>\n

That\u2019s\u00a0Frank Overton, founder of\u00a0FasCat\u00a0Coaching, when I ask him about the professionals\u2019 use of artificial intelligence\u00a0(AI).<\/p>\n

Overton\u2019s\u00a0ahead of the\u00a0data-driven\u00a0curve, with his coaching outfit\u00a0offering\u00a0assisted\u00a0input for the past few years\u00a0off the back of his proprietary AI\u00a0work.\u00a0Overton knows his stuff. And is prophetic, as AI input requests to myriad Tour de France teams<\/a>\u00a0fell on deaf ears.\u00a0Until, that is, the\u00a0progressive\u00a0folk at Team Picnic-PostNL\u00a0allowed us into their artificial world.<\/p>\n

Specific training, better results\u00a0
\n\"\"Picnic-PostNL has been capturing rider performance data since 2012 and uses AI to comb through the data. Tim de Waele\/Getty Images<\/p>\n

\u201cAt Team Picnic PostNL, data analysis and AI play an increasingly important role in optimising both performance and development,\u201d explains\u00a0Narelle Neumann,\u00a0the team\u2019s\u00a0head of science. \u201cEvery rider uploads daily training files\u00a0containing\u00a0thousands of rows of data, including power output, heart rate, cadence,\u00a0speed\u00a0and altitude information. During a season, this results in millions of data points.\u00a0\u00a0<\/p>\n

\u201cTraining files are\u00a0then\u00a0processed through analytical models that\u00a0monitor, for example, the relationship between power output<\/a> and heart rate over time. These models correct for environmental influences such as temperature and altitude, allowing to better understand changes in fitness and physiological condition. By filtering out external factors, we can\u00a0identify\u00a0more reliable trends in performance and recovery.\u00a0<\/p>\n

\u201cIn addition, we also collect subjective data via the logbook such as perceived recovery,\u00a0fitness\u00a0and RPE [Rate of Perceived Exertion<\/a>]. The team started systematically collecting and storing this information from 2012, resulting in a highly valuable database. Machine learning and advanced analytics help us extract meaningful insights from this large volume of data, enabling better-informed decision making.\u201d\u00a0<\/p>\n

The Dutch team recently cranked up its\u00a0appliance of\u00a0artificial\u00a0science, announcing a multi-year partnership with Dutch technology firm IG&H. Hopefully\u00a0we\u2019ll\u00a0see\u00a0that\u00a0collaboration\u00a0bear fruit, because there are rumours that Picnic could fold without investment.\u00a0\u00a0<\/p>\n

Ditching TrainingPeaks for Vekta
\n\"\"Decathlon CMA CGM is one of several teams using the Vekta AI coaching platform. <\/p>\n

On a sounder fiscal footing are\u00a0Paul Seixas<\/a>\u2019 Decathlon CMA CGM Team. Along with\u00a0Lidl-Trek,\u00a0TotalEnergies\u00a0and Jayco-AlUla,\u00a0they were so convinced\u00a0that\u00a0AI coaching platform\u00a0Vekta\u00a0could improve their performance\u00a0that they ditched legacy\u00a0power brand\u00a0TrainingPeaks<\/a>.\u00a0Why?\u00a0<\/p>\n

\u201cVekta\u00a0has multiple uses. Refining training is one of them,\u201d says Vekta co-founder Paul-Antoine Girard. \u201cLet\u2019s\u00a0use interval workouts as an example because modern-day road cycling is all about intervals! Say today a Lidl-Trek rider rode six 10-minute efforts within a four-hour ride. Immediately,\u00a0Vekta\u00a0will compare the results with similar sessions in the past to see how much the rider has\u00a0(hopefully)\u00a0improved\u00a0and whether they need to adapt that interval.\u201d\u00a0<\/p>\n

Vekta\u00a0identifies\u00a0the demands of each workout to\u00a0determine\u00a0how precisely the session targets\u00a0the physiological goal, be it endurance, VO2 max<\/a> or sprint power.\u00a0That might not sound too different from\u00a0TrainingPeaks, but one of the significant differences\u00a0arises from the\u00a0training model used, as\u00a0Vekta\u2019s\u00a0dispensed with functional threshold power<\/a>\u00a0(FTP)\u00a0to hang training zones. Instead, they focus on critical\u00a0power<\/a> and \u2018W\u2019 prime.\u00a0\u00a0<\/p>\n

\"\"Vekta’s AI software focuses on critical power and W Prime, which measures the capacity to power sprints. Loic Venance \/ AFP via Getty Images<\/p>\n

