Unknown

okay we are live okay hello everyone so welcome to mpow live stream event and we are very happy to welcome you here today uh my name is Chang and I'm currently a product engineer at mpow and I will be hosting today live stream event session so for those of you who are new here or just getting to know us mpow is all about unlocking the power of AI for your business so at its heart my power will help you to build AI agents and multi- Asian workflows to automate thousands of tasks across your business and

today we are proud to have welcome over 2,000s of customers worldwide so thank you all for being part of this incredible journey and my pal could not be here uh supporting many businesses without our our founding me our Founders so please welcome Mr D and miss Tom uh who uh are founders of myp hi everyone I'm Tam or siia uh you may know me through uh my past YouTube videos uh I'm very excited to have the chance to show you uh my P live instead of the pre-recorded videos uh just like my prior YouTube videos

so very excited today yeah and I'm uh also found of my p uh I'm not uh exist on video frequently as time but uh also very excited for today's live stream and the uh Q&A section too so uh to kick out today's event uh let me give you a quick uh walkth through of our agenda so first we'll take you through a live demo of mpow led by our amazing uh CEO Tom and after that we'll open the floor for Q&A where I will be hosting and both of our Founders will be joining me to

answer your burning questions and of course we'll wrap up with some exciting information that you won't want to miss so sounds good uh so now miss Tom will walk us through the magic of mow and showcasing how we can create AI agents real multi-agents workflows and use the incredible and newly launched canvas tool on mpal and let the live demo session begin thank you so much Dan uh you make it sound so exciting okay let me uh share my screen okay can you guys see my screen now very clear okay um so my pal is

a platform for you to build AI agents and multi-agent workflows uh for your business so okay before me start let me Spotlight myself first so that I would show up on everyone's screen okay um and this is the workspace you would have on my pel uh this is the homepage where you can uh easily and quickly access your army of AI agents and multi-agent workflows you have trained on myel now an AI agent on myel is a specialized AI assistant trained to do a specific task for you to uh create a new agent is very

very easy all you need to do is to describe what kind of job you want an agent to do for example let's take ghost writing a LinkedIn post as an example you can also optionally provide an example output of this job so that we have a benchmark of what kind of output you looking for this is especially useful when you are dealing with uh more uh more complicated tasks such as writing a report or an analysis for now a LinkedIn post is quite straightforward so I would just go ahead with generate with AI and what

we do is that my will automatically suggest an agent for you uh to uh that is uh well suited to do this specific job then you can save this agent into your workspace and you will be redirected to this studio where you can further customize this agent so that it works exactly the way you want here uh the background field is where you can explain the job description of this agent so any specific requirements any expectations any contexts this agent should have to perform this job well you should put it all in the background section

the design output format is where you can specify the exact output you expect to come out from this agent in this case we're looking for a LinkedIn post that include a hook a personal anecdote or Insight actually I don't think this is necessary so I'm going to remove this and some lessons and a call to action um and I don't think hashtags are even necessary so I would just remove all those things now the next um parameter you can improve on is knowledge and web search so just like it name suggests this is where you

can configure uh what types of data this agent should be trained on so on my P you can easily train any AI agent on any custom data that you have uh it could be your website it could be any file you have and the good the good news is that we Pro we support most common five types including PDF word um CSV Excel audio video so feel free to upload all the files you need here uh as well as the websites too and then you can conveniently connect the uh knowledge sources that you think are

relevant to the to this agent uh by connected here so for example uh let's say I will connect uh this agent to all the knowledge sources I have about myal so that when I ask it to write about a topic about myel it will have it will already have the context about what my is what we does what we do so that it can write the most meaningful uh post possible um web search sources is where you can give this agent uh external knowledge sources from the internet so it could be Google it could be

more academic sources like semantic scholar or ARF uh so it's up to you to decide what kind of knowledge sources are relevant to this job then you can also customize the output of this agent the output can be further customized by specifying the language you prefer uh by default the agent will answer in English but if you want Vietnamese for example you can put the language here we support a lot of languages here so feel free to pick the language that you uh that you need and another thing you can customize for the output is

brand voice this is a very very exciting feature this is where you can teach the agent to write in the exact style that you want uh we all we uh you are probably uh aware of the problem that uh AI generated content could feel really uh robotic and unauthentic uh by default uh so what you can do is you can teach it a brand voice of your own that is unique to you or your company and the AI will learn that brand voice to mimic that brand voice for all the content it generate so this

will ensure you will get the output that is that is the most authentic and humanlike possible so how can you create a brand voice well it's super super simple all you need to do is to create a brand voice here and provide some example content uh that you already uh have you already have so in this in the case of a LinkedIn ghost riter we can give it some out some uh previous LinkedIn posts that you have written so that it will understand the style that you want here so uh I will go ahead and

