New: The State of Fitness, Q2 2026. Get the free report

Guide

What members actually use AI coaching for

AI coaching earns repeat use when it turns member context into a useful next step, adapts to real life, and knows when a human should take over.

Members have a reason to return to AI coaching when it turns their information into a doable next step and adapts when their plans change. The product has to earn another interaction each time. That means less fascination with AI itself and more attention to whether the member gets a timely, understandable next step without losing human support.

Key takeaways

  • The strongest member-facing job is turning complex fitness information into one clear, feasible next action. Source ↗
  • Adaptation matters because useful coaching has to work on ordinary days, not just when motivation and time are abundant. Source ↗
  • Trust improves when operators clearly frame AI as a tool instead of encouraging members to treat it like a human expert. Source ↗
  • Good handoffs let AI notice, interpret and prompt while preserving an easy path to a coach or staff member. Source ↗
Cover for What members actually use AI coaching for

What jobs are actually worth giving AI?

Start with interpretation. Members already have workouts, wearable metrics, schedules, goals and plenty of information. Measurement alone leaves too much work with the user, according to Heather McKee. The useful coaching job is turning that information into an action a person can understand and actually take. That is a much better product brief than building another dashboard and asking members to figure out what all the signals mean. [20:01]

“It only becomes a coach when it helps someone understand what to do next.”

Dr. Heather McKee, Dr. Heather McKee, PhD Behavioral Scientist, on Habit Formation and Fitness Motivation · 20:01

The second job is remembering enough context to make the next interaction relevant. AI can ask questions and use member data to personalize programs instead of treating every visit as an isolated event, according to Anton Marchanka. “It could actually ask questions, it could actually use the data to provide those really personalized programs.” That points toward a member experience that remembers goals and feedback, then carries them into the next recommendation. Memory matters when it reduces repetition and helps the service respond to the person standing in front of it. [11:54]

“It could actually ask questions, it could actually use the data to provide those really personalized programs.”

Anton Marchanka, Anton Marchanka (Zing Coach CEO) on What Members Tell AI But Not Trainers · 11:54

Why does adaptation matter more than another perfect plan?

A plan designed for a member's best day will eventually collide with real life. Operators should design for the ordinary Tuesday when work and other commitments compete for limited time, according to McKee. That gives AI a valuable adaptation job: help members shrink, redirect or resume the next action rather than treating deviation as failure. The member does not need another idealized program every time life changes. The system should help preserve forward motion when motivation, time and circumstances are different from whatever the onboarding questionnaire predicted. [5:55]

Good adaptation also means matching guidance to what the member can realistically execute. Competency combines confidence with the ability to perform a task, and that this feeling of capability matters in fitness, according to McKee. Her practical rule is simple: “the fastest route to motivation is make that first action really small, make it easy enough to complete.” AI can be useful here because it can change the next step without turning every adjustment into a major coaching appointment. The point is not endless personalization. It is making action feel possible today. [2:41]

“Because the fastest route to motivation is make that first action really small, make it easy enough to complete.”

Dr. Heather McKee, Dr. Heather McKee, PhD Behavioral Scientist, on Habit Formation and Fitness Motivation · 2:41

Data can support that adaptation, but more data is not automatically better coaching. Spencer Tatum emphasized a problem that should be familiar to anyone who has opened a readiness dashboard and wondered what to do with it: “I think we've kind of have have given people a lot of data without the right interpretations.” Member-facing AI can close that gap by translating signals into relevant choices. The caveat is important: interpretation has to remain appropriately bounded when the recommendation moves beyond ordinary fitness guidance and into higher-stakes territory. [16:01]

“I think we've kind of have given people a lot of data without the right interpretations.”

Spencer Tatum, Spencer Tatum of TCU on Markerless Biomechanics, Wearables, and Athlete Engineering · 16:01

How should disclosure and trust shape the experience?

Trust starts by being clear about what kind of tool the member is dealing with. Ken Ford said conversational systems make it easy for people to attribute human characteristics and capabilities to machines because the dialogue feels natural. “People attribute more to it than it's capable of and attribute human characteristics to it.” For fitness operators, that makes clear disclosure part of good product design. A system should not need to cosplay as a trainer to be useful. Members should understand when AI is generating guidance and where its role stops. [4:36]

“People attribute more to it than it's capable of and attribute human characteristics to it.”

Ken Ford, Ken Ford: 40 Years in AI and Why Chatbots Aren’t Thinking · 4:36

That boundary matters because fluency is not the same thing as reliability. Ford said errors in current popular AI systems arise from how those systems work rather than from a temporary lapse in human-style reasoning. He noted that “when the AI systems make an error, and these are errors, not hallucinations, it's working perfectly.” That changes the trust proposition. The goal should not be to convince a member that an AI coach is an all-knowing expert. It should be to give the tool a defined job, sensible limits and a route to human judgment. [10:49]

“When the AI systems make an error, and these are errors, not hallucinations, it's working perfectly.”

