MTBennettMI

Thanks Anton, means a lot coming from someone who's been at this as long as you have. You named the hard part exactly, turning "feels harsh" into a change that's actually safe and right for that specific bike. Glad there's more than one of us trying to solve this issue.

@Untgrad: Haha, thanks for that compliment, I try to be straightforward and clear with my responses. You're not wrong. This project has been my sole focus for months. I've sat in classes, at home, and everywhere in between on my laptop implementing things, testing things, and breaking things. I've learned a ton. I honestly forgot about the Trust Message. I had to google it for a reminder of what it was. It's freaking awesome that you're still out there ripping one of those. Ride it until it's dead!

@boozed: You're right about the failure mode. If Suspend were just a chat box sending your message to a raw model, it'd hand you confident garbage and have no clue it did it. I didn't build that. The model isn't the part that "knows" suspension, and it doesn't invent setups from scratch. The app already has your actual fork and shock, your current clicks and PSI, your weight, your riding style, and every change you've made before. The starting numbers come straight off the manufacturer setup charts (Fox and RockShox publish PSI and click ranges). From there all the model can do is propose a specific setting moved to a specific value inside that dial's real range, like rebound from 8 clicks to 7. Then it hands that to you to approve. Nothing moves on your bike on its own. So it's less "AI invents a tune" and more: you tell it the fork felt harsh on square edges and blew through travel on the big hits, and it tells you which dials to touch and by how much, inside real ranges, off your bike's actual numbers. Whether it actually worked is whatever you feel on the next lap. That was never the model's call. Appreciate the real criticism though. This is the kind of feedback I'm after.

@VorsprungSuspension: This is exactly correct. There's only so many different changes on a bike, and these models are smart enough to realize what adjustments actually do to a ride. Someone else in this comment section phrased it as a "bridge between feelings and the dials", which puts it absolutely perfectly. Also, can't believe this post blew up so much to where proven companies like yourself are noticing! Love the stuff Vorsprung is doing!

Thank you for this response! There seems to be some confusion in the comments, thinking this is attempting to replace all suspension tuning, which it's not. It's here to do exactly what you said: bridge the gap between feedback and dials. I'm glad you were able to understand this, and would love any feedback if you give the app a try!

Respectfully, you've responded to countless comments indirectly calling the app and it's functionality junk. Have you given the app a try yet? It's good to be skeptical about AI, especially in this day and age where AI slop is being pushed wherever you look, but I actually think this app is beneficial for many riders. Not everybody knows what to adjust on their suspension, and many would rather just ride the bike than learn what to adjust. Suspend lets people just ride their bike, give any simple, plain English feedback they've got, and get a better tune from there. By no means am I saying that this is meant to replace professional suspension experts, telemetry data, or riders who have developed a deep understanding of suspension. It's just a means to get a better tune closer to normal riders. It makes no sense for the average rider to spend thousands on forks and shocks to leave them at some baseline setting that feels "alright", and Suspend is here to combat the issue. I really do appreciate you sharing your honest thoughts on the concept, and would encourage you to actually download and try the app. If you already did and this was your honest feedback on it, I respect it. It's not going to be for everyone.

That's really cool to hear, and honestly, it doesn't seem Suspend is the right fit for you. It's great that you're interested in doing that, but most riders, including myself just wanna be out riding the trails and not spending time watching tutorials, which is what Suspend is for. It lets normal / casual riders get more out of their suspension for almost no extra effort or learning. If you end up trying it out, I'd love any feedback, positive or negative.

@Dougal-SC: You're not totally wrong, but many people literally don't know what the dials on their fork does. They might be able to notice that it dives through travel on drops, but when presented with so many dials: LSC, HSC, LSR, HSR, air pressure, tokens, it might be difficult for some people to know what might fix that issue. That's the goal of Suspend, to get better tunes for the casual riders. It's not replacing pro-race tunes or telemetry.

@mcleodpulaski: Not exactly at the moment. I've got the app collecting de-identified chat sessions from users and putting them in structured JSON files, which I plan to do real LLM training with in the future, but until there's enough entries there, it's just based off your own data and prior experiences. If anybody is at all concerned about this "data collection", it's de-identified from your account (user tied information is removed from these), is never publicly released, and you can opt out at any time by emailing [email protected]. This isn't just to collect user data, if it was I wouldn't be pointing any of this out, it's to actually improve and get better results from future iterations.