copy paste some of the uh LinkedIn post that I have written before I will paste it here and then I will click analyze and then my power will automatically do the analysis for you analyzing uh to fully understand uh this brand voice uh so and teach this agent to best mimic this brand voice later on okay a new brand voice has been created here so maybe I can uh uh rename it to confident uh LinkedIn style for example then I can connect this St to this uh to this agent okay another uh aspect you can

customize is model so this is the AI model that is that will that would power this agent you guys are probably familiar with the gbt models from open AI because J gbt is very very popular and when you are using J gbt you are interacting with the gbt models from open AI but in reality there are a lot of other models coming from other companies too so the good about myal is that you can freely select any AI model that you think is best for each task that means you are not limited to any certain

model but you have a wide wide range of choices to choose to ensure that you always have the best model to do uh one task so for example uh CL 3 profile Sonet is uh is um famous uh to be very good at creative writing and coding tasks so you can use it for task like writing a LinkedIn post but maybe for tasks that require more structure output like writing a report following a template you can use gbt for that and we also have models from Google Gemini and uh Lama from meta for you to

choose here too okay you can also customize the max output tokens so how long the output should be here and the creativ the temperature or put simply the creativity level for this agent too and tools tools is where uh you can give extra capabilities to this agent uh by default the AI agent can only write text for you right but what if you want this agent to generate image generate audio uh search for stock photos or access any custom tools you have well this is where you can do that you can easily select aside any

tools to this agent depending on the job to be done of course like generate image uh generate image with flux B generate speech uh screenshot the website search for news headlights search for stock photos video photos and many more you can even add your Custom Tool too many of our users have using this Custom Tool feature to connect to many capabilities like search for 3D models from a 3D model library or uh or many more things so this is where you can give your agents more uh capabilities uh than the default uh settings okay now

that our setup of this agent has finished let's start working with this agent so the most most straightforward thing is that you can call this AI agent here and simply give it a task uh I will uh I will uh make sure to connect uh this task to all the myal customer support uh to uh materials and ask it to write let's say a post um why uh explaining explaining why um multi-agent workflows could be useful for automating business processes okay let's hit send and what this agent does is that first it will read uh

the whole knowledge sources to um gather the relevant information about the request and then it will uh generate the post for you as you can see here and uh we are we ask it to uh explain why multiagent workflows could be useful so it all it focus it focuses on the benefits of multi-agent workflows here including uh leverage unique strengths of different AI models that's correct delivers more comprehensive and balanced outputs that's correct because we have multiple agents working together here and enables uh complex test automation which is correct too and if it uh it

get the data from my knowledge sources so that's how I can ensure that the uh that the output here is based on uh actual data and not and it's not something uh this AI makeup so that is this is how you can work with an AI agent on IAL and following this same logic you can build virtually any AI agent for any task it could be um like um the a newsletter writer for you a graphic designer to generate illustration for you uh a uh video script generator here as you can see a customer success

story writer um or uh job job application reviewer so many so many AI agents you can build but uh in reality however we have a lot of multi-step complex processes that not a single AI agent can handle for example uh when it comes to writing social media posts in a workflow we would we often would need to write post for different channels not just one so uh not just uh LinkedIn post we probably would also need a Twitter post an Instagram caption threats post and blog post and email or when it comes to let's say

uh writing a blog post for example um the blog post gener writing process itself could involve multiple steps from researching for keywords coming up with the outline writing the final blog post um checking that blog post against some content uh content quality check uh criteria that we have uh and many more so and uh so for such complicated processes uh a single AI agent is of is me is often not enough to handle because when you try to squeeze one instru like a big instruction into uh the AI agent it often backfires because the agent

works best when it focuses on one specific task and um and that's when that's when our multi-agent workflow feature comes in very handy just like it name suggests this is where you can connect multiple AI agents on myel together into a system so that these AI agents can collaborate with each other to solve uh more complicated processes so uh for example let's take the exact uh example use case that I mentioned before taking a piece of content and repurpose it into different uh channels and formats uh for example uh to set up an AI agent

AI workflow like that uh it's also very simple just type down what kind of workflow you need here okay I think actually I don't have to type because my P has built in uh Speech to Text here so I will just click here to explain given a YouTube video link um turn it into a LinkedIn post a Twitter post a blog post and a marketing email to promote that uh YouTube video okay so this is the process description now let's generate this AI and my P would suggest a uh multi-agent workflow for you based on

your description uh okay so uh what what what this workflow does is that it takes in this YouTube video link from me then um actually this step is not needed I will just delete this uh then it will create the LinkedIn post create a Twitter post uh write blog post and draft marketing email sounds very good now what I can do is that I can assign an agent I already have in my library to execute this step so I will select the LinkedIn um post writer that I have trained to help me write LinkedIn post