Critical power is the highest power output a rider can sustain indefinitely without accumulating fatigue. \u2018W\u2019 is the finite reserve above that threshold; in other words, the anaerobic battery that powers accelerations and sprints.\u00a0\u00a0<\/p>\n

\u201cIn our eyes, it gives a more complete view of performance than FTP,\u201d says\u00a0Girard. \u201cYou can better unpick the physiological [cost of]\u00a0an effort. How deep the rider went and how deep they can go.\u201d\u00a0<\/p>\n

As an example,\u00a0Vekta\u00a0says this detail\u00a0and AI processing\u00a0more accurately captures work done. \u201cA mountain stage with repeated short climbs carries a very different physiological cost to a long rolling stage with a final summit finish, even if the average power numbers are similar,\u201d says\u00a0Girard.\u00a0\u201cWe feel critical power and \u2018W\u2019 prime shows this\u00a0more clearly.\u201d\u00a0<\/p>\n

\u201cVekta\u00a0also highlights the durability of a rider,\u201d he adds.\u00a0\u201cThis is clearly a buzzword in cycling at the moment\u00a0and\u00a0shows\u00a0how well they can sustain their performance\u00a0after many hours of racing.\u201d\u00a0\u00a0<\/p>\n

\"\"The top riders can sustain performance after many hours of racing. A.S.O.\/Thomas Maheux<\/p>\n

Durability is one of the defining characteristics of elite professional cyclists. Research by coach and sports scientist James Spragg\u00a0discovered\u00a0that while under-23 riders can match professionals when fresh, the biggest difference appears after significant fatigue accumulates. Top riders, particularly\u00a0GrandTour\u00a0contenders,\u00a0maintain\u00a0high power outputs even after thousands of kilojoules of work,\u00a0demonstrating\u00a0superior fatigue resistance.\u00a0How is a multi-faceted process that combines nutrition, functional strength, aerobic endurance work and having the right genetics.\u00a0<\/p>\n

\u201cVekta\u00a0allows the user to analyse power files according to different levels of fatigue [from 0 to 50kJ\/kg],\u201d says Girard, \u201cplus the current kilojoules\u00a0expended\u00a0is displayed before an interval during a workout.\u00a0It\u2019s\u00a0all about precise training.\u201d\u00a0<\/p>\n

The AI crystal ball\u00a0
\n\"EFData from devices such as Whoop can be used to power real-time analysis of a rider’s capabilities. WHOOP<\/p>\n

Alongside precise training, Vekta helps predicting performance, adds Girard.\u00a0\u201cFor each of the 21 stages of the Tour de France<\/a>, the teams will upload the GPX course files.\u00a0With the individual rider\u2019s power data, you can then see what\u00a0they\u2019re\u00a0capable of based on the\u00a0parcours.\u201d\u00a0<\/p>\n

As an example,\u00a0if\u00a0it\u2019s\u00a0a sprint stage with rolling terrain,\u00a0Vekta\u00a0can\u00a0determine\u00a0which riders\u2019 strengths it might play to.\u00a0During\u00a0the race,\u00a0it\u2019s\u00a0also possible\u00a0to\u00a0update\u00a0recovery metrics from\u00a0tech like Whoop<\/a> and Oura<\/a>. That means heart rate variability<\/a>,\u00a0sleep<\/a>\u00a0and psychology data from answering a wellness questionnaire.\u00a0All of this goes into the AI engine to give the coach a clear picture of what the rider might be capable of\u00a0in\u00a0real time.\u00a0<\/p>\n

Predictive modelling is nothing new, of course,\u00a0with the likes of Best Bike Split\u00a0used by many teams,\u00a0ostensibly\u00a0to\u00a0optimise equipment\u00a0and pacing\u00a0selection.\u00a0Best Bike Split tell us they\u00a0don\u2019t\u00a0directly use AI for modelling \u201cthough we do use some machine learning for data analysis\u201d,\u00a0plus\u00a0they\u2019ve\u00a0recently added an AI workout generator, primarily aimed at recreational riders.\u00a0<\/p>\n

Analysing the Queen Stage
\n\"\"Riders will have already climbed around 4,500m on stage 20 of the 2026 Tour de France by the time they tackle the Col de Sarenne. <\/p>\n

