Suggest New Features 💡

I propose that You.com include, within the Pro plan, a paid add-on to activate Zero Data Retention (ZDR) and No Model Training (NMT) without requiring the $200/month Enterprise plan. Currently, You.com cannot be used by lawyers, academics, doctors, or corporate employees under the Pro plan because it lacks privacy guarantees. Many users don’t need the full Enterprise benefits (like ARI or an expanded context window); they simply want to keep their data out of training and storage. Offering ZDR and NMT as an add-on to Pro would: Increase Pro subscriptions. Allow privacy-sensitive professionals to use the platform without violating regulations or confidentiality.

Dear You.com team, I truly love using your platform, but the 64k token limit on the Pro plan is holding me back. Please, could you consider increasing it to 128k for Pro and 200k for Team? Many of us are working with long documents and really need more space. This upgrade would mean so much. Thank you sincerely!

GPT-5 is currently available in your model selection as "OpenAI's flagship model for complex tasks", but unlike other models, it doesn't specify or offer a High Effort reasoning mode option. I request that you add "GPT-5 (High Effort)" to provide access to GPT-5's maximum reasoning capability

Specifically on mobile, when I click one of the 'Jump back in' options, it automatically fills the input field with pre-written text. This is annoying because I then have to delete that text before I can type what I actually want to ask. On top of that, the 'You may also want to ask' suggestions feels unnecessary from a UI perspective, especially on mobile where space is already limited. I think removing the pre-filled text and the 'You may also want to ask' would significantly improve the overall experience for most users. I've attached pictures to show what I'm talking about. (I already posted this same thing on the Discord, but I have a feeling it's more likely to be seen here)

I have a Team account with you.com , and while I find it superior in many ways to Gemini Pro and Perplexity for research, it has a significant commercial flaw: ARI. Under the Team, which I had hoped to extend to other analysts in my team, IO get 5 ARI generations in a month. However... when these generations return clearly incorrect data, and I request a search to fix that information, a second ARI credit is used, even when the requirement is relatively minor. A case in point is a research study into ICT project costs, where the staff size of the companies examined was found to be ridiculous. For example, an organisation with tens of thousands of staff was listed as having just 5 staff. Requesting the staff sizes of the listed organisations to be re-researched incurs a full ARI credit. The result is that in practical terms, you.com is not a viable research service: at best it can start a search, but I end up needing to move flawed reports to other services. While it is possible to move reports into other streams of prompts, that provides no research productivity compared to keeping everything inside Gemini or using other AI workflows. I wanted to provide this feedback so that you can reconsider the way the ARI workflow is deployed. A simple fix would be to automatically switch from ARI to a general prompt once a report has been completed. A more effective solution would be to address the ridiculously low monthly research threshold. I'd be happy to pay a higher price, or a per-research call, but not with the current workflow.

The idea is that agents are able to use multiple llms. e.g. o3 pro for the tinking and concept power and after that step the ouput gets carried to Claude for the coding skills.

I think we need this model to test it full capabilities.

The prompts context window on you.com is very small, nowhere near the 200k claimed by team plan. When it reaches about 10k tokens, it will force a conversion to a file upload and then replaces the original prompts with "Please review the attached file: ...". This approach severely reduces the accuracy of the LLM response and is likely to output a summary of the original prompts rather than the response required by the prompts.

Just like Google and Microsoft did!

I feel far more users use apps that can parse and have pasted into them normal formatted text (not sure what it is on Android, guessing HTML) than markdown. I feel markdown for most people is too technical and niche. So specifically the behavior I would love to see would be for copy mode to change nothing except allowing the user to copy text normally, just as it is originally presented. This would also fix the problem of losing your place in the text when you toggle between text selection mode and normal mode. The "Universal Copy" app is a partial workaround - it allows for a normal rich text selection and copy, but only works for the text that is on the screen when invoked.