here uh I will ask my Twitter post agent to do this step and then I have my blog post uh writer to do this step okay let me select this and then I have my email sales email writer to do this marketing email step now uh my a workflow looks pretty good so let's save this workflow and we will be redirected to this uh AI workflow Builder where I can further customize this AI workflow so that it fit my exact needs and you can see that there is a warning message here don't worry there there's

nothing wrong with this this is just a friendly reminder from um from myel that there are currently no human input values or other steps output mentioned in this step please ensure you are not missing any input needed for this step and what this means is that when you click on this uh step you will see that uh this step is carried out by LinkedIn post writer based on uh and this is the task given to this LinkedIn post writer for this step now the problem is that uh there is no uh we haven't mentioned uh

the uh YouTube video link here so that if so if we run this workflow this uh step would not have uh information about what link uh given to make the make the post so um this is because uh uh early uh earlier I have deleted the step extract key information from the YouTube video because that step is not needed so that's why there's no um input um uh associated with this step by default here but what we can do is that we can just provided the um create a uh LinkedIn post uh based on based

on uh this YouTube video and then I can mention the human input value which is the YouTube video link provided in this first step now I will copy this and past it to all the other steps here because all the steps here would need access to the link so that it know what content is mentioned in uh inside the YouTube video okay okay now that my AI workflow is ready um you can uh always customize it further if you want like for example in this input step you can always add new steps here or you

can change the type of input um like it could uh this is a link but but if you want to process files or images or just plain text you can also change to these input types as well and uh these four agents will come together to do all of these content for me all in one go so let's try it out shall we so that we we will uh we will know how it works so this uh let's um create a new run with this AI workflow and then I will uh pick uh an example

YouTube video video here let me just pick a video uh from our YouTube channel start reading this uh link so that myow can process and understand what is inside this link okay my has successfully processed this link so we can start this workflow and now as you can see here what my pow does is is that we have uh different AI agents first is the LinkedIn post writer uh write a LinkedIn post about the video and this is a video about five AI workflows for sales so yeah so it's writing the correct one um and

this is the Twitter post uh generated uh for the YouTube video and next stop is the uh blog post and all of of these uh agents have been trained on my PO writing staffs brand voices uh Etc so that it it write uh exactly the way uh I uh want so I won't have to edit it too much which is very awesome because that means it would just take me uh a couple of minutes to get all of these um content ready for me to distribute okay now as you can see in just a matter

of minutes we have got a LinkedIn post a Twitter post a blog post and a marketing email all in one go so good right now um another another cool thing about myel is that you can actually process multiple inputs at once so in reality for example for this uh YouTube video repurpose work workflow uh what if you want to repurpose many YouTube videos at once does that mean you have to um manually input each uh YouTube video link and wait for each to finish well the good news is no because myal has a buil-in book

run feature and what that means is that you can add multiple um rows at one like this for example I can add uh two rows at once then you can provide the uh input here this is another YouTube video of my pal and then another one and as you can see the status of these two items would be turned into ready to run and that means when you select this you can run both two rows in background and uh and you can go do something else uh so the process the status has been changed in

progress now that's great so you can go do something else and my power will handle the processing in the background for you so you can totally check back in a couple of minutes and the results should be ready for you to review we can refresh it a little bit to see how the status is okay as you can see we have uh both um both LinkedIn post ready this is for my video on my power for SEO and this is for my video for my for human in the loop workflows so good and yeah the

so the update the uh the results will be constantly update updated like this for you very good right and that means you can process multiple YouTube videos at once and don't have to wait every time now uh okay so we have built an AI agent and a multi-agent workflow together and as you can see it's actually not that hard to build an army of AI agents and multi-agent workflows like this on my P but following uh but by learning this this uh process of building AI agents and multi-agent workflows you can actually build any internal

AI productivity tools for your team to automate a wide range of test for example you can just take a look at our template Library here to get a sense of what's possible with mpow with the same logic many people have come up with so many wonderful ways to leverage our AI agent and multi-agent workflows to automate thousands of tasks so it could be something like uh generate a whole ad campaign here when you can provide um the company link business type target audience campaign type things like that and this AI workflow will do everything from

brainstorming different angles uh create the sales page um create the ads for different uh channels here and then write the email sequence write SMS text message sequence voic mail sequence script um all in one go uh or people also use it for analyzing sales call for example you can upload a sales call recording here and get the transcript the call insights um the sales rep performance evaluation Next Step suggestion and even a followup email all in one go you can also use it for uh test that involve evaluation for example some people have you used

us for landing page audit so you can provide the landing page screenshots or landing page link and this workflow will perform a comprehensive uh audit from design copywriting storytelling SEO as well as suggest uh actionable um takeaways for you to improve on your website or even something like uh evaluating a job application uh by providing the CV of the candidate and uh get a comprehensive analysis covering so many aspects and these are just templates and they are free so feel free to uh to check them out and duplicate them to your own workspace for further

