sk-proj-kgC52qr0bQyRPKjOXN65hj5oUlYk1c8SC8FaRoEYdB3FGMF3W47r46VEJyCHu0pXC86Vao_OT3BlbkFJCXeoAflB--MhCIgU7m3tYbTr2fDjl1fmlJifo9MWFZuPDNWKMNbe9km8FY4RrrkURQYfdLCzAA
Do I have access to Github Codespaces?
Or do I need to make a new github and give you the info to add me to our plan?
I am trying to figure all that out today lol I have Claude access done now working on GitHub
lol no problem james, i was just reviewing our meeting notes
lol no problem james, i was just reviewing our meeting notes
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
Im going to run a pilot test with transcription with google and this thing called Assembly.ai -- waiting on some credentials -- thought you'd like to take a look
check this out -- i dont think i'll need to build out the other 2 pilots with whisper/pyannote and google text-to-speech
Also, do you have Milvus setup? I'm going to start working on a chatbot to talk about the transcriptions. I wanted to setup the longterm short term memory. LMK
Just a custom memory with local storage for short term memory that sends a few n-shot of the conversation and I will use the vector databse you chose for pulling longterm memory similar to the inputs in the caht box
We may be able to just send the entire conversation with gemini context window size, however, I wanted to add these two things because its fundamental for all the past chat bots I made
lets hop on a call i am not sure what your trying to do lol
put ur api key in a .env file
should be deleted for secutiy practices lol we are going through a whole data governance peace lol
you free for a call or we can wait till tomorrow its a little late for me 630 here
did u get the email for our vectorstore
i am skipping user the vector db right now, i wrote a script to count tokens of the transcription files and its a lot less than I thought, ill be able to pass the conversation in cheaply
yeah tomorrow send me a meeting invite, i also got a response from OpenAI about billing and my key not having enough money on it.
i need to add the card to open ai, can you use claude for now or no?
i was going to use openai for this transcript chat thing, add it when you get a chance because I can't finish it. I want to use OpenAI for this specific project because of the cost. PS C:\Users\Chrs Krecicki\Desktop\audio-transcription> py test.py โ Failed: Status: 429 {'error': {'message': 'You exceeded your current quota, please check your plan and billing details. For more information on this error, read the docs: https://platform.openai.com/docs/guides/error-codes/api-errors.', 'type': 'insufficientquota', 'param': None, 'code': 'insufficientquota'}}
let me know when you get the card on it so I can finish up this leg of the project
ok let me do that now give me a minute so u can test
boss i am going to go for a walk for about 15m around the building
is it hot as hell in vegas yet lol
got the payment confirmed try agasin when you get back
yeah man we need to come to the beach to talk shop
i cant go to the beach until next weekend ๐ช
haha working form the beahc is best i got a waterfront place her ei bring my laptop downstairs and work from the beach a lot beautiful water views
a dream bro, i can see angel park golf course from my kids windows in queensridge
it nice but id trade up for what you have in a heart beat
dream in 10 years is a place in laguna beach
i am not a cali guy at all to much traffic and expensive for me lol
NICE! The transcription chat bot works like a charm
oh whats really ocol put this in rag and be able to ask questions about all of the calls
i need to move this also to github codeespaces so u dont have to use everything on the laptop
All custom. No external things used except for the LLM API calls. I built this using Flask. I may need to integrate this MVP into this thing called Sanic python because thats what Dustin is using for the portal.
Let me push this to github so you can check it out
i made a script to count token size, ill add a feature to use Gemini or somethign else if the context window is too big but GPT4-micro I think its called has a 1M context window
I believe I invited u to our ChatGPT enterprise account as well. Sandi python I havenโt heard of I use streamlit for this product I have been developing
yeah flask is similar with the way routes and stuff work
I love openai for their API and I love Claude for code because of the artifacts panel
were using assembly.ai for the transcript
Can you setup an assemlyai account with a credit card and give me the API key so I can update it at Google Secrets? We are on the free trial right now under an account I made for development. I need it done before we can let it into the wild in the office. @Nicholas McFadden @James Scott
i htink you sent this to me in our dms idk if nick got it you coulda posted this in our ai channel i can do this
here is the api key to use, i dont think they have a native enterprise have to contact someone but this is my api key with the billing attached
anyway to increase our API limits? i am starting to clean db records and enrich data and were stuck at 10,000 requests per day -- and we need the requests for the transcription app I made
do we need that many request? thats a lot
don't worry about it, i setup ollama locally and and using llama3.1:8b
im cleaning huge datasets and enriching them
gotcha yea idk if the cost is worth filling the whole bakc end
nah its not i had to make some more custom logic for parsing and some things but the local llama3.1:8b is doing just fine -- only using openai for the transcriptions and pipeline for legal, things that need that huge 1 million context window
nah its not i had to make some more custom logic for parsing and some things but the local llama3.1:8b is doing just fine -- only using openai for the transcriptions and pipeline for legal, things that need that huge 1 million context window
When you get on I have your code want to talk about how the visualization is done
all the things in the GUI are coming from a table I have in a local docker postgres database
hey good morning, yea i got access to the github this morning, wanted to know the .py file that has like how to set up the visulizations you had yesterday, woud i need docker? cuz i have like everything else all done just wanted to integrate it, or whats the name of the app and i can look
ohh have you ever run a sanic project?
