Guest: Divanny Lamas (LinkedIn)
Company: Transposit (Twitter)
Show: 2024 Prediction Series

Transposit uses AI to help operations teams better handle incidents and alerts. Divanny Lamas, CEO of Transposit shares her predictions for 2024.

While everyone is currently using mega AI models like OpenAI right now, it is anticipated that smaller, niche-focused AI models will emerge in 2024. These focused applications and domain-specific models will not only enable people to get more cost-effectiveness out of the models but will increase the accuracy of the predictions which will be beneficial for people rolling out AI into production workloads.

Secondly, AI will move beyond the chat interface. These tools are expected to go mainstream into the business workflow although people may not even be aware that the tools they are using are leveraging AI behind the scenes. Nonetheless, we can expect to see AI fitting into the applications we use on a day-to-day basis, connecting to them seamlessly.

Although RAGstack has been a hot topic in AI, the examples of it are basic and primarily used for keyword searches. MLMs are the best tool for ETL and we will see more sophisticated pipelines for processing data, adding application context, and doing classification and categorization in domain-specific areas before ingestion. We will start to see AI being used as a cognitive engine to help correlate across different data sources.

Key challenges that Transposit’s customers are experiencing include working out how AI is going to fit into their workflows. It is difficult for buyers to differentiate between tools that are all promoting AI and to work out which will drive a business advantage for their organization. Although organizations want to leverage all their data using AI, the data itself is still very unstructured and has not necessarily been cleaned. Organizations are looking at ways to prepare their data so that they can take advantage of the tooling from the AI ecosystem.

Transposit will be focusing on unstructured data in the coming year. This human-generated data includes things like Slack conversations, transcripts from video calls and interviews, and post-mortems from incidents. Transposit is working with customers to help them take advantage of the underutilized data in the context of on-call alert and incident management. Unlocking this data and making it usable will help with solving issues and reducing their reliance on institutional knowledge.

This summary was written by Emily Nicholls.