customizing ation if needed so that these workflows would work just the way you want now uh before before I uh uh I end the demo uh section I would like to show you yet another exciting feature on my pel that we have just launched earlier this week so so far you have learned how to work with AI agents and multi-agent workflows in a linear stepbystep structure uh that means you can uh run any AI agent or multi-agent workflow and get the results uh in the familiar um like chat interface right but what if you can

have an infinite uh canvas to run any AI agents any multi-agent workflows and even many of them at at the same time in uh side by side or in sequence all in one space that this is where our canvas feature comes into place and uh on this canvas uh this is uh very familiar like this is still the magic panel that you used to run the AI agents and wol flows previously but instead of uh getting the results in the chat interface I will show you the difference you can select it here you can uh

okay you can still connected to all the um knowledge sources and then you can ask it to write about the same thing uh write a post explaining why multi-agent workflows are useful for business process automation let's hit send and uh myel will still uh handle the process uh the knowledge sources to get the relevant information and then this LinkedIn post will write the post for you here and but the but here's the difference this is no longer the linear chat interface this is actually an infinite canvas so you can uh so that means you can

uh like do cool things like for example you can use this same um same and this same uh instruction but for the Twitter post post agent for example write a post and then it will also take the same instruction and then help you to write the post here and as you can see it's like you are leading two conversations at the same time with AI you don't have to create a new chat it's it's all happening in one space like that and this is even more powerful when comes to running AI multi-agent workflows right in

this space so I can do something like repurpose uh a blog post for example okay um I think let's go with this uh okay and then I can provide any blog post uh link here this is a blog post from myal and then I can run the AI workflow right in this space so as you can see this um this means you will get everything in one goal in one uh space like this you can run AI agents you can run multi-agent workflows uh or in one same space so this is uh it has uh

repurposed this blog post to a Facebook post for me and then it will continue on for the rest of the formats too and you will get all the results shortly and this is um this is very useful when you are having a big project that involve multiple AI agents and multiple AI workflows you would be able to run all of them here get all the results without having to switch contacts or switch tabs and you can even chain outputs from one AI workflow to another so the output of workflow a could become the input of

workflow B and that's where this canvas would be very very useful okay I think uh that's it for today's live demo I hope that you guys have uh fully learned uh what is my power about and how you can maximize the value of myel to automate any tasks and processes in your business be it as something as simple as writing a LinkedIn post to something as complicated as analyzing a sales call comprehensively uh thank you for listening and I will return the stage back to my uh uh my MC uh jity so uh thank you

miss Tom for the very very uh comprehensive live demo so uh how do you guys feel about the live demo is there any part that you want us to dig deeper into or like is there any burning question that you have so as coming uh as coming up right next we'll be having the Q&A session this is your chance to ask all the questions you have been saving up and we'll also have our Founders right here ready to listen and answer all your question so feel free to send your questions in live uh first we'll

start off with some uh frequently asked questions uh in which we have gathered from you guys and then we'll be taking the live questions from the chat uh starting first with the first question of how is mind power Asian different from Custom GPT um B I think you you can and uh answer this question yeah so uh compare with uh the custom gbt so uh we have several things to uh differentiate uh but I think there are uh three uh most important thing uh the first thing is that um my power agent is the uh

can be connect with any kind of last language model which is the engine behind the AI agent and different from gbd which is uh close and be St with the open AI ecosystem in my power you can connect uh the agent with any model uh from DVD 40 to anic or Google seminars and each model actually have its own uh strength and and weakness uh for example um CL uh very good at uh logic uh re listening writing and and coding uh zi from Google it have a massive uh context plan so it's it's optimiz

for the task for um that require reading a lot of document or when you need something fast and and um a lot a little bit cheaper to run uh um Lama from BET is uh better than the other and the last we model SP is uh rapidly evolving and as I know there are a lot of model um comparable with uh gbt even better in some aspect so it will be a big leverage when we can uh freely use any kind of model and the second thing is that uh um on my power we design

agent uh specializ for business purpose uh different from a generic um assistant that you just can provide a background and a knowledge and something we know that in a business we often uh have some uh integration need or brand Bo need or or knowledge Source need or some more Advance setting to connect with the tune that we you in the business for example uh time share the you SK about using to connect with a uh 3D uh model API that uh we support one of our customer recently use to publ uh VR and AR move

from them or you in like the brand voice to make sure that the output from a is exactly what you want for the brand and fully like production ready for the marketing effort um or connect with the necessary tool um to uh to uh integrate with the way you work so that is the second thing and the final thing is it can be connected with the other because we work in the business uh with a lot of process that we have been familiar with and to fully automate the process uh we need those agent to

