DrMalpani webmaster

OK< have stopped the YT ads

Youtube promo band kr dena chahiye. Ghatiya traffic hai.

*Dr Malpani Weekly Performance Report: Jun 30 - Jul 6* =================== *OVERVIEW* *Paid Traffic:* 9325 users (↑ from 222) | 4s Avg Engagement (↓ from 11s) | Top Cities: Patna (879), (not set) (697), Ahmedabad (523), Agra (490) *Organic Traffic:* 1936 users (↓ from 2198) | 89s Avg Engagement (↓ from 91s) | Top Cities: (not set) (105), Bengaluru (67), Mumbai (64), London (58) *Insight:* The 42x paid jump (222 to 9325) is the YouTube promotions turned on this week, not a search-campaign change. This is top-of-funnel awareness traffic by nature: 4s average engagement and a tier-2/3 city mi…

No. Let’s get on a call post 7 pm and we can discuss

So should we stop running ads ?

Most of it is not bot traffic. It’s ad traffic because we are running ads. You can add filters to assess only organic users. I’ll get on a call to show you how

PLease call me. We need to fix this problem once and for all

<media>

Hard to analyse the traffic

Too many bots :(

<media>

Working on it. I’ll share the costs by tomorrow 3 pm.

We need to start doing an A/B split test on the drmalpani.com website where we offer the patients the regular chatbot on every alternate day and the AI twin on every alternate day. I still need an estimate about the costs for the AI twin please.

One more compliment - and suggestion !

<media>

Clever !

MiceLabs had requested this feature “total time to read” to be shown on articles on their website. I have extended this feature to all wordpress websites under mgmt - ivfindia, startupmentors, donoreggs, etc.

<media>

Thanks!

Tasks - - Building a custom VAD filter ( Issue it addresses - It amplifies the user's voice and suppresses the background noise and filter them out ) . It is something i have seen being used for better noise cancellations and talking in voice chatbots so would be an exp try to see how it works - Fixing out audio related bugs in the AI Twin ( which are still there as noticable in the transcripts of the chats in the admin dashboard )

@analyst summarise the conversation between Dr Malpani and Saurabh since 19:45 today in this group

So feel free to use it as well and give any feedbacks

For the time being , you can also test the chatbot withoug acutally going on the video call to save tokens and cost directly on the admin dashboard for answers quality and relevance. I have just added it here for now so that admins can play around with the actual RAG engine behind it . It shows all the metrics as well as the retrieved chunks as well.

Yes, you are right for the issues part . and we can use it across the other chatbots also. But since our chatbot is already so good and has been developed for so long ( Like actual dev time ) so ditching it for any newer wouldn't make sense until we have some real need for it

My point was that the issues would be similar for any chatbot

Pankaj sir could clarify it better . But for the AI twin current changes are live based on the feedbacks u had and issues flagged 🙃

Can we use the pipeline and RAG model you used for the AI twin for the existing chatbot?

Okay

There are also many other technical aspects in how the whole pipeline works so pankaj sir asked me design it as a seperately for the AI Twin as opposed to just using the existing BeyondChats pipeline and wiring it on the AI Twin.

Use case - - Latency budgets for voice chatbot : We need to have it tight so every component is desgined to reduce this . The STT that takes your speech and transcribes it , the LLM optimized for fast Time to first token , The TTS engine that powers the agent vvoice and the Heygen Avatar which also has its own latency. The reason we did not use the BeyondChats pipeline for this one: Latency was the main reason . The whole point of desinging the new RAG engine for the AI Twin was to have it match with the constraints we have. SO we do not have a fast pipeline with Super slow RAG Engine

Why not ? Please educate me because I thought the whole point of developing stuff is that it should be modular and we can reuse and repurpose it

Nope, for voice rag like ours , we need to have seperate optimizations, so we can't just duplicate everything we have for the chatbot in the AI twin one to one

Aren't we just duplicating work unnecessarily?

The pipeline we have for the AI twin is completely seperate.

