Fluid AI’s Agentic AI Platform: Launch & Live Workflow Creation

[Music] he he [Music] [Music] hi everyone welcome to another webinar by fluid and I think by far the most exciting one uh that will be conducting in the last couple of months because we're launching the new agent Air Platform right yeah yeah absolutely and what's exciting about all of this is that you know there's so much conversation on agent in terms of what it will make possible and what's going to happen with it y uh but can you walk us through what the team has sort of done at fluid to sort of really get this out there and get this big yeah I think uh you know largely what we seeing is uh this whole shift from you know I think rag was very popular 2025 2023 and now we're seeing this big shift towards agentic models and agentic approaches to a lot of these things so what the the reason the shift has come is because the models have become a lot more intelligent okay so suddenly they're able to do a lot of things that they they weren't able to do earlier okay and uh can you just tell us what rag is just just to refresh us yeah yeah so rag is retriever augmented generation which basically means uh it's an AI system which can you know it's it's like reading and like you know the llm is trained so it has certain knowledge okay and uh then you're giving it you know access to additional information in real time it's like an open book exam right and then you know you're asking a question they bringing in real time information and it's answering using that so that way like even if the llm doesn't have access to it wasn't during training didn't have access ACC to something and now you give it it it kind of uh can understand organizational knowledge and information essentially got it and then agentic takes us a step further right in terms of now it's not information it's action correct corre so you're now dealing in an action oriented World rather than just information in and out yeah absolutely absolutely so where two things right one is if you want to agentic does two things right the word agentic comes from the word agency Okay so means that these models have uh their own um intelligence if you might put it right so their own agency of decision making so they're not bound by just saying okay do this one task like in rag it's like here's a question here's a bunch of sources answer the question okay uh with agents it's like okay we need to solve this problem and you have access to these tools or functionalities right and uh you are like you figur it out okay it's very intelligently deciding what to do next almost on the flight like llm is using its intelligence to decide what to do next which is kind of a very neat uh way of of leveraging language models right um and um and then when you start trying to do that the problem that used to come is that you know one agent can only do so much correct yeah because it's it's limited by tools correct yeah so like you could give an you could give an agent access to like one agent access to 10 tools 20 tools 100 tools right um and then you give it it prompt like it's instructions and everything right but it starts to with more complex and you'll see a few examples today like with more complex examples it starts to get tricky okay like uh how does the llm kind of how do I put it like like okay like if it's having to decide between 20 tools right let's saying a bank support system right hasde 20 tools and uh each tool has a different output needs to understand whe the user needs to be authenticated and those Ops have different kinds of Journeys suddenly it gets very hard to kind of program like not just prompt that and into one agent okay and that's where you start creating a series of Agents with different specializations got and each agent has access to its own tools so it's kind of like an office I got it with the specialist in the office exactly right so you kind of get different specialists in the office at different points in time uh so if you're like running a project you'll have like oh this guy's great at writing code and then this is person's great at marketing this person's great at research okay and then you put them together in a team right uh versus trying to say oh let me have this one person that's great at everything like a super generalist so you have a team of Specialists corre team doctors correct corre yeah yeah absolutely absolutely so uh let's talk a little bit about fluid's approach to this right this has been a long time coming for us in terms of uh you know I think some of the link post have compared it to the 911s launched by Porsche or the bin by heres yeah uh why is this such a gamechanging agent care platform before we sort of really go into yeah absolutely and if we can just switch uh our positions yeah just swap it over right right to left because I think otherwise we're looking like for the viewers yeah now we're looking the right place yeah um so uh I think when you know like our a agents are going to be like the new apps right like you have language models that are like phones correct right like oh the iPhone came right and revolutionized the way we work it's like oh this totally different way of interacting with this device and a new type of device okay language models were that right right they totally new way of interacting with computers systems cool uh and they have this intelligence to them and agents are almost like apps on top of that okay so you could build a like an app which is like a sales Agent app or a call center Support app all right and and uh and and that's the way we've kind of looked at it like so I think the few key items that everyone's looking at when they're building agents is one they're extremely programming oriented like I feel like it's like all the programmers can create these they're not available for regular folks like a no code low code yeah exactly like a business user can't go like oh I this is a way I work and can I just put it down and have a series of Agents do this for me okay so I think that was one big gap right no matter like you see a crew or a lra or whatever the open- source uh what do you call it framework Frameworks out there they're all limited to this approach of saying that okay you have