work together and it's uh it is uh the thing that leverage the first thing I said when we have multiple agent working on specific thing with different model Leverage The the the advantage of Morrow we can have a nearly perfect uh uh AI tune for for any kind sky in business yeah fantastic uh answer I totally agree with th and I just want to highlight more about how how powerful it is um to uh be able to connect multiple AI agents coming from different AI models in one same work flow I have seen so many

impressive ways people have um have leveraging uh this feature for example in one workflow you can have both uh gbt model and Cloud 3.5 onet working together to do something for example if you wanted to brainstorm something for you instead of relying one model what if you can have uh multiple AI models uh gbt 40 CL 3.5 uh Google Gemini Pro all come up with the different V uh uh the the different ideas for you so that you can pick the best one uh to go to move forward with or you you would be able

to do like um super comp uh super uh complicated use cases uh that I have I even uh see many customers building AI workflows that have 20 or 25 different steps in one workflow and the ability to handle so much input so much data and produce such a like I would say comprehensive and uh meaningful output it's a huge um it's a huge power that uh if we know how to leverage uh wisely it would be really GameChanger for any business yeah um I think uh we we are having uh quite a lot of uh

Live question from the YouTube here and uh I think we should uh answer each question one by [Music] one uh okay so uh one question we got uh from Cole uh is that can can we get it to speak like J judb open AI uh real time uh so uh in my pal we we have uh we have the T fature and in the tune feature you will see the option to uh we have a default tune in the tune template that uh that is the Zer voice with open Ai and we use the exact

uh uh take to speed model for open AI for for that use Cas and either case that you want to uh there and speak uh let the agent to speak orate the voice orate the podcast for you after you have already have the script you can uh turn on that tool and assign it to the agent okay great another question is uh from Antoine uh could you give an a use case for canvas in a real business uh so uh canas is uh uh a very new a very news and very fresh new uh feature

on myp and uh we we are still uh observing uh user to to see what uh what they actually use canvas in the business but uh I can share a little bit on why we build the canas and uh what is the like the inspiration why we use the canas feature from the US okay so uh uh the the Inception of canvas uh came from when uh I I work with uh a lot of uh customer that I that they request support from me uh to build a lot more complicated workflow uh not just the

workflow have four or five step but the workflow with maybe 10 or 20 step and uh it require multiple workf flow together um for example we we can have a workflow to to Z per campaign um and then for a campaign we have uh 10 or 11 element in the campaign and from each element we will have another workflow to develop a more detailed version of The Campaign from that so basically the output of the first workflow will be the input of the another workflow and in another workflow um after we get the detail um

uh campaign version we will often time require to edit make some change edit and double check and and maybe let the agent to regenerate uh the the output or we want to compare the output from multiple model for example and basically uh previously uh with the linear chart interface we need to back and forth a lot of time uh like we had to go to this Stu this stuff and we copy and paste a lot of time and it's it's really annoying when when the scope of the product it get bigger and bigger and uh

and for example what it in the process of the asan and the workflow uh why is generating we we have some genius idea uh and and we we want to to to like uh uh add that idea into the flow what did we do in the linear chat so if will be not possible in just St so we we we thought that uh maybe we we we need to um have a nonlinear version of running the workflow to make the collaboration between many workflow possible and uh have the user when they go into the deep

work mode um they can they can think and work like human thing like we think is a nonlinear way we back and forth with idea and we uh we often bring out our thinking into multiple uh version and we try to compare them to Pi what is the best work first so that is why we bu canvas for so in z um uh from uh from that reason um I think canvas is best for some kind of workflow or section where the scope of the product is big uh for example you you have to build

a um content schedule for for the company in the next two month for example um and the scope is B and you need a de work section on that it involve multiple workflow and it require collaboration between team member in your team for example I I generate something first but I I can ask them to generate other other person from herion and we can compare um the version in in in the in in the in the ranging thread and we compare that and we pick the best for first uh with the with outs so I

think that is the way uh the the E that um I I think uh will be the best for the conas and another case it could be for researching uh purpose um or writing a book purpose I see that uh I saw that in the uh education uh when we deot the canvas uh people immediately they think about writing the paper and and writing a book because uh a paper uh we often have a lot of idea coming up in the process of we doing research it it could took months and it involve a lot

of small idea uh but it could be novel but the the important thing is that we need to connect a lot between a lot of uh uh idea that we in some way we just have an idea and and in the canas um it's very easy uh you can select the the card uh the zated card and you compile all of them into a workflow na to generate and brainstorm idea with you and it could connect multiple do and you can visualize it in the way that support the thinking uh instead of being constraint in

the linear chart interface yeah thank you uh we also got another question from Bruno uh this is for the tools uh tools uh of Agents can we protect the tools with login and password um currently uh we we will uh we we do not have the the mechanism to protect uh the tun with uh Lo in and password and and uh we uh for for for the for example in the case you share the tun uh for example you share a chatboard or you share a a workflow to other people uh the um the only