Overton\u00a0is a keen fan of AI for modelling. Take\u00a0stage 20 of this year\u2019s Tour de France<\/a>, which\u00a0is the Queen Stage and, unless Tadej Poga\u010dar\u2019s long since wrapped up his record-equalling fifth title, will decide the race.\u00a0It\u2019s\u00a0171km from Bourg\u00a0d\u2019Oisans\u00a0to the second successive Alpe\u00a0d\u2019Huez<\/a>\u00a0finish, with\u00a0a brutal\u00a0parcours\u00a0that ticks off Croix de Fer,\u00a0T\u00e9l\u00e9graphe,\u00a0Galibier, Col de\u00a0Sarenne\u00a0on the way. By the time the GC contenders touch the base of Alpe\u00a0d\u2019Huez,\u00a0they\u2019ll\u00a0have\u00a0roughly 4,500m of climbing in their legs plus 3,032km of racing.\u00a0<\/p>\n

\u201cHow would we use AI to analyse the performance and the training that it will take to win? In layers,\u201d says\u00a0Overton.\u00a0\u201cLayer\u00a0one: peak power up Alpe\u00a0d\u2019Huez.\u201d This, stresses Overton, is the\u00a0obvious starting point.\u00a0\u201cWhat\u2019s each GC contender\u2019s predicted ascent power up the final climb? That comes from modelling each rider\u2019s best 30-to-45-minute power outputs, scaled to W\/kg on the day.\u201d\u00a0<\/p>\n

\u201cLayer\u00a0two:\u00a0power output under fatigue.\u202fThis is where layer\u00a0one\u00a0falls apart for most of the field. The relevant question\u00a0isn\u2019t\u00a0what a rider can hold up Alpe\u00a0d\u2019Huez\u00a0fresh.\u00a0It\u2019s\u00a0what they can hold after 19 stages of racing and four hard climbs in the same day.\u201d\u00a0<\/p>\n

That\u2019s\u00a0a durability question: how much power does\u00a0a\u00a0rider sustain at threshold and above once\u00a0they\u2019re\u00a03,000, 3,500, 4,000kJ deep into a stage?\u00a0Then\u00a0there\u2019s\u00a0the recovery\u00a0factor. For instance, what recovery metrics do they need to hit in weeks one and two to thrive in week three? Some riders barely fade. Others lose 10 to 15% of their threshold. That difference wins and loses Tours.\u00a0<\/p>\n

\"\"Frank Overton uses historic data to produce a fatigue-adjusted power output number for each rider on a final climb. A.S.O.\/Thomas Maheux<\/p>\n

Overton says they\u00a0can run an\u00a0historical analysis on each GC contender\u2019s data, particularly their\u00a0previous\u00a0Tours de France dataset, and quantify exactly how their power decays as\u00a0kilojoules\u00a0accumulate within a stage. That produces a fatigue-adjusted power prediction for the final climb. Not a fresh-leg fantasy.\u00a0<\/p>\n

\u201cLayer\u00a0three: GC modelling.\u202fThis is where we\u00a0combine\u00a0layers one and\u00a0two. Given the race situation entering the stage, we can model the power output a contender needs to hold on Alpe\u00a0d\u2019Huez\u00a0to (a) close a specific GC time gap or (b) defend a specific GC lead.\u201d\u00a0<\/p>\n

This involves assessing\u00a0the\u00a0likely GC\u00a0dynamics on the four\u00a0previous\u00a0climbs,\u00a0plus\u00a0the energy costs and the rider\u2019s historical fatigue profile. Feed all that information into the equation and see if the\u00a0maths says\u00a0the\u00a0lead is\u00a0defensible\u00a0or the\u00a0deficit closeable. Whether\u00a0it\u2019s\u00a0the former or latter dictates what questions need answering.\u00a0If in the head, what\u00a0wattage\u00a0is\u00a0required\u00a0and for how many minutes\u00a0to\u00a0maintain\u00a0that advantage?\u00a0If chasing, how far from the top does the rider need to attack and hold what wattage? How does this change team tactics? If the model predicts the rider\u00a0can\u2019t\u00a0make up a GC deficit,\u00a0that\u2019s\u00a0the kind of knowledge a Directeur Sportif and a team\u2019s performance staff want before the stage starts.\u00a0<\/p>\n

\u201cThese\u00a0are exactly the types of questions that properly trained AI models can answer very well and in real time,\u201d adds Overton.\u00a0\u201cThat said, analysing peak power is the easy part. Prescribing the exact training a rider needs to meet those demands 11 months in advance is the secret sauce. The durability work, the fatigue resistance, the balance of training and race volume against a rider\u2019s recovery kinetics.\u00a0That\u2019s\u00a0the marginal gain that will unlock the next Tour de France winner.\u201d\u00a0<\/p>\n