Ken Ford, Ken Ford: 40 Years in AI and Why Chatbots Aren’t Thinking · 10:49

Ford offered a better frame for that relationship. He noted that he prefers augmented intelligence because it keeps the human at the center, rather than presenting AI as a standalone replacement for human capability. Marchanka made a compatible point from the fitness side, emphasizing that AI is not a substitute for human interaction or community. Taken together, those views suggest a useful trust test: members should be able to see what AI is doing for them without being asked to mistake automation for the human relationships their gym already provides. [16:09]

When should AI hand the member to a person?

A member-facing system needs an exit ramp. Karl Foster observed that Perfect AI Chat can be configured around a threshold that determines when an interaction should move to a person. “The way you can configure it is to know when it that threshold kicks in to hand over to you.” That is an important design principle beyond any single product. The experience should not measure success by how long AI keeps the conversation. It should measure whether the member reaches the right next destination, including a coach or staff member when needed. [7:06]

“And the way you can configure it is to know when it that threshold kicks in to hand over to you.”

Karl Foster, Karl Foster (Sport Alliance Head of AI) on Running Your Gym on Autopilot with AI · 7:06

Handoffs can also start from behavior rather than a member explicitly asking for help. Foster explained an example in which a member attends a first class but does not return, creating a prompt for the operator to contact that person. This is where AI can serve the relationship instead of competing with it. The machine notices a meaningful change, but the human carries the conversation when personal outreach makes sense. That separation of jobs gives operators a cleaner model for member-facing AI: detect, interpret and prompt, then escalate deliberately. [7:06]

What operators are doing

Zing is building around continuity across the fitness journey. Marchanka explained a member experience that asks about goals, tailors a plan, draws on wearable information and gathers feedback after a session. He also described communication that continues beyond the gym rather than ending when the workout does. For operators, the useful lesson is not simply to add a chatbot. It is to connect the member's stated goal, current context, completed activity and feedback so the next interaction has a reason to exist instead of starting over from scratch. [16:48]

Sport Alliance is working from a different angle, with member communication tied to specific moments in the lifecycle. Foster observed that Perfect AI Engage and Magic AI are designed to time communications around those moments, while safeguards limit message overload. That restraint matters. “The messages are not overwhelming and they they drive impact when they need to.” More communication is not automatically more useful. A repeatable member experience depends on relevance and timing, especially when several possible triggers compete for the same person's attention. [14:06]

“So the messages are not overwhelming and they drive impact when they need to.”

Karl Foster, Karl Foster (Sport Alliance Head of AI) on Running Your Gym on Autopilot with AI · 14:06

McKee's work around wearables points toward an even simpler experience: show the member what matters now and make the next action obvious. McKee added that technology should make life easier instead of adding complexity, particularly when raw metrics leave the individual responsible for interpretation. That complements Tatum's emphasis on integration and interpretation across technology and coaching. These approaches differ in scope, but they keep returning to the same member problem: information has limited value until somebody, or something, turns it into an understandable decision. [23:36]

What should operators build for repeat use?

Repeat use has to be earned at the consumer level. Marchanka added that an operator can require staff to use a system, but members have no such obligation: “you cannot force your members to use a bad solution.” That is the standard worth keeping in mind. Member-facing AI should solve a recurring problem quickly enough that returning feels natural. Interpreting today's signals, adjusting today's action, remembering useful context and reconnecting a drifting member all qualify. Making members maintain another complicated tool for the sake of having AI does not. [0:08]

The strongest product philosophy here may be the least futuristic one. Ford emphasized human-centered AI that extends human performance rather than obsessing over a standalone intellect, while McKee described that technology should point clearly toward the next step a person wants to take. That puts the test in plain terms. Give AI the jobs where memory, interpretation, adaptation and timely prompts help. Keep disclosure clear, make escalation easy, and protect the human relationship. Members do not need AI theater. They need help making the next useful decision. [19:20]

Explore the Tech and AI topic hub for more source-backed conversations.

Frequently asked questions

What should an AI fitness coach actually do for members?

An AI fitness coach should help turn information into an understandable next action, rather than simply displaying more metrics. McKee outlined that data becomes coaching when it helps someone know what to do next. Useful jobs therefore include interpretation, practical guidance and adapting the next step to the member's current situation. Hear the source ↗

Should a gym tell members when they are talking to AI?

Clear disclosure supports an appropriately bounded relationship with the technology. Ford pointed out that natural dialogue makes it easy for people to attribute human characteristics and abilities to AI systems. Operators should therefore make the system's role understandable rather than relying on members to distinguish fluent conversation from actual human judgment. Hear the source ↗

When should an AI coach hand a member to a human?

The handoff should happen when the interaction crosses a defined threshold where human attention makes more sense. Foster pointed out that AI chat can be configured to recognize that threshold and hand the interaction over. That keeps automation focused on the jobs it can handle while preserving a clear route to staff support. Hear the source ↗

Can AI make wearable data more useful for gym members?

AI can help when it reduces the interpretation burden instead of adding another layer of metrics. McKee recalled that wearable information needs to be timely, understandable, meaningful and tied to a feasible next step. That makes interpretation and action selection more useful member-facing jobs than simply generating a more complicated dashboard. Hear the source ↗

The Future of Fitness Newsletter

Free, every week, plus our quarterly State of Fitness report. Leave anytime.