protection uh we are supporting right now is the uh payment note when people have to for example feel in a form for lead generation or purchase uh uh the price for the for the workflow to continue but uh uh we are pretty open and and if if the business is SC of the uh L Union password is something uh that is the common interest of the community uh you you can feel free to add that into the the uh the to our road map in uh my part of K iio and we will consider it

and we after we discuss and we see that that is the common interest in the community we will try to add it into the product uh in in the shortest time okay now we got a questions from uh just uh um how do you easily manage and understand the number of credits that are being used in each workflow um so uh uh is a workflow uh uh we uh so let's me explain a crate first and then I will explain how we can manage in the workflow so uh in my pal uh the credit and

all the AI credit is the uh is the unit of execution of and every time the model Z something something we will C the credit into Z work and the number of credit consumed by each asent will depend on the last language model you use uh and the bigger model and a I I can call a smarter model will uh consume more credit than the smaller model and uh for example uh if you will CBT 40 min is Will consume only one credit uh but if you use for example um gbt 40 it will be

10 credit or cl5 stet it will be 10 credit and uh either Cas for the workflow that involve multiple AI agent the total credit consume uh by running a workflow will be the sum of the credit of xation in that workflow so for example uh in the case of uh repurpose YouTube video like th share previously we have four step in the workflow and uh for example uh if uh uh every agent in in the workflow is DB for many the total credit consumed by the workflow uh when is running one time will be four

credit but it uh come change the the um the model it one step to be clock Sonet it will be 10 plus 1 plus 1 plus one it will be 30 a crit in one run and that is the way uh we um we uh uh calculate the AI credit for Asian and for workflow it essentially for Asian and workflow you can think of it as the sum of the asan included in FL okay thank you uh uh we have another question from just can myow create chatbots that can be embedded on websites uh the

answer is yes uh for every agent uh that you have viewed on myel you can create an AI chatbot and you can share that chatbot with others via a sharable link or uh via embedding on your website as an iframe or a jat Buble widget that will stay that will float on top of your website and stay in the corner of your website Um this can be done very easily on myal and not not only that uh on mypal you can even uh publish the motivation wflow you have viewed as a form and share it

with others with a public link or by embedding that whole workflow into your website too and that that has been how many of our customers have been using myal to build uh to build not just lead magnet tools like build these AI jet Bots and multiagent workflows as tools as Le magnets to put on your on their website but also as things they can share with their audiences or customers um to add into the existing uh package of products and services uh and another question we have here from Alex is uh can you help uh

us with uh the best way in terms of uh formand and structure to make notes to use as knowledge base so notes uh if you are for some context notes is another uh feature on myel uh where you can uh create notes and edit it in a notion like editor um and you can uh constantly add uh update information to your notes and the good thing is all of these notes can be used at as knowledge sources for any AI agents so could you help with Alex here how can we like what are some best

practices uh of using notes as knowledge sources for AI AG yeah so um uh so uh uh so let me share this thing that is uh when we when when we uh first released the note feature last year we we call it the working memory of agent because uh uh uh the note when we release uh we have two purpose the first thing is that uh for for writing for example when agent is finished there is something but often time we will need to make some edit uh maybe a big edit or maybe just a

small edit but essentially we need to copy it into something and make some change so uh so uh uh that is why we build a node because uh we cannot like copy and paste into another SP like Google do it's is so not convenient so we build the note so you can open the note directly after the asent there is something in the thread and uh you can use the note as like a notion like uh editor it have full function for you to uh writing quick AI continue writing or edit a specific part in

the in the article with AI as and uh but another thing that note is useful is that you can connect a note into an also as a knowledge source and every time you make some change in the note that information that knowledge will be sing with a agian so that's that is when for example uh you build an agent uh as a customer support chatboard for example and uh you have a list of maybe frequently Asked question or a list of information about your company and uh uh if you put it into a website or

or or on a PDF and for example uh in some uh in some days uh the policy change or you have some campaign some new campaign that are going to be launched for example you have a sale campaign for example and that information is not available in the customer support Chan that the uh previous way to do is that upload another PDF or relearn the the we side is so not convenient for a team of many people to come and edit together and make sure the chatboard uh provide accurate information so the note is the

way that you can just edit the information in the note and the chatboard will will have fully s knowledge for you and connect the not with an agent is actually pretty simple as time share in the demo in the in in an asent uh there is a section called know source and website and in the know source and website you have three TP uh the first TP is internal no Source it means the the file the website the video the the the um PDF the word document the PowerPoint that you uploaded into my P the

second thing is website where you can select Google or or semantic scholar or um uh all some engine to uh sh for stop images or stop video for you and the final thing is the note that is a smart note that where you D your uh TI of working working memory because it's could be updated uh instantly uh uh when the note change and we are even we have the plan to connect the canvas as the new noise source to because canvas it actually could be even more advanc way to connect the uh memory to

