I installed it on an iPad Air (M2). On launch, I am given two options:
- Get started (primary CTA). Tapping this takes me to a blank screen.
- Sign in (secondary CTA). Tapping this and choosing 'Continue with email' gives me a login screen. I tried entering my email and choosing a password but it says "Couldn't find your account"
Isn't video generation expensive? Are you letting/making users generate series of videos without any plan? I think content depth and breadth would be good to check pre-generation at a minimum.
We are generating the videos using Remotion via React so its not as expensive as using a AI video model like Veo. There are three places where you can customize the course before you generate:
1) The initial prompt where you type what you want to learn
2) There are a couple universal follow up questions that ask the depth and breadth you want to learn at
3) There are then some course specific intent questions we ask that ask content specific questions. For example if you type "History of NYC" we might ask "What time period do you want to focus on?"
Thanks! We might end up doing longer videos, but we just found that there was a happy-medium for our attention span with the videos and making them bite sized kept us the most engaged. However, this decision was made before we made some significant improvements to our video generation workflow and so we definitely have been talking about whether longer videos could make more sense now. Our goal is to just deliver the content in whatever format helps people learn and stay engaged the best, so if that ends up being longer videos or some variety of short-form and long form than that works for us.
Are the courses meant only for personal consumption or can/will they be offered to other people? (for ex I have a Linux/DevOps training & assessment platform)
Would you mind giving more insight into how you make the videos (sourcing clips or generating them for Remotion videos, agentic?, and how you handle rendering issues and QA for the Manim videos) and costs?
Both workflows follow the same broad pattern. First, break the video into a set of scenes, then plan the scene, write the code, render it, and iteratively refine the result.
For Remotion, the AI determines what resources the scene will need, whether that be images, generated assets, or prebuilt visual components. This part doesn't happen in Manim because its more for math based visualizations.
Re costs, both workflows have a series of presets that make the implementation less expensive and more consistent. The reduction in price for the Terra model series has been particularly helpful for us.
Haha, no that makes sense. I think we've seen a few other people building toward what you are describing, but it definitely would be interesting to see if people end up generating courses in this direction. I think there was a company in a recent YC batch called Clicky (https://www.heyclicky.com/) that built something adjacent to what you are describing (https://x.com/FarzaTV/status/2041314633978659092).
I get blocked on the phone number screen on sign up.
This has two issues:
1) It seems to require a US phone number (555-555-5555)?
2) Ideally I don't want to share my phone number at all. It says it is for finding friends, but I think many people don't want to share their phone number this lightly.
The app also tries to get access to my contact list - not ideal for just trying something out :)
Thanks! We are currently working on enabling international sign ups. Regarding the contact list and sharing your phone number, we wanted to make the app so you could interact with your friends courses as well. Sharing your contacts is also skippable. If it seems like users aren't using any features that would make interacting with users helpful, then we'll definitely remove requesting phone numbers in onboarding.
Yea we generate the course icons with a quick image generation model so total accuracy in representations is not guaranteed. But the actual courses themselves use research to construct the outlines and videos. Feel free to email us (founders@trykeet.com) if you end up taking that course and have any other thoughts.
We use pretty comprehensive web searches to find examples and real worked problems to use in courses and base the curriculum and roadmaps off of publicly available syllabi, so from our experience the content has been accurate. We haven't seen anything inaccurate but that could certainly be possible, just like learning via the internet or from any other source.
Not currently, but we are going to add some features in the future to allow users to report courses that have content that needs to be reviewed for false hoods or any other reasons. In the meantime, if you do see this feel free to email us and we can address it.
Sorry about that. We wanted to implement more social features on the app to let users learn with their friends which we figured letting users find friends from their contacts/phone number. Understand that isn't for everyone and we might make some changes to make those social features opt-out.
This is a totally skippable part of the onboarding. Sorry if this is off putting. I had just really enjoyed this part of other apps like Substack where I can see when people I know join and follow them. Totally understand if this is something you want to skip or don't find it appealing.
HN is an unrepresentative crowd. Don't focus too much on what people here hate/want, unless it's your target market. Very successful businesses have famously not found approval here.
Not about your app but every and I mean EVERY app implements these social features, and I don't know anyone in my circles actually USING whatever social features - even if we talk about our experiences with common apps. Are you sure you're not just investing programming effort into nothing?
We've seen some of the people on the app using the social features, but yea I think on some apps it doesn't get used. I think if no one ends up using these features we'll just remove them.
Thanks! Right now, any course you generate is freely accessible to anyone on the app. Sharing the course requires other users to sign up for Keet. In the future, we could build in some features to allow users to extract the course outlines and videos, but we haven't seen that requested so we don't have any licensing in place for that.
I don't think there's a business here, sorry. If people are in the mindset to have AI teach them something, they will either just ask a chatbot like normal, or at most they will use something like Gemini's Guided Learning. They're not going to pay for this.
That's totally possible, but I think from personal experience we haven't really enjoyed those tools because they struggle to form a structured curriculum that has videos and assessments. We love using guided learning or a chatbot to teach one of f concepts, but for things that we enjoy passively learning about it's just not really that engaging. I think there are plenty of people who teach themselves things for fun and and want to have structured courses to learn from. I'd be curious if you still felt this way after trying it out.
i will pay for it. I dont like chatbot ux for learning. for example if a reading about transformers and i want to take a detour into mlp then i can potentially fork the thread chat about mlp there and come back or pollute the current context. i dont like either of those.
We've had the same name for the past few years, but hadn't seen this other app. Sorry about the confusion. If it ends up being something that we see confusing potential users we will take a closer look.
I installed it on an iPad Air (M2). On launch, I am given two options:
- Get started (primary CTA). Tapping this takes me to a blank screen.
- Sign in (secondary CTA). Tapping this and choosing 'Continue with email' gives me a login screen. I tried entering my email and choosing a password but it says "Couldn't find your account"
1) The initial prompt where you type what you want to learn
2) There are a couple universal follow up questions that ask the depth and breadth you want to learn at
3) There are then some course specific intent questions we ask that ask content specific questions. For example if you type "History of NYC" we might ask "What time period do you want to focus on?"
Good luck with the venture
Would you mind giving more insight into how you make the videos (sourcing clips or generating them for Remotion videos, agentic?, and how you handle rendering issues and QA for the Manim videos) and costs?
For Remotion, the AI determines what resources the scene will need, whether that be images, generated assets, or prebuilt visual components. This part doesn't happen in Manim because its more for math based visualizations.
Re costs, both workflows have a series of presets that make the implementation less expensive and more consistent. The reduction in price for the Terra model series has been particularly helpful for us.
Most of those subject matters - to me - are fun to chat with LLMs with to learn (Gemini hooked to web research works nicely).
But the expectations of easy-to-use software is rising. A solution to this problem would be extremely valuable.
Sorry, this comment feels a bit like https://www.youtube.com/shorts/61ZqBW30LLA but just rambling
I get blocked on the phone number screen on sign up.
This has two issues:
1) It seems to require a US phone number (555-555-5555)? 2) Ideally I don't want to share my phone number at all. It says it is for finding friends, but I think many people don't want to share their phone number this lightly.
The app also tries to get access to my contact list - not ideal for just trying something out :)
Would people not be able to share their courses via a url?
Anyway, I don't want to sound negative here, good luck!
I sense a potential problem here.