WebDev

the same one I’m using

Which account do you use?

<media>

Claudelens was much needed. Will help us improve as a team and be more mindful of how we use AI

but my weekly limits are still not finished either?

<media>

<media>

Achaa okayyy

This is EXTRA usage

You have not crossed your weekly limits

Why is mine 0

Damn

Ok sir , got it

That’s in last 2 days

<media>

@117111057871026 Turns metric is confusing. Please record total user msgs instead. Is the logic of turns ki agar claude ne 5 bar internally msg kiya after 1 user msg, then you set it as 6 turns? Whatever the case, please have total user-msg per session as the metric

100s of turns in one session is not good practice. Because entire session history gets sent with every prompt and costs unnecessary tokens

its a preset of rules that update the memory of the project like beyondchats and clinic os and maintain a single memory throughout the daily works

Plus, keep your sessions small. One small task in one session. Then switch session for new tasks

Also, you are exclusively using opus5 for everything. You should use sonnet for most things, and opus5 only for planning.

<media>

Isiliye tumhara usage khatam ho ja rha

Please dont do that moving forward

There is no scenario where we would need orchestrator for our current usage

@202611475136752 So one BIG issue I see with your claude usage is that you use /orchestrator frequently. That sucks up tokens like there’s no tomorrow, and it's wasteful.

Updated tasksheet with staging MRs for this task.

Added demo videos for following tasks in the sheet: - ai supressions - ⁠customised signin trigger - ⁠universal ticket creator - ⁠dashboard chatbot - ⁠rbac ui/ux revamp - ⁠changelog section - ⁠llm traces section

@30150821474450 @159738440466644 guys please make it a habit to add video demos to task updates too. EOD updates are very helpful, same updates can be attached to the task updates column

this is only live on dev. I think it had some blocker mentioned in sheet. that’s why staging MR got delayed. working on staging MR now.

sure, I’ll update them.

need all of these ss added to their respective task updates column

add the demo video to it too

I’ll look into that. please give few minutes.

LLM traces staging MRs (Updated in task sheet)

@159738440466644 staging MR for this?

Good SOP. Thanks for coordinating this, @111561876209776

@267155589353519 assigned a ticket to you, please do check: https://beyond-chats.netlify.app/ivfindia.com/support?review_id=1341

chalo theek hai

Han

ye to individually ek ek ko krna hoga

You will see the option to delete the organisation at the end of the list

<media>

EOD update All assigned della and ivf india ticket closed

*Progress today:* shipped LLM traces visibility (approx costs) on dev with a god only toggle. (revamped UI) *Upnext*: staging MR

Nope, the way this cost is calculated is all based on formula of how many tokens / million cost and how much did we use

so, is there a way to show the accurate costs then? just confirming!

When I worked on the pipeline I used the costs which was listed on their site, so those costs are approx

Those costs are approx though

<media>

@analyst start dev server

@30150821474450 FYI these were the tasks i mentioned in the daily connect

Saurabh is the one building this

@117111057871026

Ask it to refer to Claude.md files for better understanding. To run it just type opencode on terminal, and /model and select your choice. I mainly use deepseek and min max

We hit limits not down

So if someone works on dev server and needs Claude, can use opencode

Btw on the dev server Claude is down, so I have setup opencode there, its just like Claude and performs on a similar level just a bit slower

@102095734710467

I’m not sure why it’s showing 3 sessions for 1 chat.

Yes Updated in the sheet

is this done?

@30150821474450 assigned you some 5 ivfindia's tickets today. And some 2-3 della tickets as well

ok

@analyst Share everything that has gone live in last 2 weeks. And share everything from last 2 weeks that is not yet live but development is complete.

All users with by default false toggle (except god)

Dashboard chatbot cta

This task is scattered across 2 repos and 55 commits (with two features combined), so aligning all of them carefully at single MR took more time than expected. and also dded two new things on both staging and dev environments: - A new card for manage dashboard assistant page in god section. - ⁠Locked chatbot access so it is turned off for all users by default except God users.

Staging MRs for Dashboard Chatbot along with Universal Creator: - ⁠Backend: https://gitlab.com/dev_management_beyondchats/chatgpt3_backend/-/merge_requests/1615 - Dashboard: https://gitlab.com/dev_management_beyondchats/beyond-chats-dashboard/-/merge_requests/953

We should show them separately.

