When I was in college, my coach knew everything. Not just my splits and my mileage — he knew I'd been up late studying, that I was stressed about finals, that I ran better in cold weather, and that I needed to be yelled at before a track workout but left alone before a race.

He'd built that knowledge over years. Daily conversations, watching me warm up, reading my body language, remembering what happened last time I looked tired on a Tuesday. That contextual understanding is what made him a great coach — not the training plan itself.

When Joe and I started building Vici, the question that kept us up at night wasn't "can AI write a training plan?" That part is relatively straightforward. The real question was: can AI learn a runner the way a human coach does?

The honest answer is: not fully. Not yet. But the gap is smaller than you'd think, and it's closing.

The Context Problem

Here's what makes coaching hard — and it's not the exercise science.

A human coach sits down with you before practice and asks how you're feeling. You say "fine" but they can tell from your posture and the bags under your eyes that you're exhausted. They adjust on the fly. They remember that last time you said "fine" and looked like this, you pulled your hamstring trying to hit tempo pace.

An AI coach gets data. Strava uploads, numbers, timestamps. It doesn't see your posture. It doesn't hear the hesitation in your voice. It can't tell the difference between "I'm tired because I trained hard yesterday" and "I'm tired because my baby was up four times last night" — unless you tell it.

That's the gap. Not intelligence. Context.

Bridging the Gap

So how do you give an AI coach context? You can't make it omniscient, but you can make it a good listener.

This is something we think about constantly at Vici. Every piece of context that flows into the coaching system exists because we asked ourselves: "What would a human coach want to know right now?"

  • How did you sleep? A human coach notices you yawning at practice. We ask directly — and we remember your answer when we brief you on today's workout.
  • How does your body feel? Sore? Energized? Heavy legs? A coach watches you jog the warmup and adjusts. We ask you to check in, and we factor it into every recommendation.
  • What happened on your run? A coach asks you after the workout. We give you a space to journal, and we actually read it — not just store it, but understand it and bring it back up when it's relevant.
  • What's your stress level? Life stress affects training. A human coach knows about the big presentation at work. We ask, and we use that information to temper our expectations for the week.

None of this is revolutionary in isolation. Asking someone how they slept is obvious. What's hard is making the system actually use those answers — connecting your Tuesday sleep report to your Wednesday workout recommendation, and then referencing both in your Thursday morning briefing.

The value of context isn't collecting it. It's connecting it.

What AI Actually Does Better

I don't want to pretend AI coaching is just trying to catch up to human coaching. In some ways, it's already ahead.

A human coach is one person with one brain and 24 hours in a day. They can't sit down every morning and cross-reference your sleep quality, yesterday's training load, your 30-day fitness trend, and your journal entries from the past two weeks — all before writing you a personalized briefing at 6 AM.

But that's exactly what Vici does. Every morning.

A human coach forgets things. They might not remember that three weeks ago you mentioned your left calf felt tight after hill repeats. The system doesn't forget. It holds weeks of your history in memory and surfaces patterns a human might miss.

A human coach has office hours. Vici is there at 5 AM when you're debating whether to run or go back to sleep. It's there at 10 PM when you're wondering if tomorrow's long run should really be 16 miles given how the week went.

This isn't about AI replacing human coaches. It's about doing the parts that AI does well — pattern recognition, consistency, availability — while being honest about the parts that are still hard.

What We're Still Working On

I'll be honest about what Vici can't do yet, because I think transparency matters more than marketing.

Vici can't watch you run. It doesn't know your cadence dropped in the last mile because your form broke down. It doesn't know that you overpronate when you're tired. It can't see the difference between a 7:00 mile that felt effortless and a 7:00 mile where you were grinding.

It doesn't know your personality — not really. It doesn't know if you respond better to tough love or encouragement. It doesn't know that you always sandbag the first repeat and then negative split the rest. Over time, the system learns these patterns from your data, but it's learning from the outside in — numbers first, nuance later.

And it can't have a hallway conversation. It can't bump into you at the gym and notice you're limping. It can't call your training partner and ask how you really looked on that run.

These things matter. We know they matter. And we're building toward closing these gaps — not by pretending they don't exist, but by acknowledging them and designing the system to ask for what it can't observe.

The Goal Isn't Perfection

Here's the thing: most runners don't have a coach at all.

The choice isn't "AI coach versus great human coach." For the vast majority of runners, the choice is "AI coach versus a PDF training plan downloaded from the internet." Or no plan at all.

Joe and I both raced for years after college with no real coaching. We figured it out ourselves, made mistakes, got hurt, came back, made more mistakes. If we'd had something — anything — that remembered our history, checked in on how we felt, adjusted our training when life got complicated, and actually explained the reasoning behind each workout, we would have been better runners and probably healthier ones.

That's what we're building. Not a replacement for the best human coaches in the world. A real coaching experience for the millions of runners who don't have one.