to write code okay um which is a very typing even for a developer because takes a lot of time to then you learn the framework really get the nuts and Bulls going right correct yeah you have like a thought flow in your head so how you get that running fast so number one that's that's something we try to solve for okay uh the second Big Challenge when you want to use agents is how do you use them in like you know like agents these days have been great for like these linear flows okay so no back and forth yeah exactly like customers having back and forths with the agent Stacks uh like even if you're try code that in one of the agentic Frameworks is very complicated okay like how do you maintain State and then how does it go back and forth and that becomes important customer interaction because the customer Say Hey I want to change something or you want some he's not going to give you everything to start maybe a user and the company can but a customer exactly exactly so customers will want a lot of like you know back and forth switching yeah uh and managing going on so that's where the ability for these agents to be a lot more intelligent okay uh to handle that state is very important so that's that's one key key thing that's the second thing thing we solve for and third is uh I think a biggest problem out there today is there are so many agentic Frameworks okay it's very hard to choose should I go with this should I go with crew should I go with llama index and then tomorrow you're going to be like okay let's say you bet on langra and then six three months to six months later crew AI is the hottest thing right now crew is the hottest thing things so then you know lra comes ahead and this this happened once before so if all of you remember the machine learning days right there used to be like when Google launched TZ of flow okay and I was like oh machine learning now open source at another level and you know everyone's trying to create these neural networks using tza flow and Facebook meta at that time I launched something called byor okay which didn't pick up and everyone was T of flow T of flow T of flow right and uh like a couple of years later I don't know what happened in the market but suddenly py to just took everything and it was it py yeah like now py is the defao people don't even I mean people still use Tor for it's a great library but fch like all the latest stuff that's happening with Transformers gen a lot of it is p toch okay right like if you pick a hugging face model there's a high probability that it's on P toch and llama series is all running on top of pyos because of you know being the same parent had met up but essentially py has become like a lot of transform models are all written in py you go to any open source library on GitHub uh you'll probably see py under the cor so so same thing with agents right it could be that one is picking up steam right now but you could get locked in so we kind of built interoperability of Frameworks underline okay uh and that's the third pillar that we we thought would be important that you should even think from a general perspective even if you're not using flu AI just you thinking about hentic that's that's I think a key one cool so uh I think just before you know obviously very excited to go into nuts and boats but I thought we'll just spend maybe a minute or two just walking folks through a little bit about what agentic is and what we're thinking and then get right to it so this is a little bit about us obviously you guys know us been working in this space for the longest time uh you know we work we launched this beautiful book it's still aable on Amazon called Bridging the AI Gap written completely by AI but this is before it so I think version two with agentic maybe yeah even better even better yeah and in terms of what we think are going to be some really cool agents uh we think data analysis where agents is going to be really powerful where agents can you know really go into company data company systems pull push information customer facing agents I think we touched upon that but both marketing as well as customer support I think these are definitely going to be agend flows and that's why I think fluid platform really shines through um another key area of help desks HR it all your regular help desks out there and finally comes this sales and marketing you know you want to create a newsletter a blog post uh and in the end we are surprised even any influencers possible using agent absolutely um so I think we've already talked about what agentes are but these are essentially multimodal right if you give them a goal and they'll sort of just go towards that goal and they'll adapt to the environment using the tools they have yeah absolutely right right and and how they sort of transform business models is it's their modular right so like you can literally have an agent let's say you're submitting a email or a newsletter and you sort of have a planner agent which sort of plans out what kind of content is interesting then it says okay now we need to research what's the latest in that field yeah then it goes down to an actual newsletter agent and then it drafts an email out of it so four steps and four Specialists rather than one system trying to do everything together correct correct all right and again just the llms are independent right you can use any llm out there right you can in fact each agent could be on a different LM true yes so the planner agent could be on llama the writer agent could be in Gemini we could have anthropic opener so it's almost like you could have a team of llms support you through this yeah absolutely so let's let's go into it let's let's see some you know we are all excited to see the platform the unveiling of it uh and actually you know sort of really excited to see what what what what comes out absolutely so let me go ahead and share my screen [Music] yeah all right so you know let's let's start from the the the the very barebone Basics right um so just just walk what what are you showing us right yeah yeah so what we're doing is you can create you know any kind of agent workflow okay and so you can create basically each you can treat each agent and yeah this a nice stter like a