riding with technology: CoachGPT

Happy Saturday, friends! Hopefully you’ll forgive me for the off-day post, or perhaps just the diehards will read this, in which case — excellent, you are **exactly** my target audience today. Bc I have a favor to ask of you!

ooooh who’s the new roan in the herd?? 

Obvi it’s somewhat inescapable these days to make it longer than 5 seconds without somebody extolling the virtues (or perhaps impending doom) wrought upon society by the burgeoning field of AI in general, ChatGPT in particular. 

I won’t bore you with the details, but I’m actually a fairly heavy ChatGPT user in my professional world — mostly for writing code and all that, and also a bit of copy editing for reports etc. 

oh wait haha, nvm it’s just a Dooz of a Different Color!

For some time now, tho, I’ve had this idea of utilizing the generative and multiplying effect of GPTs to support the creation of some sort of real-time pocket coach app. Not like, an actual app that you could download from the app store, and certainly not something I’m trying to monetize or whatever. 

But basically, just another tool to add to my existing kit of helpful technologies to improve my rides. Recall, I’ve been evangelizing for literal years now on the benefits of riding with a metronome app (mine is set to trot tempo) and an interval timer (mine dings every 2min). 

chatgpt helpfully suggested three distinct ride plans to be interspersed in our weekly routine

So finally I got serious about the idea. 

I started by feeding ChatGPT background info about me and Doozy, specifically relating to our current goals of relaxation and transitions. It spat out some ride plans and helpfully offered to craft a weekly scheduling guide that could be printed and left in my tack locker. Thanks, Chad!

But I pushed it for more — I took what I liked from these different ride plans, and asked it to create a moment-by-moment coaching script that could be turned into a recorded ride guide. 

my prompt for generating coaching cues throughout the ride

It took a fair amount of iteration and refinement, especially relating to timing. In fact, idk about you all but imho gpts can’t be trusted to math for shit, so once we had the bones of a script put together with time stamps for each exercise, I moved things over into excel to fine tune and create symmetry. 

From there, we plopped the script into an AI text-to-voice app (also recommended by ChatGPT), created an audio file (in a British accent, you’re welcome), and uploaded the resulting Recorded Ride Guide session to Google Drive. 

If I set things up correctly, that recording is now available to anybody (like you!) with the link.*

(*There are no trackers / pixels / counters / cookies / **whatever** associated with that link either, it’s legit just a link to a file saved into my google drive that you can either access directly online, or download / save locally as you see fit.) 

for those of you absolutely bored to tears by all things AI, however, there’s also new tack!! this time with exactly the type of aggressively anatomic v-shaped crown piece that i know works on Doozy

So. Back to me asking you all for a favor — I want you to try it out, let this recording play along in one of your next flat schools, and then provide critical / constructive feedback on what worked, what didn’t. What you’d change or adjust in any future iterations.

it’s a used Montar Normandie, from ebay — isn’t it elegant?

First, some details on what, exactly the ride entails + my preliminary thoughts after riding to it once (last night). It’s a 25min session, starting off with about 2ish minutes of walk before embarking on a trot warm up exercise of riding sweeping 3 loop serpentines, transitioning to walk for crossing over center line / changing bend, then trotting off again. 

This warm up stage lasts about 7 minutes, and includes changing directions about halfway through so you can serpentine off both reins. My impression was that it’s a fair bit too long for this specific exercise, tho an appropriate time frame for general trot warm up. 

well. ok. the rolled leather snap-on browband is clearly not a fit, womp, it’s so pretty and understated tho!
($50neg. including shipping if you think it’ll fit your creature! fraidycat {dot} eventing at gmail!)

For Doozy, getting straight into serpentines and transitions was a bit premature, so the next time I ride it, I’ll probably ease into that exercise a little slower, since there’s plenty of time. More advanced horses might also intersperse other common warm up exercises into this time, and you could add a brief walk break to reduce overall ride intensity too. 

Next up comes a block of trot spirals. Again it’s fairly generously timed, so the intrepid rider probably has enough time to spiral in and out, then back in and out again before the change of direction to rinse / repeat.

everything else works nicely, tho, the straps all lay nicely against each other vs layering weirdly like the last bridle i tried. we have a new (used) browband arriving next week to try too.

From the spiral, you step into your first canter. Starting with canter on a 20m circle, then beginning to intermix shallow one-loop serpentines of counter canter in between 20m circles of canter-trot-canter circles. This whole block of work from warm up serpentines, to spirals to canter lasts about 16min.

After the first bit of canter, the recorded guide brings you back down to trot and then walk, for about a one minute break, then finishes with another roughly 8min working session — taking you first back up to trot for another (shorter) spiral cycle to lead into the second canter. 

That canter follows the same pattern as before — circles with trot transitions and shallow counter canter loops — before transitioning you back to the very first exercise: three loop trot serpentines with steps of walk over center line, before finishing with stretchy trot.

the full picture is coming together! finished the kit with Wintec webber leathers, and some black Royal Rider irons that i already had on hand

Some real talk: Our experience did not start out…. great. Doozy was super tense and distracted and rushing. And I actually paused the recording for a bit during the first warm up trot segment to kinda regroup and get my bearings (and shorten my stirrups). 

But then… Actually, as we progressed through the exercises after restarting the recording, it felt extremely helpful. My experience may be different from yours in that it’s hard for me not to get kinda baited into the tension, to kinda devolve into an almost survival-mode “just get through the paces” kinda attitude. 

Having this recording tho, especially with the consistent gentle coaching prompts, really really helped keep me centered, AND — most importantly — kept me proactively giving instructions to Doozy. Instead of just being like, “omg dear lord slow down!” I could actually stay focused on an exercise and just help Doozy find her balance within those steps. 

we’re clearly going for that muted palette bc obvi doozy has no trouble drawing attention to herself already!

And gooooo figure, the ride ended much more successfully than it had started. Like finishing with another stint of trot serpentines with trot-walk-trot exercises might be a bit remedial for some horses, but it was just the ticket for Doozy and she was able to execute them much more smoothly at the end vs the beginning. Which, ya know, is the whole point of schooling. 

So yea. I made a thing. I’m excited about it. I’d like to create more things like it — with variations in exercises, timing, intensity, etc., presumably with progressive complexity to continue skill development.

You know how it is, tho. Grand ideas crashing into mundane reality, this stuff takes time to do and is so far of untested / unclear utility. Will it be one of those tools that, over time, I stop using bc it’s not helpful? Or will it be like the interval timer or metronome (which, personally, I played simultaneously with the recording and may possibly integrate into future recordings pending interest) and turn into something that I use for virtually every ride? 

It’s to be determined, I suppose. In the meantime tho, I’m interested in hearing from other folks. Do you plan to try this recorded ride guide out? And if so, would you please consider leaving feedback?