the agent compared to the note so uh next up uh we also have a question from uh from uh a frequently as um of uh myal user is how is myal multi- Asian workflow different from tools like make or zapier so do you want to uh answer uh this question uh miss Tom um okay so uh uh myal AI multi-agent workflows uh we uh with myal multi-agent workflows we focus on helping you connect connect multiple AI agents to do Aid driven tasks like uh research analysis evaluation content generation uh while make or uh zapier uh

this uh cross app integration platform focuses on uh helping you connect multiple apps uh into a comprehensive business logic uh I don't think uh myal um um and make or zapier are competing against each other actually on myal you can use myal in combination with make or zapier to perform more complicated uh cross app uh business automation logic uh with our uh public API and web hope support you can trigger any AI workflow on myel to run uh from anywhere from any other app uh and you can even uh after uh an AI workflow completes

the results you can also send the results to anywhere you want so this logs uh use cases like for example we can you may receive uh like you may have a form that uh receive um submissions from people and uh for every uh new submission received you can run an AI workflow on myel to do something maybe processing analyzing it things like that and then you can the the finished results can be sent somewhere uh like it could be sent to your team or to the user uh as a report or something like that uh

and uh all uh all of the use cases that involve multiple apps like this can be achieved by using my house uh public API and web hop in combination with zapier or make any uh cross app integration platform of your choice yeah I think uh I I think I I add some uh some some more points with some comparation here I think in the big picture uh uh the thing is that uh what an automation platform not just s here and make and the other uh platform trying to do is that uh connect between multiple

app together to automate some some kind of thing for example data entry or uh web scraper or updating the database updating the spreadsheet or posting on social media that is the the common use SK of automation platform why there are many um UK that is AI native for example writing the content for example that is not uh the thing uh those platform I desire to do in the first place and uh when we come to more specific and more details in the UI and the elix that we uh that we do in the app for

example in my pal um if if for example the UK about research something and then write something uh that is the AI native UK uh often times you you need a streaming interface for example uh this agent work together to to to do the thing for you and you want to have the human in the loop uh the human can will that will check the result give some feedback for AI to re regenerate or to continue to work on on that specific part or even with canas to work in another way to to uh connect

many workflow together and in my power um the way we design uh the workflow the automation is that we abstract everything into patient like like we work with human uh we we don't Define human like a tune for example we don't Define human like Facebook tune or or Excel tune or Google sheet tune we we Define human as a ro for example uh SEO writer uh researcher uh investment analy or or something like that and the way um a Asian work will be by our natural language we we we def a task to uh for

agian um tell them about rules and let's some connect to each other via process and that is natively the way um how human work and those AI native us SC could be only be abstract in into the process to do because uh for example um uh I think last month uh I have seen uh I have met an user he said uh he can do something like my power in for example like the process to to write something with the brand voice and and then he Shar with me a massive uh M scenario uh he

made on mik that is involve of I think nearly 20 step to to just something and for example a common my power workflow in of our customer involve for example 20 step it could be a uh 200 or 300 step on M because M connect up is not connect as so it's need a lot of customization to do such thing and it's not optimal because it will let you wait for example uh like um Zen or even 30 minute to just complete that process uh for simple process for in my because my power is designed

for the AI R UK and I think in the future it will be more and more uh complicated uh use SK uh with a I uh involves uh the way we execute T in parallel for example or maybe in the future we don't need to uh explicitly tell the agent to work in that process we just Define the process and we have an manager agent to tell the agent what to do uh in a even nonlinear way in the native way so I think that is the the the real differ here so a next question

is what are the most impressive AI agents or multi-agents workflows you have seen people build with mind pal so can I invite uh miss Tom first uh yeah thank you uh I actually I have seen a lot of impressive AI agents and worklow Ms with myel uh but I think uh one of the most impressive ones I have seen firsthand is um uh is a a set of AI workflows one of our customer have viewed to help him um uh process medical appointment recordings so uh for um uh this is a Healthcare company and for

a typical Healthcare company they have to uh write meeting Recaps for a lot of medical appointment recordings like one month uh in uh in any given month month I think they would have hundreds of meetings like that and uh normally uh pre PRI prior to myel they often had to hire um a PhD level staff just to write the uh meeting Recaps because it's not just any meeting recap it's a meeting recap that follow the clinical standards and it needs to cover a wide range of aspects from the clinical standard nodes to analyze in uh

sentiment of the uh of the patient evaluating the staff um whether they are moderating the section well um then coming up with a personalized plan or program for that user and so many more and not just that there are different categories of medical appointments too so there are appointments for completely new uh customers but there are appointments for for returning patients uh and many more so what he built is a whole system uh of different AI motivation workflows each workflow for each C category of meeting and now every time he has a meeting he just

