Anduke kaka business lo digithey better manam

10 hours ago, Joker_007 said:

Can some one help my Friend here.. How AI is used in day to day activities in typical software company 

AI is being loosely used everywhere. A lot of people are using it synonymously for automation. 

But one of the use cases down the road ->There will be chat GPT kind of agents within organizations that can connect to the various sources with in the organization and can get the data you are looking for without needing analytics or integrations with plain simple English (kind of like ChatGPT now). Help them with insights that need not be written by people. 

For e.g. You are an executing in a company and you need to find out some data. Like what are my margins for certain product with in a particular region. Lets say if an existing report is not there, he needs to rely on IT to create it. It takes at least 2-3 weeks to be productionalized. With LLMs and agents, its now just a simple question. 

On 10/8/2024 at 4:34 PM, usetime63 said:

https://www.windowscentral.com/software-apps/generative-ai-wont-render-coding-dead-in-the-water-but-80-percent-of-software-engineers-must-upskill-by-2027-to-keep-up-with-the-revolution-building-ai-empowered-software-will-demand-a-new-breed-of-a-software-professional-the-ai-engineer

I upskilled and got a decent job.

But had to compete with Interns who worked 12 hours a day and they already knew those technologies from college.

So I Downskilled myself and got a better offer and my competition was 50+ uncles. I am their best coder now 🤣

11 hours ago, Hitman said:

Please brief what does your "Software company" do? is it a product development company? or solution provider? 

 

The company provides insurance to business. Typical products are Personal Auto, Commercial Auto, Commercial Package (Property and GL), Umbrella Policies, General Liability 

3 hours ago, Konebhar6 said:

AI is being loosely used everywhere. A lot of people are using it synonymously for automation. 

But one of the use cases down the road ->There will be chat GPT kind of agents within organizations that can connect to the various sources with in the organization and can get the data you are looking for without needing analytics or integrations with plain simple English (kind of like ChatGPT now). Help them with insights that need not be written by people. 

For e.g. You are an executing in a company and you need to find out some data. Like what are my margins for certain product with in a particular region. Lets say if an existing report is not there, he needs to rely on IT to create it. It takes at least 2-3 weeks to be productionalized. With LLMs and agents, its now just a simple question. 

For our discussion lets take an example of an Insurance Company which has the below systems.

1) A Policy Admin System developed based on Java/J2EE technology stack.

2) A Rating System which interacts with other Enterprise Applications like Claims, Billing and Sales 

3) Data Lake System which is used for generating reports.

Any idea how this AI gets integrated to the above Enterprise applications.. Does it connects straight away with the backend systems or Databases to fetch data..? Does a developer is needed to streamline the data flow between the AI and the application data. 

sure

57 minutes ago, Joker_007 said:

For our discussion lets take an example of an Insurance Company which has the below systems.

1) A Policy Admin System developed based on Java/J2EE technology stack.

2) A Rating System which interacts with other Enterprise Applications like Claims, Billing and Sales 

3) Data Lake System which is used for generating reports.

Any idea how this AI gets integrated to the above Enterprise applications.. Does it connects straight away with the backend systems or Databases to fetch data..? Does a developer is needed to streamline the data flow between the AI and the application data. 

The key for all of this to work is establishing data lineage via a Catalogue. Define the business terminology for technical fields in different systems. Data lineage tools can establish the flow of data from one system to another. They are also working on them to decode the Integration or the code written for data pipelines to establish or understand business rules.

once the catalogue is established, its easy to configure the LLMs, Agents and Bots to convert the user language to Technical fields in different systems, create queries, use the connections to go to those systems and fetch data. 

It will take 2-5 yrs for these tools to come.

In your example .. Lets say you are using SQL Server. The policy details are in SQL Server and the technology is Java. Claims, billing and Sales are in lets say an SaaS Billing system. Lets say the data is stored in a different DB. 

You could potentially have a data lake where you pull all the data from these systems into it. Or leave them as is and have the lineage (flow of data) defined by a catalogue tool. Pulling the data is easy as described above.

1 hour ago, Joker_007 said:

For our discussion lets take an example of an Insurance Company which has the below systems.

1) A Policy Admin System developed based on Java/J2EE technology stack.

2) A Rating System which interacts with other Enterprise Applications like Claims, Billing and Sales 

3) Data Lake System which is used for generating reports.

Any idea how this AI gets integrated to the above Enterprise applications.. Does it connects straight away with the backend systems or Databases to fetch data..? Does a developer is needed to streamline the data flow between the AI and the application data. 

The key for all this to work is having quality data. That's the key. Else hallucinations will happen. 

On 10/8/2024 at 10:02 PM, Thokkalee said:

Nenu Intaku mundu work chesina company lo oka application user names and passwords plain text lo log chesevallu 😁😁😁

ilanti companies and legacy applications unnanni rojulu we are safe… 😃

One of my clients used to download excel sheets and update with pen which will be shared with db team. Cloud laga gen ai koda oka mandatory skill avtundi.. everything still same to same

26 minutes ago, Konebhar6 said:

The key for all this to work is having quality data. That's the key. Else hallucinations will happen. 

Data labelling ee kada they need to have 1000s of them to work 

27 minutes ago, kevinUsa said:

Data labelling ee kada they need to have 1000s of them to work 

That’s different