simpler one so this is like a very simple agent so it's going to uh basically you know it has a journey to it which is like a stock market research agent oh and we said okay you know let's have uh you know an investment research planning person a research analyst and then a final research advisor okay and let's kind of put them together to do something interesting now like the you just click on an agent you can see that how we Define an agent it's all it's all very simple like that's that's what I wanted to drive home that is can you walk us through the screen so I see system prompt yes what does that mean yeah so so for you give the name of the agent and you just describe what the agent does right so like this is the planning agent right so his job is to plan the system is the main instruction to the agent okay right like what is this agent going to do okay um and so in this case it said okay you are an expert on investment analysis and you'll break down the incoming question into a list of simple independent questions that a research analyst can research okay so the users obviously going to just give something very complex but you this agent's job is to break it into its pieces correct correct and um and then so it's basically saying thing through an i TR come up with the final list of questions and output those questions okay and give me about two to five research questions okay right and it said hey the main question is and and then this is the input by the user right whatever the user gives it essentially okay so that's how you define the agent uh for techies this is how you decide which agentic platform you're running on okay so I'm seeing Lang chain I'm seeing Lama index what do these mean these are basically open source platforms okay um agent yeah platforms but they're only code accessible okay right so the the AI engine underlying so what you like I talked about how you don't want to get into vendor lock in right so this is how you essentially take your agents and say okay they can run within a particular agent Library okay right um and the library is being llama index like correct so let's say it's running on llama index and you're like okay llama index have come outdated I just do this okay and boom I've Rewritten all my entire code base to run on Lang chains langra platform got like one one one drop down and so this is super exportable exactly so no lock in no no getting stuck okay here's where you select your model so which llm model all right um and uh here's where you do a simple prompt like there are different types of Agents okay uh uh we'll get into this uh but in a nutshell like a simple prompt agent is just by its very nature simple like like a you ask a question you get an answer okay you have react agents which are reasoning agents basically it follows a Chain of Thought approach okay so it's like and we'll show you the chain of thought that it does okay but it it's actually quite beautiful right the agent is it's like thinking through its steps so it's like okay I need to do this um what you know what action can I take and then what's the observation of that Action Now what's the next thought I can do and it kind of loops loops in that so around agent kind of orchestrates or decides okay out of all the agents I have who do I use for what situation and then function calling agents if you have like a rest API or something you want to call you can just use these agents too okay and these are all the tools that the agents have access to so tools are basically functionalities okay right so I'll give you an example so you created a knowledge base all right and you want the agent a knowledge base of like all your documents all your documents PDFs PowerPoints yeah Okay Internet research right theet research is basically agent capability to go into the internet get like get information from get information from the internet okay pin authentication account balance oh okay so so some of them are informational tools like get information from our knowledge based on and some are like pin authentication so I'm guessing this is in a banking scenario authentical pin so it's connected to the pin system exactly then you know onetime password generation validate that so who creates these tools up like are these the apis yes so some a lot of them are apis some of them come out of box with fluidi okay and uh yeah those are the typically the two formats that you can so it's your apis plus the preformat like the knowledge based the internet one those are's own tools correct and then you can add let's say you're a bank or your Telco you can say okay these are my system apis and each of these could just be to correct right so that that's our first agent okay then we have a researcher agent so the researcher agents actually been told that you know hey below the questions your uh below the questions that have come from the that you need to research on okay this is the output of the previous agent the planner agent essentially got it so the planner agent is s of broken it down from one question to five six questions and now we have the step okay and then it's saying okay so you know uh yeah so this is the if if you saw this was the uh yeah this is the investment planning agent so then this is this is output comes here okay and then and then this is this has access to the stock analysis tool and it's a Chain of Thought thinker so it will hit the llm multiple times if needed got it and yeah and then it kind of goes ahead and and does the answer so it's saying use the analysis and then come up with a crisp summary of every every research that you've done got it and then there's a final agent that puts takes the output of This research analyst right right along with the initial question right and says now just give the summarization of the entire answer that you've thought through essentially all right so so that's where you moved from a question you've you have divided the question to Pieces that's what the planner has done then the research anal got that information and the final research devices just put it together in a presentable format right so it's almost like a three person team yeah three agents exactly sorry um so let's run this and let's see what it looks like so