Suman,
Many thanks for your prompt reply to queries raised by Manoj
To save your time, I have copied below, some of the mails that you sent to me in the recent past ( which may answer Manoj’s queries ) :
Manoj Queries to Suman / Past Replies from Suman
On Fri, Dec 23, 2022 at 6:26 PM Manoj Hardwani <Manoj.Hardwani@atidan.com> wrote:
Hi Suman,
Request you to please provision and share API credentials or API Key for Hemen Parekh’s Personal.ai account.
As per our understanding, currently we are only going to make Message API call for Q&A implementation.
Regarding Topics, per explanation from you end, “what do you know” question will provide us list of only 50 topics as of now.
Is there possibility to share all the topics in the response and not just 50 items. This will enable us to synchronize the topics at our end on daily basis.
We are also expecting the API that will specifically developed at you end for returning topics, will cover pagination and returning all the items.
Also, do you have idea when we can expect below details from Message API:
# Score or confidence, mood, related category, related topics, etc
# Accepting the user rating on the response, that may be required by you to improve
# Metrics related to questions asked
Regards,
Manoj Hardwani
24 / 12 / 022
Manoj,
Most of these questions were answered in a different thread. Let me find and forward it.
Meanwhile,
Api key is in the settings, follow doc here
https://docs.personal.ai/docs/
“what do you know” will update as there is new data and changes to topic information density.
If you want latest topics you can use “top of mind”
Scores api update is on roadmap; refer to thread.
PAST REPLIES FROMPERSONAL AI :
Suman wrote on 26 Nov :
Hi Manoj
1. Message API is in the same link on the left. You can find the api key to use in Hemen personal ai account in settings. Here is direct link for message api.
https://documenter.getpostman.
2. Most emotion is auto processed. We may need to extend api to read the emotion on the message. We will add this to roadmap to expose to external api.
3. Metric api also need to be exposed to external api. We can aim to update these in jan and you can have placeholders meanwhile.
Can you add this request to https://roadmap.personal.ai please ? That will help prioritize.
4. There may better way to promote Hemen ai and invite them in our newsletter and social media channels.
For privacy reasons member list may not exposed via api but we can promote via marketing channels and put links to Hemen ai when you are ready to launch.
5. Zapier links below
https://docs.personal.ai/docs/
==============================
Sharon wrote on 20 Dec :
Hey Hemen,
Would such separate / single ( unconsolidated ) uploads manage to create my “ Voice Clone “ , which can “ speak out “ in my voice, ANSWERS when questions are posed to my personal.ai ? Or , to create a “ Voice Clone “ require us to merge all 123 files into ONE file, before uploading ?
Right now - it only needs about 30 minutes of speech for the voice clone to happen. It actually doesn't matter that it's separate, the 30 minutes are usually split into single utterances (one sentence / utterance).
Holidays is a little tricky as I am waiting on a few people to coordinate on working on your voice clone but I hope to have good news by NY. The one that we created for me is very promising and I can't wait to hear yours as well.
Cheers,
Sharon
Re: FW: UPLOADING MY AUDIO FILE ( 20-12-22 ) :
Hey Hemen -
We are actually working on a audio upload version where you can do it yourself. Should be in production sometime EOW here and I can send you a note/instruction on how your team can do the upload.
Cheers,
Sharon
Suman replied on 17 Dec :
Lensa plan is to simply integrate the generation of your avatar for your ai within app. There are other options we are exploring but it’s an option if the user want to generate a few choices for their ai profile page.
Ai profile page is nothing but a simple home for your ai that you and your visitors see on your ai domain.
You will be able to invite users to join your ai workspace. Once they join you and them direct message much like any other messaging apps like WhatsApp although the difference is your ai can draft the responses for you alongside. Thu data overtime will be used for the personal chat model. We are looking at 4-6 weeks for this to be completely flushed out. Then you’ll be to invite and manage your own lounges.
Suman wrote on 16 Dec :
Makes sense. We will likely switch show memory to show sources or similar.
We added lensa integration as well to the roadmap for your avatar.
We will send your voice clones soon to get your feedback.
We have ai profile coming up.
Try message api with “top of mind” to return the topics. I need to test this myself.
Not sure if you tried image generation in editor yet. Go to editor by click write icon in message in your ai, type a prompt like “I was a kid playing cricket with group of Indian people in Mumbai urban area”. You can use that to power outputs for your writings and inspiration.
