InterSystems held an analyst briefing day in their Windsor offices on 23rd September 2025. The company was founded by Terry Ragon in 1978, and he is still their CEO. The company is privately held, with a single investor and no debt. The company is headquartered in Boston, and in 2024 the company achieved $1.1 billion in revenue, up 10% from the previous year. Their software manages over 80% of all American healthcare records. – -InterSystems is a cornerstone partner of Epic. InterSystems also processes 2 billion real-time financial transactions daily. Their cloud business increased by 36% last year. In the UK specifically, InterSystems have around two hundred customers, with many in the healthcare sector but also in retail, financial services, supply chain and logistics, publishing, public sector and consulting. 60% of all NHS data flows through InterSystems Health Connect, with a quarter of the English population covered by their Shared Care patient record system.
At the core of their platform is their multi-model, multi-workload data platform IRIS, a high-performance data platform with real-time analytics and integration capabilities. On top of this are a range of applications and solutions for specific industries, including ones built by partners. The product has a reputation for very fast performance and affordability. A multi-model database stores data in a specific internal format but can project out that data into different formats such as relational tables, documents, objects, graphs, time series or vector format without duplication. This makes it particularly effective for applications that deal with multiple data formats. IRIS supports vector search natively, which is important for large language models.
The IRIS data platform can ingest almost any data format, including relational data or documents, and can connect to various external data formats without moving that data. IRIS can run on popular public cloud platforms such as AWS, Azure and GCP. It has full support for containerised implementations such as Kubernetes and Docker. Their InterSystems Data Studio allows low-code data fabric application development. Industry-specific wrappers are delivered on top of this e.g. for supply chain and financial services. Data fabrics typically require many different components to deal with the various aspects such as connectivity, data lineage, security, semantics and data integration. This can be quite a daunting prospect for customers contemplating building a data fabric. The InterSystems Data Studio brings these capabilities together. It has a series of core capabilities such as adaptive analytics, plus some industry-specific modules. One customer, Harris Associates, saved $200k after automating processes for reporting using this technology.
The vendor continues to embed AI capabilities within their products. A key theme is to enable AI to have access to enterprise-specific data, often through retrieval augmented generation (RAG). IRIS supports vector embeddings and vector search, effectively the language of LLMs. You can combine vector search with full text search, and there is support for Langchain and templates for GraphRAG, which combines LLMs with knowledge graphs. One customer, Biostrand, built a research platform for genomic structures using this technology. There is native support for the MCP agentic AI protocol. They add IRIS authentication, which helps address some of the known current security and authentication limitations of MCP by adding an extra security layer.
IRIS has a common data plane providing services such as transmitting data packets that eliminates the need to copy data among different functional capabilities, and a common compute plane, which handles the processing layer. The EPIC system had 172 million database operations per second at one EPIC customer. For context, Amazon’s Prime peak day is roughly similar in scale. InterSystems have shown ten times this speed in lab tests.
There was a demo of a medical application which recorded a doctor’s consultation and produced an AI generated summary of the consultation, including bringing in patient record data. The summary could be output in multiple languages at the touch of a button, which is useful in a multicultural context such as the UK. What I particularly liked was that there was a very extensive, multi-step testing process that had been put in place to check the AI-generated summaries, to try to ensure consistency of output. This approach is something that is far from universal in AI tools.
There was a customer presentation from:
Gerd Karnitschnig, Head of Software Solutions International at SPAR, a food retailer with 14,000 shops around the world and 94,000 employees and €19 billion turnover. SPAR used InterSystems as the basis for an ERP style system for their shops with 2,900 database servers. SPAR have a full range retail offering with 50,000 stock keeping units. They undertook a six-year project to roll out new till software based on IRIS across their network. A single server now handles communicate with 6,500 cash registers, over 8,000 PDAs and 18,000 scales. They are rationalising their ERP systems into a core retail platform with central master data, all based on IRIS, which they have had positive experiences with for over thirty years.
This was clearly a very happy customer.
The InterSystems Supply Chain Orchestrator platform has a supply chain data model, with supply chain applications like scheduling and analytics based on top of IRIS. This can ingest data from many data sources such as SAP and output into a variety of formats.
In terms of vision and strategy, Gokhan Uluderya, Head of Product for Data Platforms, described how InterSystems want to build what is effectively a data and AI operating system in the cloud. The idea is that the services should be able to handle many of the data management functions that users need, e.g. hiding explicit file formats from users and handle resource management, so shielding users from having to deal with cloud infrastructure and operations. An example was given where a search was made for a specific vertical solution, where an application was discovered via a search. The application has metadata, and the new platform could discover this metadata and load it into a data catalog, which itself has features like data lineage and knowledge graphs. The product utilises the existing orchestration capabilities to build human and agentic workflows and data pipelines.
This was an interesting vision, though actually turning this into reality will involve some challenges, both technical and marketing. Some elements of this are already there, such as the Data Studio product and the company seems to be well-positioned to build on existing capabilities and strengths
Summary
InterSystems has some differentiated technology that is the foundation of some of the most challenging real time applications deployed today, such as the huge healthcare system. Their underlying IRIS multi model, multi workload data platform is particularly well-suited to the needs of modern AI applications, which need features such as AI and vector search. InterSystems has a strong customer base and is investing heavily in research and development. They spend 20% of their revenue on R&D, which is unusually high for a billion-dollar software company (the industry norm is 13-16% for companies of this size). This should position them well in the coming years.