um so let's say so right now you're showing it from the back end right this is but like just to show us the steps but otherwise from a user perspective it could just be question answer correct correct okay but right now we sort of drill through yeah so we can see what it's looking like yeah so if I ask it like uh what is the margin uh difference between ultr Tech cement and ucha cement right so now the cool guys you'll see like the different agents start to fire off okay so if you didn't notice the investment planning agent has essentially done its uh you know its first pass which said okay I'm going to split this into two questions what what is the ultratech cement margin current operating margin and the ambuja cement current operating margin correct um and what you'll notice here is this is the second agent which is a research analyst agent okay um uh and it's thought through its process so it's saying okay I need to I need to go to the annual report I can see that yeah yeah exactly and it said okay the first question I need to ask is the ultratech cement current operating margin okay and it's used a stock Insight reference um like a tool right and um it's it started to get this data so it's it's essentially starting to get different annual reports and different kinds of data sets U that it's getting and once it puts all of this together uh it'll it'll be able to generate your answer with the margins and the differences and everything okay so that's how you know you can have like just a very simple agent that's that's kind of doing it so now if we ask a very different question like what is the market share of uh let's say acsc bank right okay um totally different question now it's realized okay uh it's come up with these four questions right so market market capitalization okay and then total number of deposits and then total number of deposits held you know in the banking sector so it's looking at market share in two ways it's saying either it could be as market capitalization or it could be deposits correct and it's trying to say okay those are the two ways I sort of exact so now it's running its various thoughts and processes okay and getting getting different kind of answers from W actually going Bank by bank and saying okay what are the different banks deposits Tot so it's kind of done that analysis and if you see um you know it said total deposits is at this insane number as of 2024 these are the different banks yeah uh these are total deposits and then essentially uh you know come up with with what that number looks like awesome and then it it sort of really you know pulled a lot of data and analyzed it very quickly correct and and still you know coming out with that one you know precise answer so that's more of a you know a research kind of approach but one of the key things we are very interested with agentic is also customer interaction front of the house yes so let's talk about that so let's go to a totally different place so within customer interaction right is both support and sales uh but it's largely the same right you're interacting with Key Systems uh the customer has to authenticate that's a key part of it yeah uh because he's onboarding new product and a large part of those conversations are going to happen on voice right because that's the current Paradigm so it's difficult to force the customer to say Hey you change where you interact with so you want to have voice going you want to have maybe text going as well live chat you want to have email and the functionality just balloons right from where the stock agent was looking at just maybe information yeah now you're sort of saying oh account balance but I need to validate before giving you the balance uh I need to maybe take an action like blocking a card and then somewhere they just throw in a gener query like what time is the bank branch open right so it's it's so much more complex absolutely yeah right so here here's that you know it's this is that example of how you start with okay the the router sorry the the router agent kind of getting the initial um you know figuring out where the customer authenticated these are all the available agents figuring out where the agents need to go um and then sending them you know there's a card balance agent there's an account balance agent there's an authenticated agent and they're also deciding when to do these transfers right like who transfers it Backes actually this is where I was coming to when you start to think about really complex scenarios of also being able to answer customer Knowledge Questions right but also do transaction and that doesn't need authentication I don't need authentication to know what kind of BR is exactly yeah and so programming this in a bot flow would have taken like days because then like each Journey each question sending the user back out but now you just give it the apis and you're kind of done and it's intelligently handles goof ups and situations right customers on the phone will say oh okay wait no I want to First discuss this now I want to do this they're jumping all over the place which like a traditional system would be really bad at doing so if you look at a card balance agent it's actually quite simple it just giving it a description a set of instructions uh your instructions are just plain English and giving it access to the card balance tool okay and that's about it so so and the card balance tool is typically the API to retrieve a credit card balance corre correct corre so now if I run the workflow and I say I want my credit card balance okay it's hit the router and now it's hit the authentication right that saw the or yes exactly so they lit up so the decided okay to do this uh the user probably not o so I'm going to send to the Au agent okay and then say okay sure I can help you with that the O agent is taken over so now now this is where the power of having multiple agents as opposed to One agent is right okay now o agent has got all these instructions that okay you know what what you need to ask the customer for ETC how do you need to ask it uh the router has its own set of instructions so you know imagine trying to fit all of this and then the card balance has its own set of instructions true uh so