Task sheet updated with Staging MRs for these: - Team management page enhancements - ⁠Change-log section for admin dashboard

https://gitlab.com/dev_management_beyondchats/beyondchats-embed/-/merge_requests/357 staging mr

Token counts are not counted incorrectly. Reason for the inflated numbers - totalTokens: inputTokens + outputTokens + cacheReadTokens + cacheCreationTokens, Summing per-turn usage across turns is correct in principle (Anthropic reports usage per API call, not cumulatively) — but Claude Code resends the whole growing conversation context on every turn, and most of that resend comes back as cache_read_input_tokens (a cache hit), not fresh input_tokens. Because cache-read tokens get added into totalTokens with the same weight as real input/output tokens, a conversation with N turns and a large c…

Thanks Nithin for the feedback

@117111057871026

also the token usage seems too i probably worked on this task 3 hours max , token usage also seems quite high, im not sure if its crt

@102095734710467 , in claude lens the duration calculation wrong i think it is calculating from when the chat was started and till current

Good. What was the issue

@102095734710467

the fix is on dev

https://drive.google.com/file/d/1_ieuQNWR0OeVs9WdL7BssT27darVyti4/view?usp=drive_link

Please join @202611475136752 , @159738440466644 , @30150821474450

https://meet.google.com/hjf-twmv-ofj

yes

@102095734710467 r we connecting at 2?

This?

Im not aware of this task , where is it , could u share the task details

@30150821474450 can you fix this today?

<media>

This doesn’t work if chatbot is not on the msg page. 🙁

i will be unavailable

Available at 2 pm?

@111561876209776 , @159738440466644 , @30150821474450 , @202611475136752 I need to get on a short 15-min GMeet call w you all to discuss claude code plan we need and how to learn from each other's usage.

So it’s live?

That is the expectation

Yesssssss

i have no re collection of this task. Can you confirm if this is the expected behavior: https://drive.google.com/file/d/1e4XTceRee1v9SuiQoPWgabykWwPqO_s8/view

4) toggleBEChatbot with query param → Row 176 "toggleBEChatbot should take as input - user query, so websites can trigger it themselves" • Status: In Dev | ETA: 18 May | Owner: Pavan @111561876209776 Please make this live. 18 May se atka hua hai. And share how to use it.

Please take sick leave and rest so you recover better. Inform in meetings group for the same.

..

So it the filter not working kya? Data remains same for all days filters

<media>

<media>

@159738440466644 please create MRs for dashboard chatbot first thing today

everything on /analytics is real

@111561876209776 can you confirm what data in analytics of whatsapp bulk is real and what is demo right now?

Hi team, I’m down with a viral infection due to the sudden weather change. I’ll be working, but my pace may be a bit slower than usual while I recover. Thanks for understanding.

Haven't updated them yet , will do it today

and please fill task updates for: “Add a "copy context" button to each AI message” on sheet

@30150821474450 updates on this?

Progress Today Resolved all assigned Della bugs Added resolution in the support tickets

Changelog page (with god only edit option)

<media>

@analyst start dev server

Don’t paste all the links in the chat but paste the link in a file and share the file to the perplexity

Another way would be setting up open claw, then fire crawl, but it will take longer + credit heavy

Yes I follow these steps: 1) Go to perplexity computer (it’s basically open claw but on the cloud) or even deep research 2) Share a file with all the links 3) explain perplexity that you are working on creating a RAG for a company from these specific web pages and we need to scrape them and store them in a csv with the format Title, Description, Link. Title will be the page’s title/name, Description will be the content and Link will be the link to the page, we need to not scrape the Header and footer and Duplicating ctas, We need to also convert the tags to md so that the data is easy to und…

Ki agar new org k kiye chatbot train krna hai on website and pdf data, how do you do it, and how cna I do it

Do you have perplexity ?

Nooooooo that's what I want you to help m with!

Did you already scrape ?

Alright

And 5 PDFs

I want 1 org trained on those 5 websites specific webpages. I have a list of 500 webpage URLs

Yes. How do I create that CSV?

Make different orgs for the 5 different websites

Then go to mindmap section, and go to csv, in csv section Upload the csv, and there should be a tick box which bypasses the approval requirement

This format

Title, Description, Link

You need a csv for webpage data

I have 500 webpages from 5 different websites, and 5 PDFs

After setting up an org, how do I train the chatbot on website data?

Are you talking about setting up an organisation from the dashboard ?

@54125194641609 @111561876209776 please share doc / video of how I can train chatbot on 50 webpages on my own. By generating a csv of well scraped content

Who can then re-share with clients

Will help us update the entire CSM & Sales team about these things

Please share such updates in demos group also

https://dev.beyondchats.com/dashboard/ivfindia.com/whats-new

its in CI and will be deployed soon

This was so needed

@54125194641609 @30150821474450 @267155589353519 mind-map now supports hybrid search and will show you exact matches on the top itself. - Also added most used filters: source/document type such as PDF, source link (search for vecs of a webpage etc.). - ⁠There is a “Hygiene checks” button now which will reveal any duplicates, vecs that have never been fetched. This should help us with mind-map cleanup related tasks and bug fixes.