Aren't these already standard present for the existing chatbot on Dr. Malpani?

It is for the AI Twin since I am handling AI twin. Will be sharing updates for daily changes and work here as well for the AI twin based on the feedbacks and all so we address everything as the dev works go on

How do we apply all these changes to the AI twin as well?

Great !

EOD Updates - - The transcript window now opens by deafult for users on desktop view and auto scrolls as the chat keeps happening - Added the drop down of Auto detect in the language selection droopdown at the lobby - Improved the Prompt & answer quality : Fixed length contradiction ("be complete" → "answer only what's asked"); dropped the "up to six sentences" allowance, Banned the reflexive upsell ("Would you like…?") in the prompt; removed next-step-offer language from the "hesitation" branch, Added prompt-secrecy rule (never reveal/paraphrase instructions) + meta-question handling (ident…

Thanks!

<media>

So how do I get rid of it ?

Yes

Most probably it is the chat which is already selected by you before applying filter

Why is a 4-day old chat being displayed on the top of the dashboard

<media>

They will all need to have to talk to each other and the staff when we set up a clinic OS

Yes, we can add an option at the end for users to connect on whatsapp / enter their whatsapp number But right now, AI-twin is a very separate project. It’s not linked to the beyondchats ecosystem where insta, whatsapp, website - all chats persist. I’ll note it in tasklist to explore this

Iterating on these things so the agent doesn't keep on talking endlessly and confuse the user. It is in the tasklist to work upon already.

Can they switch to WA from this call ? Communication should be seamless

<media>

Need to fix this please

<media>

Yes please

@analyst Collate and share all feedbacks about ai-twin shared by Dr. Malpani in this group since last Thursday.

I tried the digital twin . Sending my feedback across below , also filled it in the feedback bar after the call . Interface was great , but both times I wasn’t able to hear doctor My audio settings seem fine . Not sure if there is any specific setting i should change to . Haven’t tried with headphones though I don’t have access to them currently . It clearly understood the questions i was asking and transcribed correctly . Haven’t reached the medical questions yet so unable to give feedback on that . Personally i prefer chat vs phone call , so would also have liked to have a chat option i.e…

She logged in from the US

Ria Mirchandani

<media>

Yash Panditrao

That way we can assess their specific chats to get more context

Who sent it? Would be great if you can share name of person who sends feedback

Wow, pretty detailed. Very helpful.

Great feedback!

Hi Dr. Malpani, I tested the Twin again. It works much better than last time. The questions are not breaking, and the language selection is a nice touch. The download transcript feature is also useful, but I think a better version would be to generate minutes of the conversation instead of a raw transcript. Most people will not read a full transcript. Right now, the transcript downloads as a .txt file, which opens in Notepad and is hard to read. It may be better to assemble the transcript or minutes in a clean PDF format, so people actually feel like reading it. The current .txt solution fe…

This link ain't opening, tried multiple times

<media>

Currently the feature is not there , so for time being these pseudonyms are randomly generated so we can track same users and have a better view whole looking at the dashboard. Once we start taking users name or number or some sort of PII , I'll swap it with that

What are all these pseudonyms ? I thought we were going to nudge patients into sharing their WhatsApp number and email address if they asked more than three questions

<media>

Hi {name}, I'm Dr. Aniruddha Malpani. Over the last 30 years, I have helped thousands of couples with their infertility journey. My patients often have a lot of questions, and I would love to answer each one personally. This is why I created my digital AI twin. I have trained it on my 40 years of clinical experience, and I’ve ensured that it gives you reliable answers personalised to your situation. No hallucinations ! Now Patients can talk to my digital twin any time they need! I request you to please try it out: https://ai-twin.drmalpani.com/ And share your valuable feedback. Your hone…

@179035711746252 Can you proof-read this msg we are planning to blast on email and whatsapp?

