20 Aug 2024
Artificial intelligence in general evokes images of humanoid robots performing tasks which are normally done by humans or autonomous cars without a human at the drivers seat. Is there more to it? How does it play out in the enterprise space? CXOs of today are faced with the question of what can be done with AI? How can AI be leveraged to accelerate the business? How can I run X times faster with AI etc. The answer really is to look at various aspects of an enterprise and how to apply AI to move the needle faster.
We will look at the four major facets of the enterprise landscape viz.
The focus of this blog series is to start with Market Intelligence leveraging AI. Market intelligence requires sifting through myriad sources of data, cleansing, perform metric computations, correlate trends and generating inferences out of it. Traditionally Datawarehouses were built feeding from structured enterprise databases which houses the OLTP schema and the feeds were seeded into an OLAP schema with Facts and dimensions. Subsequently we saw the evolution of Big data with very large hadoop clusters. With the evolution of Natural language processing and document processing, vectorization and graph databases and orchestration of all of it with tools like langchain and LLMs for query generation, summarization and inferences, we live in a slightly different world today. Lets look at some of the usecases where we can meaningfully leverage the latest developments in document processing, NLP, Large language models and the astronomical computing power available through hyperscalers.
A Marketing research analyst has to deal with myriad of data sources which have some hidden insights about customer behaviour, evolving market conditions, Pricing dynamics, brand awareness, Campaign effectiveness, customer segmentation using A/B testing, Pricing sensitivity, white space analysis etc. All of these problems involve consolidation and cleansing of data from variety of sources and produce insights leveraging LLMs. Campaign generation for a deeply pesonalized content does require a very scalable ways of generating deeply customized images based on a deep micro segmentation of the customer base. Platforms like Ongil.ai can solve precisely these kind of problems by leveraging LLMs along with langchain orchestration and models for data extraction from Elastic search, MongoDB, vectorDB etc. by building custom models for natural language to specific query languages. More on this usecase in later blogs.
The problem statement here is to deploy and test the actual responses from the user and use that behavioural data to figure out which segment of the user base should be targeted in the next Iteration and which ones should be left out. It will be optimal to leave out the customers who continuously dont show interest in the touch point. At Ongil we built a Multi arm bandwidth algorithm implementation to understand the customer responses and figure out the segment to go after in the next iteration. Coupled with a generative tool for campaign content and images, this is a perfect solution for campaign management at scale. Deep personalization of campaigns in possible now with Generative AI which was not the case before. More on this usecase in future blogs.
Advertising Platforms has to be able to generate creatives which are most relevant and personalized for the user segment. With millions of users, the variety of creatives that are possible is very limited in the human context. However, with AI generated Imagery, it now becomes possible. Also, its possible for AI algorithms to sense the behaviour of a microsegment and control the next set of actions for the user. Advertising is one area where AI is going to generate huge gains. Already companies like Google and Meta are leveraging this in a big way. It also has more room for creativity as content generation , image and video generation at scale will create tremendous opportunities. One critical application of Machine learning is to classify the images and videos and eliminate objectionable content thus avoiding breach of Law of the Land for the advertisers. This alone could potentially save the companies from Legal proceedings. It will also save the brand reputation and make internet a safe place to browse for everyone.
Campaign generation and measurement of campaign effectiveness is one of the key areas of marketing which always requires creativity and huge computing power at scale. With current generation GPU architectures, its possible to develop marketing intelligence unseen so far. Ongil.ai has developed targeted content generation for campaigns as well as Image generation to automatically produce campaign content. Coupled with algorithms for A/B testing,
Social media platforms like Meta has billions of users generating terrabytes of content. To meaningfully process that content and derive insights out of that is a humongous task which is where artificial intelligence coupled with Big data systems comes to the rescue. As we move towards Multi-modal content, there is a huge opportunity for startups specializing in processing variety of content sources and produce insights. Ongil.ai platform is designed precisely for such usecases in mind. The interplay between the trends in social space to the surge in e-commerce sales is a very interesting problem we worked on. For e.g. a Concert in Europe virally spreads through social media channels across geographies and increases e-commerce sales in North america. Businesses who predict this well ahead will be able to capture this opportunity which otherwise will produce disappointed customers who are willing to shop out of momentary trend.
The channels for an enterprise evolved from static web, dynamic web, mobile apps, chat interfaces which are statically guided flows. Current generation is the conversational channel which can interact with the customers at real-time. Such conversations can provide a very deep insight into what the customers are looking for and by generating the right questions, can capture the intent of the user precisely so that the customers go through the business flows without much friction and get what they want in the least period of time.