To that end, Overton says your AI system is only as good as the information you feed it.\u00a0\u201cThere\u2019s\u00a0an illusion that the likes of ChatGPT and Claude deliver good training advice.\u00a0But\u00a0it\u2019s\u00a0not trained, meaning it\u00a0doesn\u2019t\u00a0answer like a coach or AI using a relevant dataset.\u00a0It\u00a0takes\u00a0information from the general internet, meaning\u00a0there’s\u00a0no filter.\u00a0So, information from Reddit, forums\u2026 everywhere.\u00a0There\u2019s\u00a0no thought leadership.\u201d\u00a0<\/p>\n

Overton says his AI\u00a0CoachCat\u00a0app\u2019s\u00a0the result of his proprietary work\u00a0that\u2019s\u00a0involved years of\u00a0uploading\u00a0rider training data, his blogs,\u00a0podcasts\u00a0and training plans.\u00a0\u201cFor AI, data\u00a0is\u00a0rocket\u00a0fuel. The more you have, the more insight it can provide, the better it is.\u00a0But it must be the right data.\u201d\u00a0\u00a0<\/p>\n

\"\"Picnic-PostNL developed an app to support the team’s nutrition planning. A.S.O.\/Charly L\u00f3pez<\/p>\n

So, you must feed the correct\u00a0ingredients,\u00a0or the recipe will be a disaster.\u00a0Cue the\u00a0Picnic team\u00a0working with\u00a0IG&H\u00a0to develop\u00a0an application to support nutrition planning during training and races.\u00a0\u201cUsing data models, the app predicts riders\u2019 energy expenditure during races based on\u00a0parcours\u00a0information such as elevation and distance,\u201d says\u00a0Neumann.\u00a0\u201cSimilar models estimate energy expenditure during training sessions based on workout descriptions. Based on these predictions, the app helps nutritionists and cooks to plan strategies and daily meals, ensuring that energy intake is aligned with the demands of training,\u00a0racing\u00a0and recovery.\u201d\u00a0<\/p>\n

It\u2019s\u00a0a similar idea to that employed by\u00a0Visma-Lease a Bike\u00a0for a few years now.\u00a0Ahead\u00a0of the Tour de France, the nutrition team analyse the\u00a0route and make a series of nutritional predictions for each stage based on\u00a0team line-up\u00a0and a rider\u2019s physiology. For each stage and the potential role and goal of each rider, the team then predict their power output and intensity of effort, and so could predict energy expenditure and carbohydrate use. With the advent of AI, which also accounts for weather conditions, Visma\u00a0suggests accuracy of nutrition predictions\u00a0varies\u00a0from\u00a0a moderate\u00a052% to\u00a0an excellent\u00a082%.\u00a0\u00a0<\/p>\n

Tech-led scouting\u00a0
\n\"\"Teams are using AI to analyse performance and identify the next Paul Seixas. A.S.O.\/Thomas Maheux<\/p>\n

AI\u00a0isn\u2019t\u00a0solely about maximising the moment. It\u00a0possesses\u00a0great potential\u00a0in\u00a0identifying\u00a0tomorrow\u2019s\u00a0Paul Seixas or Mads Pedersen.\u00a0\u201cLater this month, we\u2019re launching an\u00a0academy with one of the\u00a0WorldTour\u00a0teams,\u201d says\u00a0Girard. \u201cWe\u2019ll be using a big data approach to scouting and\u00a0try to\u00a0predict\u00a0how good they\u00a0could\u00a0be in\u00a0two\u00a0or\u00a0three years\u2019\u00a0time.\u201d\u00a0<\/p>\n

They\u2019re\u00a0not the only ones looking to refine talent identification through AI.\u00a0Earlier this year, then Ineos Grenadiers (now\u00a0Netcompany-Ineos) pronounced\u00a0they\u2019d\u00a0partnered with\u00a0Swansea University\u00a0to explore how data science and\u00a0AI\u00a0could transform talent identification. At the heart of the collaboration is a project to develop\u00a0\u2018digital twins\u2019 \u2013\u00a0data-driven profiles of riders\u00a0\u2013\u00a0using a combination of the team\u2019s internal performance metrics and publicly available race data.\u00a0\u00a0<\/p>\n

The aim is to build an automated system capable of tracking junior riders and flagging\u00a0up\u00a0standout performances, helping the\u00a0British\u00a0team\u00a0to unearth future stars.\u00a0Swansea\u2019s A-STEM research centre, which has\u00a0a track record of working with\u00a0elite sports organisations,\u00a0including Swansea FC and the Ospreys rugby union outfit,\u00a0leads\u00a0the academic side\u00a0with a\u00a0dedicated PhD student\u00a0working\u00a0directly with the team.\u00a0\u00a0<\/p>\n