Hi {name}, I'm Dr. Aniruddha Malpani. Over the last 30 years, I have helped thousands of couples with their infertility journey. My patients often have a lot of quesions, and I haven’t always been as available as I’d like to answer every one personally. So I’m experimenting with creating a digital AI twin for myself. I have trained it on more than 30 years of my clinical experience, and I’ve ensured that it gives you reliable answers personalised to your situation. Now Patients can talk to my digital twin any time they need! I request you to please try it out: https://ai-twin.drmalpani.co…

We are now actively promoting the AI twin and I keep on checking the dashboard but there seem to be very few conversations and I was wondering why this is happening

15:10

When did you realise this?

Oh no!

I’ve increased subscription tier now and added unlimited usage credits so this doesn’t happen again

We had exceeded our ElevenLabs subscription usage limits :/ So all its services got paused - ai-twin, audio-chatbot

<media>

This is what we send to patients

Clinic Mobile: 9867441589

Mumbai 400 005. India

Malpani Infertility Clinic, Jamuna Sagar, SBS Road, Colaba

Dr Aniruddha Malpani, MD

Regards,

We look forward to helping you to have a baby !

Taking treatment at a world-class clinic will maximise your chances of success and give you peace of mind you did your best !

If there is a problem, then we can treat it !

infertility.

With these test results, we can determine what medical problems are causing your

Please send me ALL the detailed test results and medical reports . You can scan them in as a single doc file and email them to me at aniruddha.malpani@gmail.com with a copy to frontdesk@drmalpani.com

The vaginal ultrasound scan on Day 2-3 and again on Day 10-11 should check for the following. a. ovarian volume b. antral follicle count c. uterus morphology d. endometrial thickness and texture

HSG ( hysterosalpingogram, X-ray of the uterus and tubes, http://www.drmalpani.com/knowledge-center/articles/hysterosalpingogram) on Day 8 of your cycle ( to confirm

blood tests for you for the following reproductive hormones - FSH ( follicle-stimulating hormone),LH ( luteinising hormone),PRL ( prolactin) , AMH ( antiMullerian hormone) and TSH ( thyroid stimulating hormone) on Day 3 of your cycle, ( to check the quality of your eggs). Do this from a reliable lab such as SRL . Day 1 = Day the period starts.

Read more at http://www.drmalpani.com/knowledge-center/resources/book/chapter4b

semen analysis for your husband ( to check his sperm count and motility).

You need to do ALL the following simple medical tests:

Do a complete workup before starting treatment haphazardly.

Yes

https://www.drmalpani.com/knowledge-center/articles/can-i-move-my-frozen-embryos-safely

https://www.drmalpani.com/knowledge-center/articles/why-is-my-pregnancy-taking-so-long @179035711746252 Do you have a 'Fast-Track' Infertility Diagnostic Workup service as mentioned in this article?

Thanks !

Some of the suggestions are plain wrong / hallucinated. Ex: Sitemap being 2018 deprectated files

I’ll implement the suggestions that are more relevant from the security loopholes side

Much of it is not relevant for crawlers like Google & bing. But the suggestions are still good-to-have things we can incorporate to make our website follow all schema best practices. But no real benefit will be visible from these changes to end-user.

Please show full webpage screenshot

Please hard refresh, sign out (top right) and sign in

You don’t seem to be logged in

There is nothing to fix

<media>

Please fix this

<media>

What about this ?

The 2 confirmed problems are correctly identified. Seems to be really old pages that never got identified as problematic because they can’t be reached from anywhere on the website. But recent sitemap revamp must have made them visible again. I’ll fix this before eod

Please share feedback about these suggestions and how we are going to implement them

# Broken-Link & Crawl Report — drmalpani.com *July 6, 2026 · SearchFit broken-link check* **Method note:** I can't run a full headless-browser crawl from here, and your site renders most links in JavaScript, so this isn't an exhaustive every-link sweep. I pulled your real URLs from the sitemaps and directly fetched a representative sample of pages plus the links embedded in the content I could read. Everything below is a confirmed observation, not a guess. For a complete status-code map of every link, run Screaming Frog in JS-rendering mode or connect Ahrefs — I've flagged where that's worth…

Amazing job, @117111057871026 ! ❤️

She is a nurse in the US

Feedback from a patient

Dr. Malpani, I am still trying to think of more questions but I have tried it for a little bit. And so far, AMAZING!!! Very accurate of what I would think how youd answer! It feels so real, i feel like im really talking to you.