Also perplexity computer is also open claw but it’s on cloud and does not stress my local computer

i have referenced the onboarding doc alot of times now but still no team member seems to have any idea of it 🥲

there is SOP for this on onboarding doc but emon has found openclaw to be better so we have moved away from using the node scraper now

it was raised on 17th july, its under QA and will be made live today

It’s sad that there is no standard SOP for these things, and @111561876209776 you are clueless about it. Setting up new orgs is fundamental to our growth. Please do better.

I used open claw, but now I just use perplexity computer

Nope I do not

- The card got wiped from the god page during a prod-to-dev sync on 21 Jul by someone, use this link for now please: https://dev.beyondchats.com/dashboard/ivfindia.com/god/chatbot-docs - ⁠Prompt is written in code. - Haven't got feedback on it yet, so held off on the staging MR, happy to raise it once ur testing is done.

Please let me know by when can this go live.

@111561876209776 1+ month ho gya hai is micro-requirement ko. Please address it urgently yar

@analyst Weren't we supposed to ask for emails and WhatsApp numbers after answering five questions for Dr Malpani? When does this go live? Please check product ToDo sheet and inform me

@111561876209776 Update on this?

Sure

if any bug is redundant and already worked upon then say so as an internal reply on the ticket and point to the canonical one

assigned della resort tickets to you today. check those

@111561876209776 what bugs to work on next

dashboard assistant section seems to be gone from god options

@159738440466644 1. how do i train the dashboard chatbot? I dont see a god section 2. ⁠where do the prompts live for it? 3. ⁠are staging MRs created for it?

Okay, will add.

no. add a button for gods on the announcement page itself

It Needs some ui/ux improvement. Once done, I'll share a proper demo + links.

There's a separate page for God users. It was already live on dev.

Would like to know this before Thursday (30 July)

@111561876209776 , @267155589353519 , @117111057871026 I want to know ki since 22 July, how many users clicked on video call, how many clicked on chat What was the chat duration for video call users, what was the chat duration for text chatbot users. More such analytics. Please discuss amongst yourselves, and share this info. And please ensure we track every unique button click for every new feature in GA.

@54125194641609 do you use existing scraper of node? if using something else then please do share

no, today is the day to make things live unless there will be other realtime requirements

And please share script / docs for scraping content

@111561876209776 Dashboard chatbot for reporting issues - not live yet?

Good job

how to add new announcement?

EOD Update Bugs Progress - covered della assigned bugs , added how to fix them once verified by siddhant will add them to production

Progress today: changelog implementation finished. Live on dev. Upnext: ui/ux revamp, staging MRs

@analyst start dev server

@analyst start staging server

@111561876209776 updates on the priorities I shared?

Okay understood! can you please also confirm this: “wouldn’t prod MR webhook help”

added task details

anytime something goes live on dashboard, there is no organized way of notifying clients or keeping track of it. We just want a new section where we will be able to document whats new. It will be manual process but we should be able to add tags, title, content and authors. Check this: https://www.oration.ai/changelog

also can u please provide some context on changelog in admin dashboard task. like for what operations specifically it’s needed.

@111561876209776 I think out of these 5 few are already present. so only which are missing, those will be required for this specific task.

scopes required: pages_show_list, pages_manage_metadata, pages_read_engagement, pages_manage_engagement and pages_messaging

Oh I’m sorry I thought you’re asking about Instagram scopes. My bad. just give me a min I’ll verify.

will fb comment automation work out of the box?

i dont see anything related to fb/messenger in that

.

They’re only the required scopes. also please do refer this doc. I hope it helps.

do we need to request any additional permissions from meta for fb?

Not that urgent weird up the code for it , working on other fixes as per mom

how urgent are these APIs? I have other priorities on my plate right now

Yes on onboarding and when they want to update location generally in case of clinic shifting or so

oh okay. So this will only be used a few times per clinic during onboarding?

This will share shortly

For getting precise locations of the clinic where user enter can enter their clinic address or select manually via maps

and an approximation of how much it will cost us

@202611475136752 will do, can you share what these are needed for?

Bro can you enable maps js API , places api and geocoding API in google project and give me api key need for clinic os

No, nothing else. Only these 2 MRs.

- https://gitlab.com/Pankaj-Baranwal/chatgpt3_backend/-/merge_requests/1469 - ⁠⁠https://gitlab.com/Shoney13/beyond-chats-dashboard/-/merge_requests/737

@159738440466644 any other MR to get fb comment automation working?

Weren't we supposed to ask for emails and WhatsApp numbers after answering five questions for Dr Malpani? When does this go live?