Talk soon. Let’s get your ai to the finish line.
Alex wrote on 13 Dec
Hello, Hemen
Thank you, again, for taking the time to speak with me. I enjoyed learning about your experience with the product and vision for an infinite virtual avatar. Your perspectives on influence and impact have me thinking.
Have a nice rest of your day.
Regards,
Alexandria
Suman wrote on 12 Dec :
this is very insightful Hemen;
can't wait for the simpler DM experience to be out with respective profile page with voice and avatar interfaces.
the interface and the entry point makes all the difference for personal.ai to be absolutely personal.
Suman wrote on 9 Dec :
Hemen,
Hope you are avatar is coming together. I know Sharon is playing with your voice clone to attach to your personal AI.
I need your help and your brain on a discovery process we are running.
The Personal AI market strategy team is conducting research to understand the challenges that thought leaders like you face when it comes to remembering personal knowledge, sorting this knowledge, and sharing it with the world.
Alex de Aranzeta (she/her) is new on the team and is interested in learning from you about your perspectives on this subject.
Would you be open to a 45- minute call with Alex?
Hemen wrote on 02 Dec :
Suman,
On right hand top corner of this page, there are following counters :
Ø Level 5
Ø Size
Ø Quantity
Ø Vocab
What are these and which one is more important ?
What action on my part, makes these numbers go up ? How do I focus on raising these numbers ?
Suman replied on 02 Dec :
Hemen,
You will have to upgrade to unlock level 5 which will create a generative ai mode from your data that helps generate content from your own stack.
The size and quality will continue increase with more data which will determine the breadth and depth of your stack respectively. Generally the higher the numbers the better and at one point it will start to plateau for one corpus of a mind.
Hemen wrote on 28 Nov :
Hello Suman,
This feature is encouraging and will enable me to upload those “ relevant “ news-links on my AI page ( not lounge page ), as suggested by you
I suppose, my AI will read the text / parse it to find “ keywords “ / calculate “ frequency of occurrence “ / create memory blocks / save in stack
Question :
How will AI “ associate / link “ the news item with those few PAST BLOGS on the same / similar “ subject / topic “ ?
How will AI conclude that this just uploaded NEW content belongs to the SAME FAMILY of my past BLOGS , cantered around SAME / SIMILAR topic ?
Eg :
Only this morning, I came across a news item concerning “ Sharing of information among 25 government agencies / departments / ministries / States “
On this subject, I sent my first suggestion in 2014 ( a BLOG sent to Cabinet Ministers as e-mail ) , followed up by 17 more blogs / emails
Followed by this blog / email of to-day > https://myblogepage.blogspot.
Will my AI notice that quite a few KEYWORDS ( highlighted in BOLD ), are the same in 2014 blog and in today’s news item ?
Suman replied on 29 Nov :
generally, memory block dates determine the time decay and responses, so technically it has an understanding of the time
Suman wrote on 28 Nov :
Hi Hemen, auto scrapping web is a bigger undertaking although we want to be able to automate data inputs from known sources and integrations such as emails, text, blogs etc.
We'll likely partner with some when the time is right.
With regards,
hemen
From: Suman Kanuganti [mailto:suman@personal.ai]
Sent: 24 December 2022 07:18
To: Manoj Hardwani
Cc: Alexandria de Aranzeta; Hemen Parekh; Kartavya Chitalia; Sahiti Bondelepati; Xiaoran Zhang
Subject: Re: Request some Clarifications
Manoj,
Most of these questions were answered in a different thread. Let me find and forward it.
Meanwhile,
Api key is in the settings, follow doc here
“what do you know” will update as there is new data and changes to topic information density.
If you want latest topics you can use “top of mind”
Scores api update is on roadmap; refer to thread.
On Fri, Dec 23, 2022 at 6:26 PM Manoj Hardwani <Manoj.Hardwani@atidan.com> wrote:
Hi Suman,
Request you to please provision and share API credentials or API Key for Hemen Parekh’s Personal.ai account.
As per our understanding, currently we are only going to make Message API call for Q&A implementation.
Regarding Topics, per explanation from you end, “what do you know” question will provide us list of only 50 topics as of now.
Is there possibility to share all the topics in the response and not just 50 items. This will enable us to synchronize the topics at our end on daily basis.