Trying to fix the issues as they arise

But it uses a lot of outdated codes. Ex: There is a piece of code that finds common keywords on the article, then searches the website for articles matching that keyword, then embeds a link to best matching article on that keyword. But it’s poorly implemented and is leading to unnecessary latency in page load. But when I removed some of these codes, the chapter page broke - header has moved to below the article

https://www.drmalpani.com/resource/books/chapter1a I’ve also started updating the outdated book chapters in clinic website.

I’m hoping that we’ll have 1 version ready by tomorrow eod.

Update 3: I have started work on analyst@ivfindia being accessible to everyone. It’s significant work - there is no rate-limit, no authentication, no way for us to view sent mails, received mails, no concept of “threads” if user replies to our mail, etc.

I’ll utilise these functionalities to generate better articles for the websites under management

Update 2: Now seoforge tool is able to: -> Auto-Generate a new article if just title is provided -> Create a website brand personality & TA automatically for better article generation context -> Auto-Generate topics for writing articles on -> Allow user to add a raw thoughts dump that can be used to frame a new, polished article -> Update an outdated article into an up-to-date one using top-ranked Google articles on the topic.

https://docs.google.com/spreadsheets/d/1bXL8lZ-k39esUXZNvqLWl7ZjDgeK1wfWXV-VzRPkCRU/edit?gid=519389466#gid=519389466 Here is the list of all articles that have been removed

Update 1: 56 irrelevant articles have been deleted from clinic website. As per my review, no non-IVF articles exist on website now

I’ll go over it later tonight to review + share a timeline

Please reply

Changed the greeting to this. ::)

Please implement this advice after reviewing it. Thanks !

# SEO Audit — drmalpani.com (Malpani Infertility Clinic) *Prepared July 6, 2026 · Full-site audit · Run without connected SEO tools (Ahrefs/Similarweb not yet authorized), so volume and ranking figures are directional estimates from web research. Connect an SEO tool to auto-populate exact numbers.* --- ## Executive Summary drmalpani.com sits on a genuinely rare asset: 34 years of doctor-authored content and one of the deepest infertility knowledge libraries in India, updated almost daily (the articles sitemap was refreshed today). That depth and the clinic's E-E-A-T — real named doctors, I…

Okay sure

Hi, I'm Dr. Malpani's AI Twin. How can I help you?

Please do this also

Thanks

Sometimes it does say explicit stuff , working on refining those .

I like the features you have added, thanks!

Also I am not sure where the sexual intimacy answer came from

The introductory message is too long and too boring so I think we should shorten it

<media>

<media>

<media>

😀

<media>

<media>

<media>

Feedback feature wireframe ::)

Feedback on Medixcel report generation: Overall it looks good. 1) A few treatment points were missing. Pls run a thorough screening using the AI tool so that everything is captured. 2) Datewise chronology shud be maintained. 3) Doctor's name shud be mentioned in all the procedures

@102095734710467 I have resumed the ai twin chatbot with the correct avatar and tested it end to end . The chatbot now loads correctly and the sessions are initiated successfully E2E. ( The previous top up unlocked on avatar slot which was given to the previous avatar that we had , due to which the calls were still not getting connected ).

@179035711746252 Credentials for you to log in to the admin dashboard at admin-ai-twin.drmalpani.com Username - drmalpani Password: sSTs2Sj-nG_DFe9d

Great!

