bless you!

I see a true investigator. Keep up the good work.

This is a great example of how to create dynamic screens using AI. The example you just shared has a lot of potential and can be applied to many other things. Thanks for sharing!

@planetnocode - I see how to integrate via API, and upload data (text) via your demo. AS a no-coder, I"m unclear on how to connect Google Drive/etc. (as it looks like React is needed for integrations. Do you have a tutorial (on planetnocode) for this?

Would love to beta access - looks great!

Nice overview. What’s your go to ACL look like for module/feature level access?

Wow ))))))) Awesome !!!

thankssss

Well done Matt

Great video. Has there been any updates on this matter?

This video is not all that helpful. You should have shown how to save multiple repeating group input tags or dropdowns.

Great video! Such a helpful way to solve issues👍

hey can you please do with gemini

What about accuracy?

Mini series!!

working thanks

Im working on a project using Assembly and its great so far. Ive decided against storing the transcript (well the utterances section) abd instead call the Assembly API to pull it down whenever needed. Do you have any advice if i wanted to store the transcript in my bubble app instead? The main purpose would be to allow users to amend/correct the transcript if its incorrect...but storing and manipulating JSON is a challenge..? Great vids by the way, keep up the content!

Hi there! I’m currently building an ai photo app with bubble, and I want to limit the numbers of photos users can generate a month for each plan. Free plan=15 ai generated photo, basic plan= 50, and pro= 100. Can you please make a video on how to do that?🙏

thank you!

Good morning, how can I adapt this name search input name by a photo search please thank you for your help

Awesome - i have the same question, how can you convert the base64 into an mp3 and save to the database?

What about Realtime API from OpenAI Matt?

This is actually very interesting. Do you know if there is a free and opensource, self hosted alternative for tamplated io ?

I would love a training on this

Thanks for sharing❤❤

Hi! How do we format the "closing_date" into a date instead of a string? So you can set a reminder of when that date is approaching?

Does Gemini allow multimodal RAG?

Thanks for taking your time providing excellent content..

I'm missing something because I did exactly the same thing and it didn't work

Thank you

did you delete part 2 and 3 or im just wrong because i think i saw it before

This is a long shot, but would you ever consider colaborating on a project? I'm trying to build an MVP for a start up I'm working on and could really use a partner/someone who knows what they're doing.. I know you're a busy guy, just thought I'd ask. Thanks again for the video!

you are an absolute legend. thanks man!

This is not optimized. It brings all users back from DB and then filter on the front end.

I keep seeing folks using tools just to enforce json output format, even though it is a workaround really. I don't condemn it, but it is still important to understand their real purpose. Tool calling (and this applies to all llm providers that have it as part of their API) was actually implemented by LLM providers in order to standardize how we instruct the LLM that it can/must call some tools usually for scenarios where the LLM is instructed to function as a multi-step agent. A tool should be really a functionality that your app can provide as a response to a request made by the LLM as part of its output. The LLM decides what tools it wants to call in a particular situation, (like a web-search or a math module). The way an LLM can instruct the app that it wants to call a tool (and with what parameters) must be somehow included into the LLM's prompt (with details about what each tool can do, what its parameters are, etc.). Instructing an LLM about its toolbox (mainly to build agents) was originally done through classic prompting in natural language, but this was tedious and prone to errors. So what OpenAI did (and others followed) was to finetune their models on examples where instructions about the toolbox were given in a consistent format so GPT was now great at following that format. Next, in order to ensure that instructions about tools are included in the prompt (by us mere mortals) in the same format that it was finetuned on, they made this available through the API. Theres some additional internal magic happening for trully enforcing json output specifically for the tool calling sections of the output, which relies on constrained sampling. OpenAI more recently allows for enforcing JSON output following a given schema for the entire output (not just tool calling), and that should be the way to go when you don't relly want to call a tool, but they are the only ones to provide this functionality for now.

Hey bro! can you do a quick video for Bubble, how to have a search input on one page, and display relevant results on another? so when someone types something in, it navigates to a search page displaying relevant results based on what they searched, would very very much appreciate it. thank you!!

Thanks so much for this tutorial! Have been trying to solve this for hours!

Would be great to see the api connection of this in bubble

Sorry, your sound and accent is like the geoguesser skit man

This is a great release from eleven labs. I will be using this for sure.

Just what I was after

Great,, unique, niche topic! I'm working with the api in bubble right now... I'm having trouble passing the raw image data to the api connector and blueskies uploadBlob endpoint properly to post with an image. If you have thoughts, or want to figure it out... hit me up

Nice as always

Thank you so much, man! I'm starting a project that uses Twitter/Bluesky's interface as it's main core. I plan to use WeWeb to export the code in JS, but definitely will learn from the process of this video. Do you plan on adding videos using WeWeb? I mostly use Bubble, but for some projects, I simply need the code.