uh added that recording and my pal will have I think each workflow has somewhere between 10 to 20 steps to process uh this uh recording and this has saved him a lot a lot of time and I think this is so wonderful uh because he now saves um a lot of resources for this mundane Tas but at the same time he has um he get got more actionable insights to improve uh his business um and yeah I think that's uh that's for me how about you what kind of uh workflows that you have scene that

are the most impressive uh I think uh um uh so um I I um I actually uh know a lot of impressive uh in in in the fields of marketing space uh for example uh SEO stop um website all stop uh generating um advertising campaign zing something related to goto Market team a lot of things like that and but uh the most impressing thing um workflow I I actually knew by accident uh because um that is a time when um there there was a customer uh she uh she build a workflow and that workflow will

uh will receive the data from the CM uh and then we trigger to run in my power to do a lot of things like and Lead uh enri the lead uh with information then we categorize the lead into segment of customer and then write a lot of things for example analysis about the lead and write collect the data write the a lot of content Tor for those audience and give some uh maybe even free resource in that workflow to that is the workflow that involve around I I remember around like 80 or 90 step even

100 if if she improves over time and the problem at that time is that she want to connect that golow to a cm and B a call with me with a and I I I I I so so then I I look into the account to take the databas about the workflow and I I I I was like completely shock because someone be a workflow so massive like that in my power and uh another workflow I think that is quite impressive too uh also about lead enrichment I I I think that is a kind of

feel that uh have a lot of impressive thing uh from a customer uh he he be a workflow to to his community uh to help people identify uh the the the the the information of the lead and uh have some clarify about the ideal customer profile I actually don't know a lot about the Feld so I may Mis uh interpret but that's so Sky flow is pretty impressive and even uh those people uh previously um they use a workflow and then they try to export it into a PDF but uh the PDF is so big

because the information uh produ is so big that even for a lot of people it cannot fit in their uh in their Google Chrome uh data and it cannot be downloaded and they thought that is a b and they a lot of customer support come from that until we identify the problem is that the five size to B for those kind of workflow and they need to delete their their their um their stories to to continue download the F yeah okay yeah that's that's the fun funny story um okay we uh have uh one question

from Cole from the live chat I would love to see some automations with myel and make uh while we are not able to show uh an automation with my and make live today uh due to the time limit uh you I uh recommend you checking out our YouTube channel we do have a video where we dive into how to connect uh myal with make uh and in that video I show you step by-step guide on how to uh connect uh my power to make to do things like um to connect to Google sheet and Linkedin

etc for uh Auto publishing of social media content so uh I highly recommend you check that video out for some inspiration so for the next question we have How can I uh get the support if I become a m palal customer so uh let me uh invite Mr TW for this uh so uh uh there are um three uh layer of support that you can uh you can be supported when you uh when you become a my power customer uh the the most um actually Four uh I I I I I calculate wrong the first

layer will be uh you can check out our community of user and our documentation uh in the case uh you you would like to take it yourself so in the community you can get the support from the Community member uh and in the documentation we outline pretty details about every single setting and configuration for each feature on my how in our documentation uh about asan workflow noise source and uh you will be able to find pretty detailed uh solution for the um technical issue when you use my power uh the second uh layer of support

will be uh our customer support chatboard uh is actually train on uh uh our uh every single p on my website and our documentation and being F tun to give uh the most uh possible accurate possible um answer to the question for you and the third layer will be our email support so in the case that uh you you don't find a a um useful enough solution uh from the chatboard or you cannot find directly or not obvious in the documentation you can email us at support at uh my heart I to to get a

support uh there so we go typically uh respond uh to Z support email in with six hour and the the the final layer is that uh only available for uh supplier in team and business plan is that we will have a dedicated uh WhatsApp uh or any kind uh uh Comm uh channel that you prefer but most people choose what WhatsApp uh group uh with uh one of uh our team member here so we can uh answer the question uh uh uh we you have dedicated uh member in our team in the channel and try

to answer the question uh on time uh and as soon as possible uh when the message AR right in WhatsApp and so I guess that's it for our um session so um thank you so much uh Miss Sylvia and Mr Twan for sharing about uh my pal uh journey and uh capabilities and also thank you so much you guys for all your very dedicated questions so before you go we have a quick reminder is that our Black Black Friday offer is still on live so we have a 60% off of all yearly plans with the

code of Black Friday 24 it is only valid until December 8th so remember to uh check out it quick so you can get this offer and um finally don't forget to uh stay in touch uh you can subscribe to our YouTube channel for more tutorials and join our Facebook group to connect with uh other uh users as uh don have just mentioned we have a lot of support between myout users and also from our team team as well so we believe that you can get lots of uh information and help and guidance uh from the

community uh so up until next time we wish you have a productive time automating business process with mpow thank you and see you thank you everyone see you soon thank you everyone have a great day and enjoy the holiday