Ai-twin dashboard is ready

<media>

All articles shared in this group are published now

https://www.drmalpani.com/knowledge-center/articles/the-most-important-person-on-your-ivf-team-is-you

https://www.drmalpani.com/knowledge-center/articles/why-i-tell-my-ivf-patients-to-keep-talking-to-their-spiritual-guru

https://www.drmalpani.com/knowledge-center/articles/the-second-opinion-trap-why-am-i-on-the-right-track-is-the-wrong-question

I do not have to publish this right?

https://www.drmalpani.com/knowledge-center/articles/reforming-egg-donation-in-india-from-exploitation-to-ethical-practices

https://www.drmalpani.com/knowledge-center/articles/don-t-walk-away-from-your-frozen-embryos

https://www.drmalpani.com/knowledge-center/articles/how-ivf-clinics-inflate-their-success-rates-to-take-you-for-a-ride

https://www.drmalpani.com/knowledge-center/articles/the-ovarian-prp-trap-how-desperate-women-get-sold-an-injection-that-doesn-t-work

https://www.drmalpani.com/knowledge-center/articles/why-every-ivf-patient-should-read-the-art-act-before-starting-treatment

Please share the URLs

There was a slight miscommunication, I thought Pankaj Sir had automated the article publishing, so I hadn't put them up yet. Now that I know it's on me, I'll have them all published by EOD

@267155589353519

I had shared a number of articles. Could you please confirm these have all been published and let me know what the URLs are?

Not yet. Just noting down all the suggestions. Focus was on building the dashboard first so we can evaluate video chat sessions more easily and to also allow people to selfreport issues. Tuesday tk this will be wrapped up. Then thoda fast-fowrad kr k issues pe work krenge

And did you figure out how to deal with the problem of patients asking three or four questions in a long sentence to the AI twin? What is your solution for that?

You also need to scrape my Gmail to automate my replies to patients

Cool

Exploring extending it to ApniPrerna so we can create an SOP for easily extending to other clients / websites

@179035711746252 This is the internal content generation dashbaord we've created for managing the needs across all our websites

Please share this with Dr Sai as well

Excellent

Ok

Basically, we don’t need to worry about this traffic. We should rather just add some filters so GA can skip over this useless traffic.

Here Are the investigation results

Regarding singapore bot traffic, cloudflare is showing 0 traffic from singapore. GA is showing such high vol. I talked to claude about it - shared it cloudflare’s data AND GA’s data.

If you are happy with this, I can get this live for you so you can generate it for any medixcel patient

Hi Sir You can find 4 patients’ mmedixcelanalysis reports here along with their raw files that were analysed: https://drive.google.com/drive/folders/1-FhNf2vno4GkztyVOtU3oFvTYg0Gx7nK

Too many bots are messing up our analytics and our traffic

<media>

Is this working ? Please track this

I am now sending traffic through YouTube where the call to action on the YouTube videos is that they go to the website

<media>

When people encounter glitches, we need some way of being able to track this so we can fix them

<media>

Meeting w Reddit sales team (Sumair Pasha) - Based in Portugal - Sen. Dedicated guide partner - Malpani infertility clinic - Sumair has Passed to Legal team for approval - Might get blocked & rejected because of industry sensitivity - Actionables: - Share clinic license - Share Creatives, headlines, etc. - Sumair will send to legal team. (takes upto 48 hours) - Once approved, Sumair will make the strategic plan, onboarding process, etc. - Timeline: Can be done on priority - Approval by Monday - Meet again by Wednesday - Since our clinic is registered in India, W…

I’ll also send the update to Zeenal

Ok

I’ll plug whatsapp notification so you receive an update form tomorrow

Already getting posted on MiceLabs & StartupMentors

So you will start posting from tomorrow ?

*SEO-Forge - Content generator updates* https://www.youtube.com/watch?v=ztSDvcQPmNo Dashboard is ready! Will make my life easier to generate new content and to also manage a lot of websites at once

*Mice Labs + StartupMentors + IndianAngel + IVFIndia updates* -> All had social media sharing / meta tags issues. Fixed it across websites. -> IndianAngel’s sitemap & robots.txt was completely deprecated. Fixed it. -> Installed YoastSEO plugin on all wordpress websites under mgmt for future automated meta tags review & warnings