oh man lol -- download my branch first
you'll want to get a local docker container setup running postgres
let me export the database and see if you can load it
to run sanic you do -- pip run main.py
you'll need to update the secondary database code that I had added to main.py
so its main.py? i am going to use that and just refactor everything to my use case but tahts all u do create local docker and container, and then just hit main.py and it creates that webpage?
no not at all that will just load the sanic application
when you go to the tort tab once oyu db is linked it'll dispaly the data
that page is a html page using jinja2 for tempalting
flask is constrained by global objects and the inability to handle async so it cannot do concurrent requests with returns at the same time
yes let me see waht i can cook up with this lol
i am here if you need help james -- may even fly to DR if they need me too ..
ahahah!! come be roomies and yes i like this solutioj as front end, those python apps i was using just wasnt cutitng it
yeah sanic seems great, dustin introduced me to it, i used flask for years but i like how this is natively async
and i can create this all within vscode? and or docker
i use docker for running the local postgres db
i just learned all this the past few weeks
im here at the office till 3pm so just reach out ill help
Iโm bout to crush this in docker in. Second
so i made updates in my main repo can we go over how to update it in the main one
Only 2 files so it should be easy to update
can you send them here? im assuming it was just the views files and the jinja one
i can swap them out and make sure it doesnt break anything
still havent heard from anyone on the Db migration stuff
yes let me send htem here
you've been doing extraction on medical documents etc yeah?
check out their medication example https://github.com/google/langextract/blob/main/docs/examples/medication_examples.md
yes i have i did that for years!
healthcare pharma and stocks is my background
i just got to the office after WFH wed/thur/friday
man wfh is the life lol ima try and get cam to get u that but i been remote for like 9 years
boss - yeah i worked for myself my entire life - -so this office thing is super demotivating
bra i feel u on that lol i just like the free time to work at my pace i cant stand looking or being confined to an area i be moving and working about
OK. I am back. Ryan is here, just met him in person.
yeah i like these guys ryan, joe -- havent met cam yet
i've got a database thing to test and then I will get your files integratec
If u want to impress Ryan talk about boats lol
Little word of advice ahah let em know when u update the dashboard so we can walk through it
have to do some DB update stuff because they changed the schema
OK working on it. I have to red some things you changed with auth. Should be OK. Will update.
oh damn youre using env for the rag_engine lol
got to give me a day or so -- lots of stuff I have to redo for env etc
hey chris for the project ryan wants me to have oversight, can you see me on the meetings and gameplans and stuff
awesome thanks man let me know i have multi agent tools set up already you can just drag and drop with strandz its set up for aws bedrock but can prob easily be changed to antrhopic or our claude api
did you ever add me onto the permissions on your project id?
can you give Dustin Surwill admin access to?
we're going to skip optimizing bigquery on an application layer and just re-write the backend for postgres
ryan prob knows better since he did it
seems to be outside of the code issue
lol bro you have worked your ass of on this
lol not all of them are used lots of steps to pull the data in from all the sources
haha i see the 4 being used in your code
getting this taken care of .. sick of waiting on other people
going to make cameron and the boys excited it'll be way faster with bigquery application layer taken off
Yes I hate big query re factor it if u can
i will 100% bro will be piece of cake
im going to get out of here in a half hour or so to head to CA
Itโs 5 here Iโm in the beach chilling
yeah man when you get into town we can take it for a rip
Hmm let me show u something donโt tell anyone lol