We are also expecting the API that will specifically developed at you end for returning topics, will cover pagination and returning all the items.
Also, do you have idea when we can expect below details from Message API:
- Score or confidence, mood, related category, related topics, etc
- Accepting the user rating on the response, that may be required by you to improve
- Metrics related to questions asked
Regards,
Manoj Hardwani
Subject: Re: Request some Clarifications
WARNING: The sender of this email could not be validated and may not match the person in the "From" field. Take caution when replying to this message.
Yep it returns top 50 for now. We can add to return more but requires some sort of pagination and handling on your end. I will add a backlog to pass in number of topics to return via api.
On Sat, Dec 17, 2022 at 9:14 PM Hemen Parekh <hcp@recruitguru.com> wrote:
Suman,
I typed “ what do you know “ , and my ai displayed 50 topics , with “ no of occurrences “ shown in bracket , next to each topic
Repeating this message , displayed the SAME words
Surely , there must be thousands !
Regards,
hemen
Subject: Re: Request some Clarifications
Use same QnA message api for topics, just send message “what do you know” and you can render per your design and sort.
Don’t have api yet for showing source but will add to backlog, hope that’s not blocking for now.
On Sat, Dec 17, 2022 at 4:01 PM Hemen Parekh <hcp@recruitguru.com> wrote:
Suman,
Thanks for clarification
As mentioned in my yesterday’s email, attached find a “ draft “ of revised User Interface for my Digital Avatar
During Manoj’s visit to my home on next Wednesday , I would like to freeze this for ATIDAN to start development
As stated by Manoj earlier , following functionalities would be developed by ATIDAN :
# Metrics
# Answer Rating System ( including data capture and forwarding to you , in a format desired by you )
# Storing / creating a database of all the questions asked by all the visitors and displaying these in box “ Questions Asked by Others / Click Any “
# Share
As far as Personal.ai is concerned, it will take care of following functionalities :
# Delivering answers against questions submitted by visitors { through API }
# “ Show Original Resource “ instead of “ Show Memory “ ( Pop up of full blog – note , which got broken up into the concerned Memory Block ? )
# Alphabetically arranged database of TOPICS ( frequently occurred KEYWORDS ) , obtained through parsing of all 96,000 memory blocks / once a week ? )
I request you to let us have your confirmation of my DRAFT U / I , by next Tuesday
Regards,
hemen
From: Suman Kanuganti [mailto:suman@personal.ai]
Sent: 17 December 2022 10:26
To: Hemen Parekh
Cc: Manoj.Hardwani@atidan.com; Sahiti Bondelepati; Xiaoran Zhang
Subject: Re: Request some Clarifications
Lensa plan is to simply integrate the generation of your avatar for your ai within app. There are other options we are exploring but it’s an option if the user want to generate a few choices for their ai profile page.
Ai profile page is nothing but a simple home for your ai that you and your visitors see on your ai domain.
You will be able to invite users to join your ai workspace. Once they join you and them direct message much like any other messaging apps like WhatsApp although the difference is your ai can draft the responses for you alongside. Thu data overtime will be used for the personal chat model. We are looking at 4-6 weeks for this to be completely flushed out. Then you’ll be to invite and manage your own lounges.
On Sat, Dec 17, 2022 at 12:02 PM Hemen Parekh <hcp@recruitguru.com> wrote:
Suman,
In your yesterday’s email, you wrote :
“ We added lensa integration as well to the roadmap for your avatar “
I do not understand how exactly will this work .
In the meantime, today’s Economic Times , shows following article :
“ Cyber experts train lens on Lensa-like apps “
Extract :
“ They point to the possible misuse of sensitive personal information such as face identification and Iris Scan and warn users to exercise extreme caution while uploading their photos on any such apps , which have of late begun to face greater scrutiny for their privacy policies and security features “
And what did you mean when you wrote ?
“ We have ai profile coming up. “
Regards,
Hemen
PS :
I await Sahiti’s explanation re process to be followed for uploading my AUDIO files on Personal.ai
source > https://docs.personal.ai/docs/
terminology ] A creator can upload information to their memory stack in the form of audio files, text files, and URLs.
A creator can sync information to their memory stack by recording their audio feed and connecting other apps that stream data continuously
No comments:
Post a Comment