imagine when you and then the knowledge based one has its own set right right like okay you know a knowledge based question now imagine trying to put all of this in one agent which some people are trying okay but it's crazy like the amount of context you'll have to send it the L's going to hallucinate like crazy and then security right because authentication happening in the same agent exactly right the ability to fool an agent becomes very easy um security kind of goes for a TOS so so now we let's go so we'll say I'll say my TRN is one two 3 4 sorry yeah what is it one two3 4 yeah 4 220 0 1 2 3 420 uh I think it's a double four as well oh so um I I actually forgot to give it my last four digits but that's cool right it asks for both typically customers forget or they may give it in a separate order and this is where it's very different from the bot because I could actually give it uh I can say both like I can say my T is this my this is this and it'll it'll autoally so that old NLP flow world where you're like no no I just wanted you can give me any order any order anything last four digits are and then I was ask me for my card pin so I one for so again something's going to light up out there I guess is the authentication agent condition Clos so now Au agent says okay please hold on I'll proceed so now this is the cool part right so it um it sends like the router agent has decided to transfer the request to the credit card card balance agent okay um and then the card balance agent uh has gone ahead and what do you call taken that request right and then it's thought okay I need to use this tool to get the customer's card balance okay and I'm going to use the card's last four digits and then the customer TRN okay uh which has got from previously in the and that's what the API needs right exactly I figured out what the API needs and most from the past conversation this has been collected okay um through some smart memory and then this is the API that responds saying okay balance is 6084 uh and ,200 is $1,200 is the balance okay so I question let's the API days tomorrow into last four digits we need something else correct would it just question so literally like you just add in the in the API tool in the card balance tool you add the additional parameter that's needed okay yes exactly and you're done like you don't need to change anything else like the systems will all know that now this is needed and the llms will ask the questions the flows will automatically like handle the situations the agents will be intelligent so all that error handling right someone doesn't provide them made Etc that's pre- taken care of you just need to add the element one place exactly got it and and so now let's say it's done this now let's say I asked a question like and this is so like I'll okay what are the fees on the sapphire credit card okay that's a gener question I OD for that right yeah exactly oh so it goes to the get card fees that's the tool that's correct corre so so then it'll kind of you know move accordingly okay in the situation because this needs a different different layer of off but now if I ask it uh what was my credit card alance again so it's come it you know it still has that memory of okay and I can even say uh um can you tell it to me in words got it so yeah so then saying your your card balance is 1,300 to exactly so it can handle like these like when it becomes humanik okay when people just start chitchatting with it and asking for different situations that's pretty you can handle it quite well and the same thing translates to voice right correct yeah yeah and that's where the very cool it gets because you take the same you take this experience okay and uh and we move it to voice yeah so in a voice element while we're going to show it of a browser yeah it can actually connect to teleph systems through S right correct so it's a a single line of code addition literally to add to an existing teleph system wherein you can take the same agent which I was showing for credit card balances Etc to actually get it activated on a voice kind of platform as well so you literally have the same tool not just on text but on voice as well so it becomes multi modal in the sense that you now have it not just text not just email but you have voice as well and it's brilliant in terms of saying that we have the ability to add this to sip by a single line of code uh rather than having to really get things complicated and you know having a lot of moving Parts going on yeah absolutely so now I'm G come back here so so same agent we're talking to her but but on voice and let's see what happens Hi how are you doing hi I'm good how are you I'm doing good how can I help you I was uh wondering what the my credit card balance is um give me a second to check that for you I can help you with that however I will need to authenticate your account first could you please provide me with your TRN tax registration number yeah my tax regation number is 123 4422 um please stay with me while I check that thank you for providing your TRN unfortunately I also need to confirm that you are authenticated could you please let me know if you have already been authenticated no uh the bank still needs to do just a moment checking that for you since you are not authenticated yet I cannot retrieve your credit card balance at this time I recommend contacting your bank to complete the authentication process if you have any other questions or need further assistance please let me know yeah so that's an example of when it hits the system at the back uh if the kyc is not done for the user he can also decide to reject so this was a scenario where the kyc Au was not complete Cor it could not handle this transaction on phone and it could give you that kind of messaging coming back out corre correct absolutely perfect all right so I think uh you know let's talk about another scenario which is newsletters right I think that's a scenario where you move a lot of information in there's a lot of planning happening and uh also by the way you know you guys are free to sort of post something in the comments if you have a question or a query on the platform you happy to take that and and answer that but yeah let's talk about user like