Name: GoodGist
Uploaded: 2024-12-06T12:07:38.709Z
Duration: 2 min 33 s
Description: An AI-driven platform that enables businesses to create custom AI agents for automating complex, time-consuming workflows without coding.

goodgist

An AI-driven platform that enables businesses to create custom AI agents for automating complex, time-consuming workflows without coding.

GoodGist

Specific demo just want to give a brief on the platform itself. So how this works is that this is a multi-tenant system which uh we have hosted in Azure. Uh this can be hosted in a specific uh private Cloud as well. Uh to provide when someone has a their own uh private Network and want to be very secured. 

So here for different teams we call something as a team space. A team space gets created and um as so here where you can create different kind of workflows by creating a providing a new agentic workflow and providing your objective. Now once this objective is provided based on whatever the objective is it there is a agentic backbone which triggers and it creates identifies how many agents needs to be created and what all their task would be and what all uh configuration it would require. 

And based on that it gives you a series of agents that will require or work in tandem to perform a particular objective. So here um so this two of the uh work flows I've tried to create would want to show them. So one is this particular workflow uh if you see and

This is how the instructions it would take is that just uh giving a very simple English uh instructions that create an agent that reads Outlook and read the most recent email with subject a service requests. 

Read and analyze the service request email will extract shipping details information and source and destination based on that. 

It'll identify what all things are required what all documents need to be provided and then reply it back. 

So this is the first U uh kind of workflow that generally typically when a customer provides or asks for queries this this can be handled through that. 

So for that it it has created uh kind of four different agents. 

So one kind of it does a email reading shipping detail extractor. 

Um additional information it does research on that. 

So now this research can we can provide saying that these documentation what is required these are very much within our knowledge repository go and use there or if some something is not around then you can also say that these are specific places is where you can look at maybe go to the. 

Share the AI Agent Store with friends.

Web or maybe go to this particular domains and and get those information from there and then collate it and um kind of reply back. So this I have just before the time purpose I've run these things and uh so this is the email that has has been provided. 

So I've created it uh for a dummy demo purpose. So this is the email that has come in uh it has read from the Outlook. I have a shipment to ship from Bangalore to San Francisco California and this is a cosmetic items of 500 units. So please let me know what to do. 

So it has read and extracted these information: the source, destination, item, quantity and then it triggers the next agent which basically identify the shipping details key informations and what is the weight. All these are required parameters so it identifies that. 

And these are the because it's a US shipment so what all informations are required it does uh find out from additional information resource what all things would be required. So these are the transportation methods it will ask back saying that do you have any kind kind of. 

Share the AI Agent Store with friends.

Preference of that necessary documents that commercial invoice would be required. Certification of origin all these things. These documents would be required. Uh regulations and standards these are different. So this is basically based on the request that has come. It has done a research on that and it has created that. And then it has done a emailed formation. So basically this is the email it has created. These are the necessary documents. Uh these are the different regulations. 

So it kind of send this. Uh email so basically which this is one. This is the email which I have got here. So now this particular thing this can be very easily changed tweaked. For example I I I can I can go here. So this just to show you how it happens. So this particular agent these are the instructions it automatically gets created. And uh we can go and tweak these instructions as well. 

So it basically looks at the connect to Outlook account. Read the most recent email with this particular request. So you can actually edit those add those. So basically this is. 

Share the AI Agent Store with friends.

First attempt done by the agentic uh Builder uh you can always uh improvise that change that add that uh to make it work very custom. So and also it identifies that there is a Outlook mention that's why this is a connector that needs to be configured. So you can configure with this is I've configured with my uh Outlook. There are lot of other connectors that um are already available and which can be used. 

So similarly it uh does uh you know if there is a web involved then it it does that. So so now next uh is that so once these uh information is sent then typically customers would be providing all the documents and it provides like a either a zip file with the documents further. And once that comes in uh it triggers this workflow where it reads that then information it extracts and then it tries to uh verify the documents if there is a mismatch or anything missing from there. 

So if you look at this is the final output what it is showing uh it has uh tried to to compare the incoming documents uh with the templates and what is required and if there is.

Any descrip it has found so it has kind of tried to identify what all information has there. Uh what is the address destination address? What is the, you know, what is the number of packages? What is about it? What is the weight? Um all those things. So it has looked at all the documents and it could identify some of the discrepancies. 

Like for example weight discrepancy. The document lists both gross and net weight to be 50 mg. So there's, uh, that seems to be a, uh, maybe some issue with that, uh, for a shipment of 100 units. So it has, it is able to recognize that some of the dates looks like the dates are missing and it is able to identify that. 

So basically if you really look at it, what it's trying to do is if there are, there's been a lot of, um, lot of effort goes in making sure that all these documentations are proper appropriate based on the shipment. This is one thing that, uh, which, uh, this particular workflow is able to identify and make sure that it is accurate. 

Now the same thing if the destination changes maybe if it becomes, uh, to

Somewhere in Europe so the requirements would be different. It will be able to identify that um create a plan accordingly. Similarly uh these are the sample workflows that I've tried to create. But the whole point I'm trying to mention here is that you can create any of the objectives. 

So you can classify uh certain information. You can uh then kind of uh interact with a freight management or maybe a transport system. Uh kind of uh connect to their API through their API and book a particular shipment and then send it back to the uh sender who for a particular calendar invite what time they will be picking it up. 

So all these automation can at the same time there's another aspect what is possible is that moment we connect with realtime data. And if there is a um iot based information if there is a real-time data coming in based on that it we can provide um you know there are workflows can be created through which it can provide the health of the systems where exactly it is in um if there is a particular shipment is kind of stuck somewhere.

Something that needs a immediate attention or there could be a probable scenario where there should be a route change required. So all these can be built here. So that's uh kind of pretty much which I wanted to show. 

And some of the other things that uh wanted to show here is that if as we are seeing here you can also provide feedback or saying that uh this particular uh information extraction is not satisfactory. Then what is it so you can actually mention that. So this goes back into the training so that it can learn back and. 

Share the AI Agent Store with friends.

Note, that there may be slight differences in text and video. Watch video for the original content.

GoodGist Video