Introduction
Over the last couple of years, Geminos has gained great traction, helping large industrial customers make better decisions using Causal AI, which allows your software to reason based on cause-and-effect relationships and direct input from your team to make real decisions that optimize operational performance, not just make predictions. We’re now expanding our platform to deliver Causal AI at enterprise scale – and we’re working with IBM to help streamline adoption and bring value to new and existing customers.
What do we mean by “Enterprise Scale”?
To date, our Causeway platform has generally been used by small teams of business analysts, data wranglers and data scientists to model a few decisions at a time. Our customers can now capitalize on Causal AI across their organizations, which is designed to help drive more informed, holistic business decision-making. This is especially important for our customers, as many of their decisions are interconnected.
For example, if an operator reduces the output flow of chemicals through one part of the PVC manufacturing process to extend the life of a critical component (such as the Oxy Reactor), that reduction will impact the next stage of production and could ultimately reduce the overall production of PVC, impacting key customer deliveries. The ability of our platform to show how all these decisions are interrelated, using Causal AI, means that our customers can quickly understand and tackle hundreds of decisions, not just a handful.
Teaming with IBM
To meet this challenge, we’re working with IBM in three technical areas:
- Enterprise-scale data management using IBM watsonx.data and watsonx.governance
- High performance causal model analytics using IBM watsonx.ai
- Combining the best of LLM and Causal AI technology, leveraging IBM Granite LLMs and SLMs and Geminos Causeway CRAG (Causal Retrieval-Augmented Generation).
We’re also excited at the opportunity to leverage IBM’s Co-Sell program to develop a joint Go-To-Market (GTM) strategy that helps bring Geminos’ approach to business decision-making to a broader customer base.
The next few sections will go into each of these areas in a little more detail.
Enterprise-Scale Data Management
Geminos’ Causeway product line uses an advanced repository based on a graph database. This offers high performance and is a great fit for the causal and knowledge graphs being created by our customers.
As our customers start moving to large-scale rollouts, we must add enterprise-class data management and governance, so we plan to embed IBM’s watsonx.data and watsonx.governance, which we expect to provide the capabilities we need, including:
- Authorization and access control
- Model and data versioning
- Management of development and production systems
- Integration with a broad range of data sources such as Azure Data Lake using open platforms such as Apache NiFi and Airbyte
High Performance Causal Model Analytics
Working with the IBM team from their Ecosystem Engineering Build Lab, we’re creating a version of the Causeway platform that can leverage IBM watsonx.ai for causal model analytics. This will enable customers to use watsonx.ai’s Machine Learning (ML) capabilities and integrate their causal models into the broader IBM Watson Studio platform, helping speed their time to value with Causal AI.
The aim of this process is to help maximize the performance of our platform on very large-scale datasets.
Combining the Best of LLM and Causal Technology
Some organizations are hesitant to adopt Large Language Models (LLMs) for business decision-making due to concerns around ‘hallucinations.’ Geminos is working to address this problem by combining LLMs with causal understanding. The core idea is based on Daniel Kahneman’s “Thinking, Fast and Slow”, with LLMs and a causal foundation providing the ‘fast’ thinking while causal modeling—assisted by LLMs and subject matter experts—handles the ‘slow’ thinking.
LLMs represent Kahneman’s System 1 thinking in a business context – they are fast, easy to use, but more prone to biases and errors. In contrast, Causal AI requires more attention, because after all, we need to build and analyze a causal model of the business problem; however, it’s designed to be less prone to biases and more accurate. Therefore, it’s similar to System 2 thinking.
As we work to add this new capability to our product line, we’re once again working with IBM, leveraging the company’s in-depth expertise with LLMs and IBM’s Granite LLM and SLM models to help our solution meet customers’ needs.
Bringing Causal AI to a Wider Market
In addition to embedding various IBM technologies in our platform, we’re also excited to work on a go-to-market (GTM) strategy with IBM that helps bring the power of Causal AI driven decision making to a broader customer base.
Conclusions
IBM is proving to be a great partner for Geminos Software as we transition our platform to being truly enterprise-class and as we continue to introduce technologies that help address our customers’ needs, such as combining LLMs and Causal AI. The opportunity to take the resulting solutions to the broader IBM customer base is also very exciting.