We contacted Swansea and\u00a0Netcompany\u00a0for further\u00a0information\u00a0but the respective outfits were\u00a0unable to comment.\u00a0<\/p>\n

These \u2018digital twins\u2019\u00a0are a growing trend, not only for talent ID but also gear\u00a0selection.\u00a0Teams can\u00a0dispense with\u00a0rider\u00a0mannequins in a wind tunnel and instead use a computer-generated twin based on a rider\u2019s\u00a0body dimensions, riding style, power\u00a0output,\u00a0flexibility\u00a0and aerodynamics. AI\u00a0can then simulate\u00a0different helmets, different wheel depths, different riding positions\u00a0and different weather conditions before the rider ever\u00a0rolls\u00a0onto the road.\u00a0\u00a0<\/p>\n

AI and the Tour de France Femmes
\n\"\"Teams see a specific opportunity to transform women’s pro cycling with AI. Getty Images<\/p>\n

AI also has the potential to transform the Tour de France Femmes<\/a>\u00a0specifically, says Girard.\u00a0\u201cWe\u2019re\u00a0working with [Demi Vollering\u2019s] FDJ United-Suez to better understand how to use AI for women.\u00a0To understand the impact of the menstrual cycle on things like power output and recovery.\u00a0It\u2019ll\u00a0help the riders and their coaches.\u201d\u00a0<\/p>\n

One rider who understands the impact of AI and women-specific feedback is 2024 Olympic road-race champion Kristen Faulkner. The EF Education-Oatly\u00a0athlete studied computer science at\u00a0Harvard and\u00a0utilised\u00a0her academic background to build her own AI. Spending up to 10 hours coding each day, Faulkner\u00a0used\u00a0nine years\u2019 worth of data \u2013 around 4,440 hours of training history \u2013 to synthesis her biometric data that included heart rate, HRV, sleep, weight, power, temperature, training load, menstrual cycle phases, bloodwork and DEXA scans.\u00a0<\/p>\n

\u201cEvery model is trained on my body,\u201d Faulkner penned\u00a0on her website. \u201cEvery finding is specific to my history. And every output is actionable, not just interesting.\u201d\u00a0<\/p>\n

Did it work?\u00a0\u201cI used this to help me prepare for the Pan Am\u00a0Championships, where I won three gold medals this year [one on the road, two on the track],\u201d Faulkner says. \u201cToday, I produced my best 20-minute power ever with training help from this app. AI is going to change women\u2019s performance research from the bottom up and I want to be part of it.\u201d\u00a0<\/p>\n

\"TeamKristen Faulkner developed her own AI to analyse 4,440 hours of her training data stretching back nine years. Oscar del Pozo\/AFP via Getty Images<\/p>\n

With her focus and diligence to\u00a0extract the most\u00a0marginal\u00a0of gains\u00a0from every\u00a0facet\u00a0of performance,\u00a0don\u2019t\u00a0bet against the American regaining her Olympic title in Los Angeles in two years\u2019 time.\u00a0<\/p>\n

All in all, artificial intelligence has the potential to\u00a0accelerate\u00a0cycling \u2013 a sport for whom data is\u00a0already\u00a0king or queen.\u00a0AI\u2019s ability to rapidly analyse every metric generated\u00a0is set to make\u00a0WorldTour\u00a0coaches\u2019 jobs\u00a0easier and more effective, by delivering specific training\u00a0workouts and plans, clearer race\u00a0predictions\u00a0and bespoke feeding advice.\u00a0As Overton\u00a0signs off, \u201cSir Dave Brailsford has said AI will help win the Tour within 5 years. I think\u00a0it\u2019s\u00a0two (or less).\u00a0Within 24 months, every World Tour team will have an AI programme. The teams that build them well will win races. Teams that\u00a0don\u2019t\u00a0will be dropped.\u201d\u00a0<\/p>\n","protected":false},"excerpt":{"rendered":"\u201cI have a theory that\u00a0it\u2019s\u00a0all smoke and mirrors.\u00a0UAE Team Emirates-XRG\u00a0has\u00a0reportedly been\u00a0using it for years. Visma-Lease a Bike\u00a0has a…\n","protected":false},"author":2,"featured_media":700141,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[554,733,4308,86,56,54,55],"class_list":["post-700140","post","type-post","status-publish","format-standard","has-post-thumbnail","category-artificial-intelligence","tag-ai","tag-artificial-intelligence","tag-artificialintelligence","tag-technology","tag-uk","tag-united-kingdom","tag-unitedkingdom"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts\/700140","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/comments?post=700140"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts\/700140\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/media\/700141"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/media?parent=700140"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/categories?post=700140"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/tags?post=700140"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}