that's complex task I know our marketing team spends a lot of time we're trying to get our first newsletter out and da agent has been a big help so you know if you can show us how that kind of agentic model can FL yeah yeah absolutely let me bring it up um and actually maybe yeah yeah let's uh let's see so in this one uh you know what we do is we'll have a research agent the research agent will you know go to the Internet get data from from different kinds of sources okay uh there'll be a planner agent that will plan what needs to be researched okay then we'll have an agent that kind of gives the the news letter output all right and then we'll have an agent that also drafts an email for that newsletter okay um so you almost have all the four steps from planning it out researching it to writing the newsletter and then writing the email that goes within correct corre correct correct so let's let's WR a newsletter on agent yeah absolutely so um agent take Ai and agent so it's going to hit the planner agent first correct okay and that s okay I need to do as two points so that's of what's divided into yeah so the the understanding of agentic AI and then the analyzing of you know the different agentic AI platforms okay and then now the researcher agent is starting to go this is from the internet yes okay uh and kind of starts it's putting together you know some some Bas baselines from what the the researcher agent is doing then we have the newsletter agent okay and uh that's created this whole newset yeah it's created the newsletter oh it's even added a PS stay tuned for more nice and and then the email draft that come which has a subject and things like that okay nice so it's literally like from topic researching the internet to the newsletter to the email we reached it within 30 seconds corre right and absolutely each of these agents can be modified we can give them tonality we can give them previous articles maybe a knowledge based we don't want it to research from the internet correct from our previous blogs yeah we can do that right absolutely absolutely perfect all right I think that was amazing U and you know this is coming out it's already launched for a couple of our customers and we are very excited to sort of showcase this to you uh do you know set up some time with us we we are conducting demos strategy sessions on how agent can be used uh so happy to take you know any kind of requests on that uh you know you can U show we have a geni community uh wherein you know we have we we are discussing these kind of things oh sorry I forgot one thing uh we have ai Afters as well possible through this technology right which is a influencer which we've launched it's called Aisha Neo so if you're on Instagram uh and she'll be on LinkedIn soon yes as well but yeah so this is another AI agenty platform which is creating Aisha Neo right correct yeah absolutely so she's using the multimodality and and creating it right and right now it's images but now we'll have voice in the next uh next few days and yeah she so I think our team put her at the Nvidia conference a few days ago um and yeah see so it's a really powerful Paradigm wherein you know Brands can actually create these influen ERS and all they need is a team which is managing the agents right the agent which is saying hey put our influencer in this location or make our influencer do this or say that right and the same kind of technology is coming with AI ofar as well right so right now we did voice but imagine you want to add a visual element to that same AAR to that same we can add Anar yeah absolutely absolutely and all of this can be done in a very secure architecture each instance is separate you can even run it on the uh on the host on our client's infrastructure yeah and we are going live with these spb quickly right as fast as 35 days corre which is a record in terms of what can be done with agent yeah and we offer multiple types of architectures we can host everything on the customer Cloud everything can be on fluid cloud and their hybrids as well so uh in terms of all these different scarios you have the options to really go through and see what scario works for you uh this is our AI community on top please do scan the code it's a WhatsApp Community we share a lot of interesting ideas on AI on geni on agentic AI and if you do want to work a demo with us you can always log on on the second QR code yeah H Kaji asked a really interesting question so he asked uh in scenarios depending on the prompt series AG agents are involved and is there a capability where based on promt context other value adding agents are launched to go beyond what the customer has asked for so absolutely um in fact let's try and see if we can show you uh that example of what that would look like so um let's say this is our agen are planning you know that are doing like card balance off Etc as well right so most of these agents typically once they give that value out they move straight to stock now you can easily let's say just add an agentic node that like you know does certain Lookouts of customer interest Journeys Etc so let's say I'm just going to create a test agent uh you know and say okay based on test and say okay based on the customer conversation so far um you know think of interesting things to to add to this conversation etc etc and we'll just save save save this one and and then you know once you have the testing agent you could actually do a double link so you link this agent here as well or you could put it in between stop and this one so we we get get rid of this Edge and it goes here and from here we go to stop now you could even say that only in certain scenarios I uh where the card balance agent um you know doesn't give a balance uh it'll come here and if there is a balance it'll go somewhere else so you could design those multi Journeys and you know breakouts as needed awesome so perfect I think thank you today for joining us and stay tuned for more information from us at fluid uh we're always developing new things and we are also launching our podcasts which are coming up next week so excited uh you you'll get to know about them on our channels thanks again and uh look forward thank you all

2024-12-24 19:55

Other news