I'm sorry. The requested resource could not be found. Please try again or contact support if the issue persists. #1076

Hi @PurpleGreenW,

I'm really sorry you've hit this — that error blocking every model essentially means the app is unusable for you right now, and that's not the experience we want anyone to have, especially as a paying user. Thank you for taking the time to flag it.

The "requested resource could not be found" message across all models usually points to a connectivity or routing problem between the desktop app and PiecesOS rather than a model-specific issue, which is something we can almost always resolve quickly with a bit of live troubleshooting. I'd genuinely like the chance to make this right for you. Two options, whichever you prefer:

1. Hop on a quick support call — I'd be glad to jump on a screenshare with you, diagnose what's happening on your machine, and get you back up and running. You can grab a time that works here: https://calendar.google.com/calendar/u/0/appointments/schedules/AcZssZ3bJHPPSJJHom2WwsPf5KcNlu-vRZsVa3DO1gJ7q4VIwExD1NUFBXlc7w8JGnEoSd2TLxP3eFbh
2. Refund — completely understandable if you'd rather not invest more time in this. If that's the route you'd like to take, just confirm here and I'll personally process the refund request on your behalf right away — no hoops to jump through.

In either case, could you please confirm the email address associated with your Pieces account? That way I can look up your account on our end, tie the support context together, and (if you choose the refund path) make sure it's processed against the correct subscription. Feel free to send it here, or if you'd rather keep it private, email it to support@pieces.app referencing this issue.

If you're also open to sharing a screenshot of the error and roughly when it started occurring (including your timezone), that would help us investigate on our side regardless of which path you choose.

Again, I'm sorry for the frustration — we appreciate you giving Pieces a shot, and I want to make sure you're taken care of either way.

Hi Benjamin,

Thanks for confirming the email. Quick update on what's been done on our end:

• Refund: submitted and processed on our side. Please allow up to a week for it to fully clear and reflect back on your payment method — that's the typical processing window once it's pushed through.
• Subscription: canceled, so you won't see any further charges.

That said — the refund and cancellation are done regardless, but I do want you to know that if you're open to it, we'd genuinely love the chance to fix your setup and make things right. The "resource not found" error you were hitting is something we can almost always resolve quickly on a short screenshare, and beyond that, I'd personally make sure you're taken care of — whether that's hands-on support, reinstating your subscription on us, or anything else that helps offset the frustration. If you'd like to take us up on that, the booking link is here: https://calendar.google.com/calendar/u/0/appointments/schedules/AcZssZ3bJHPPSJJHom2WwsPf5KcNlu-vRZsVa3DO1gJ7q4VIwExD1NUFBXlc7w8JGnEoSd2TLxP3eFbh

I also want to acknowledge the broader point you made — that the last few updates have made the experience worse rather than better. That's hard to hear but important for us to hear, and I'm passing your feedback directly to the team. Regressions in the user experience aren't something we take lightly, and you shouldn't have had to chase a refund to be heard on it.

If anything else comes up on the billing side, or if you ever decide to give Pieces another look down the road, feel free to reach out to me directly and I'll personally make sure you're taken care of.

Apologies again for the experience, and thanks for giving it a try.

Best,
Mark

Hi Benjamin,

This is really valuable feedback — thank you for taking the time to write it up in detail. Quick response on each of the three points:

Cross-device memory sharing. Good news here: we're actively in a sprint on exactly this, and the feature should land in roughly the next 4 weeks. The goal is exactly what you described — memories accessible across your devices without having to TeamViewer into your main machine just to reach them. I'll personally make sure you get a heads-up here once it ships.

Coding experience / IDE integration. You're touching on something important. We're actually moving away from dedicated IDE-specific plugins and consolidating around MCP as the primary surface for coding workflows, since it gives us one unified place to invest deeply rather than spreading thin across N plugins that all drift out of sync (which sounds like exactly what you ran into with the Rider plugin and PiecesOS being on different LLM versions). That said, if MCP felt "dumb" and generated unusable code in your setup, that's the most important signal here — I'd really like to learn more about what specifically failed:

• Which client were you using MCP through (Cursor, Claude Desktop, Cline, Continue, something else)?
• What kinds of prompts/tasks were producing unusable output?
• Were there specific tool calls that failed or returned bad results?
• For the "work with files directly" part — were you trying to have it read/edit files in your project, or surface relevant snippets, or something else?

If you're willing to share even a couple of concrete examples, that goes straight to the team driving the MCP/coding roadmap. Continuous codebase indexing and large-codebase support (game projects, etc.) is also actively being worked on, and your input would directly shape how we approach it.

Model version transparency. Fair point, and consistent with the feedback you left on discussion #1077. We're going to surface the actual underlying model name next to each tier in the UI so you're never guessing whether Fast is GPT-5.4 nano vs. something else. Granular per-model selection as an advanced setting is also on the roadmap for users who want that level of control.

The offer to hop on a call still stands if you'd like to walk through any of this live.

Best,
Mark