Skip to content
Karl Foster

Writing · articles, interviews, coverage

Things I’ve written.

8 written by me10 interviews and coverage

01 · Featured

Members rarely leave in a single moment.

Paper · Sport Alliance

The MagicAI Engage white paper: my foreword and analysis, published by Sport Alliance.

Two studies on Magicline data. Members who got a well-timed message came back at higher rates than a randomly held-out group, and a year of attendance from 431,325 new members at 1,752 operators shows how early the drift starts.

with Daniel Gonzalez ·

02 · Archive

20266 pieces

  1. Coverage

    Sport Alliance veröffentlicht erste Ergebnisse von MagicAI Engage

    Trade news on the white paper’s findings, quoting me on why a well-timed message still helps when a member’s routine changes.

    bodymedia.deIn GermanThe paper in German (PDF)

  2. Feature

    Why most fitness brands don’t see a return on their AI investment

    Lauren Parker’s feature, built on my keynote at the ATN Innovation Summit in New York: why the technology is rarely the problem, and why the return usually takes 18 to 36 months.

    Athletech NewsLauren Parker

  3. Podcast

    Why most AI projects fail

    Why change management decides whether AI pays.

    The Future of FitnessEric Malzone

  4. Coverage

    Sport Alliance launcht MagicAI Chat

    Trade news on MagicAI Chat going live for Magicline customers, quoting me on behalf of Sport Alliance.

    bodymedia.deIn German

  5. Interview

    Strategien für messbare KI-Erfolge in der Fitnessbranche

    A long interview on measurable returns from AI in fitness: the J-curve, hard and soft ROI, and shadow AI versus governed AI.

    Magicline blogAnna BöckerIn German

  6. Op-ed

    Can we all just relax a little?

    A calmer take on the AI doom cycle: energy limits, labour data, layoffs and S-curves.

    MediumAlso on LinkedIn

20254 pieces

  1. Article

    Building a PhD-level AI workout planner with Google ADK

    A nine-agent pipeline that writes training programmes for about 70 cents each. I built it because I hated writing programmes.

    Medium

  2. Article

    Building the Tinder algorithm for gyms: a data science love story

    A playful build that matches gym members by training habits, and what a gym could do with it: workout partners, class times, PT assignments.

    MediumCode on GitHub

  3. Coverage

    FIBO Future Talks 2025: the event report

    The report on the Cologne programme, naming me on the automation panel as CTO of GymNation.

    EuropeActive

  4. Coverage

    GymNation research suggests the Middle East is ripe for low cost gym growth

    Coverage of the GymNation UAE and KSA Health and Fitness Report, a survey of 9,644 people that I designed.

    Health Club ManagementKath HudsonThe report itself (40 pages)

20247 pieces

  1. Article

    “AI will never replace the human connection.” Think again

    Why I think AI and staff each have a place in the member journey, with the research behind it.

    LinkedIn

  2. Coverage

    GymNation finds celebrating members’ progress with digital messages improves LTV by 10.5 per cent

    A six-month A/B test of 7,000 members: progress messages lifted lifetime value by 10.5%.

    Health Club ManagementKath Hudson

  3. Coverage

    GymNation’s AI agents and the numbers behind them

    75% automated tour-booking success against a 50% industry norm, and a 20% lift in digital lead-to-sale conversion.

    Health Club Managementpress release

  4. Coverage

    Case study: GymNation’s AI agents

    LlamaIndex on the agent build: retrieval, ReAct agents, function calling and multi-agent orchestration.

    LlamaIndex blog

  5. Article

    The development of AI agents: how we deployed an army in our business

    How agents work (tools, reflection, memory) and how we ran an orchestrator with worker agents across the GymNation member lifecycle.

    Medium

  6. Article

    What is AI and fine-tuning?

    A 2024 primer on learning types, retrieval and fine-tuning, written after the Beyond Activ AI webinar.

    LinkedIn

  7. Article

    Untapped potential in the fitness industry with AI

    Why fitness trails other sectors on AI, the four blockers (vision, skills, infrastructure, change management), and why most business AI is still classical machine learning.

    LinkedIn

03 · Elsewhere

Want to work with me?

It starts with a short call.