Issue 3: GenAI for Enterprises| SmartApps
Sharmilli Ghosh
Building Products, Partnerships and businesses | AI, Greentech, Sustainability, Social Impact | Investor | Board Member | Entrepreneur | Freelancer
Generative AI for enterprises
Guest Writer: Rony Chatterjee, PhD
Shifting gears from saving the planet, to helping enterprises and the global economy in this edition.
According to Goldman Sachs Research, breakthroughs in generative artificial intelligence have the potential to bring about sweeping changes to the global economy. Advances in natural language processing will have dramatic impact on businesses and society. It is estimated that they could drive a 7% (or almost $7 trillion) increase in global GDP and lift productivity growth by 1.5 percentage points over a 10-year period.
Lack of unlocking insights in a timely manner inhibits enterprises to make faster decisions and results in loss of business opportunities. Enterprise Data comes from diverse sources, making integration a significant challenge. At first glance, integrating data may seem straightforward to create diagrams that show relationships between different entities. However, the challenge lies in the fact that data can have multiple sources and the data needs to be organized and standardized to facilitate factual and useful information and knowledge for humans and machines. Hence rises the complexity of maintaining and managing data models. Enterprise customers today use diverse data modeling techniques however translating human requirements into a data model presents a significant challenge. This aspect of data modeling is especially difficult because it requires effectively communicating with people to understand their requirements and then translating those requirements into a flexible data model that can evolve as needs change.
?
Spotlight: SmartApps
?In this newsletter I capture my interactions with Dr Rony Chatterjee. I have fond memories of many white board sessions in a large 16-20 seater conference room in Microsoft Corp, way back in 2017, walking through enterprise data challenges and the importance of clean data, the tools for transformations and preparing data for downstream analytical applications. Rony is a natural teacher – every scenario is explained with examples and every question is answered with depth and context.
Rony is the co-founder and Chief Product Office of SmartApps (https://www.smart-apps.ai/). After a stint at Microsoft and Amazon, Rony dove into the fascinating world of startups, with two of his ex-colleagues (Alekh Jindal?and?Shi Qiao) from Microsoft. ?
SmartApps is providing Generative AI solution enabling enterprises to get faster insights from data in less than 30 mins. The SmartApps team has invented?“Large Data Models”?allowing customers to create data models on top of the data stored in enterprise warehouses like Snowflake, BigQuery, Microsoft Fabric etc., traditional on-premise databases like Microsoft SQL Server, PostgreSQL, MySQL etc., flat files stored in Blob Stores like S3, Azure Blob Storage etc. and allows customers to connect to applications like ServiceNow, Salesforce etc. covering variety of different sources where enterprises store the data.?Many companies rely on data consultants to help them develop models that work well in the present while providing flexibility for future needs. SmartApps GenAI solution works as experienced data modeling consultants and can handle the most challenging data modeling aspects.
The Large Data Model (LDM) in combination with any of the Large Language Models (LLMs) provide enterprises with the ability to interactively operate over their enterprise data using a chat like natural language interface and enterprises no longer need to worry about complex data engineering. There is no data movement which happens within the SmartApps platform, customers data always remain secure within their environment. Drawing analogy to Psychology, Rony Chatterjee, Chief Product Officer of SmartApps says –?“You can think of LDM to be the left brain associated with logic, analytical thinking, and language processing, while the LLM acts as the right brain and is linked with creativity, intuition, and holistic thinking. However, this is an oversimplification, and just like in the physical world you need both brain hemispheres cooperate to accomplish most activities, in the data world you would need both LDM and LLM to work in harmony to get insights from the data.”
One of the issues enterprises face today, is hallucinating responses from the LLMs. For LLMs to be updated with the latest data, which is stored in enterprise data stores, Retrieval-Augmented Generation (RAG) frameworks are applied.?RAG pipelines are a tool for leveraging LLMs in production.?SmartApps builds and deploys retrieval augmented generative pipelines that keep the data models always up to date making sure that the responses from the LLMs are hallucination free. RAGs provide the up-to-date responses enabling the LLMs to access real time data and providing factual correct responses. SmartApps platform includes GPT-4 Turbo and GPT-4 models and allows enterprises to even bring their own LLM. In combination with the LDM it improves the capabilities of a Large Language Model.
Today, enterprise customers do not have a way of interactively asking questions about their data in natural language and instantaneously getting responses. They write SQL queries and while using Generative AI using features like Co-Pilot have provided some flexibility to writing SQL, we observe that Copilot can generate the right queries if they are basic and simple. When the prompt for a more complex question is given, Copilot generates a hallucinating query and even starts to reference columns that don’t exist on the database providing customers with incorrect responses. SmartApps AI solution solves this pain and allows enterprises to query the data using a natural language interface. Enterprises using SmartApps can customize how to interact with their data using LLMs. SmartApps will use the latest retrieval techniques to provide the LLMs with the relevant context and provides factual up-to-date responses to customers improving the prediction quality and relevance of their search experience.
领英推荐
About SmartApps:
SmartApps provide a flexible pricing options (pay-as-you-go and fixed) for enterprises. We cannot wait to see how you talk with your data!
Quick study tips:
Next Up: