OpenMatter adds secure AI & data tools to platform

OpenMatter Network has expanded its platform with new tools for secure application development, AI model management, privacy-preserving machine learning and data collaboration. The update is its first major product expansion since the commercial launch.

The additions include MatterSDK, Model Router and MatterML V2, now available as part of the platform. They are aimed at organisations, developers and researchers working with sensitive data and artificial intelligence systems.

MatterSDK is a new client layer that gives developers access to OpenMatter's MatterChain environment. It includes MatterVault, which uses threshold cryptography to split API keys, credentials and other secrets into shares held across multiple parties, so no single machine can decrypt the information on its own.

The approach is intended to reduce the need for specialist cryptography expertise among developers building applications on the platform. OpenMatter said the software development kit offers a simpler way to use those controls across its environment.

Another addition, Model Router, is aimed at companies using AI models from multiple providers. OpenMatter said it provides a single gateway for managing access to models from OpenAI, Anthropic, Google and self-hosted endpoints in on-premise deployments.

According to the company, users can set routing rules, switch models without redeploying applications, rotate provider credentials centrally and monitor model usage through the service. It added that provider keys remain outside individual AI agent environments, limiting exposure if an agent is compromised.

Privacy tools

OpenMatter also introduced MatterML V2, an updated privacy-preserving computing system. The product allows multiple organisations to train or run models on combined information without requiring participants to reveal their underlying data to one another or to the computing infrastructure.

The release also aims to make secure multi-party computation easier to use by offering a graphical interface instead of relying on specialist cryptographic programming. Analysts can run privacy-preserving workflows without writing code, the company said.

OpenMatter said early benchmarks by its cryptography team showed a 1,000-fold increase in efficiency for MatterML V2. It did not provide independent verification of the result.

Alongside those releases, the company outlined a feature called Communities for research groups, scientific organisations and other member-led groups. It is intended to help users organise around datasets, set privacy levels, discuss information and govern what their groups endorse.

The changes reflect a broader push by technology suppliers to give businesses more control over data use and model access as AI spreads across their operations. They also point to rising demand for systems that support collaboration across organisations without requiring the direct sharing of raw data.

OpenMatter frames that strategy around what it calls a verification architecture, which separates verification from the applications, models and infrastructure operating above it. The company argues that this structure allows new tools to be introduced without replacing the underlying cryptographic foundation.

Renee Davis, Chief Executive Officer and Co-Founder of OpenMatter, said the company did not view its initial market debut as a finished product. "When we launched OpenMatter in June, we weren't launching a finished destination," Davis said. "We were establishing an architecture designed to grow with the needs of our customers and with the rapid changes taking place across AI and secure computing. These additions demonstrate how quickly we can extend the platform while preserving the cryptographic foundation everything is built upon."

The company said that flexibility matters because AI models, providers and security demands are changing quickly. It argues that this pace makes conventional technology refresh cycles harder for enterprises to follow.

Davis said the latest additions reflect a broader design principle rather than isolated product releases. "These aren't isolated features being bolted onto a platform," she said. "They are examples of what an open Verification Architecture makes possible. Customers should be able to take advantage of new models, new cryptographic techniques and new ways of collaborating without having to replace the foundation underneath them every time computing changes."

She added that the company sees uncertainty over future AI models, computing environments and security risks as a central design challenge for enterprise technology. "The future of computing is going to keep changing," Davis said. "No one can tell an enterprise today exactly which AI models, computing environments or security challenges it will face three years from now. What we can give them is an architecture that is ready to evolve with that future."