Cisco Investments backs startups pushing the boundaries of innovation in areas that matter to our customers. Each month, we spotlight how our portfolio companies are advancing the technologies shaping tomorrow’s enterprise.
Artificial Intelligence
Mistral AI introduced Voxtral, a suite of open-source speech understanding models designed to enhance real-world voice interactions. They offer two variants, a 24B model for production-scale applications and a 3B model for local and edge deployments, both released under the Apache 2.0 license. These models provide advanced features such as long-form context handling, built-in Q&A and summarization, multilingual support, and direct function-calling from voice inputs, all at less than half the cost of comparable APIs
Scale AI explored the integration of verifiable rewards with multi-agent learning to advance AI training environments. They demonstrated that combining reinforcement learning with a student-teacher framework, where a teacher provides targeted hints upon failure, significantly improves model learning efficiency and generalization, particularly in complex reasoning and software engineering tasks.
Cloud and Data Center
Dremio announced the 1.0 release of Apache Polaris, an open-source metadata catalog tailored for Apache Iceberg. They introduced features like official Helm charts, JDBC support with Postgres, and a policy store with schema evolution, enhancing deployment and governance capabilities. This milestone signifies Polaris’s readiness for production environments and its commitment to open, community-driven development.
Collaboration
Aisera unveiled a comprehensive System of AI Agents designed to transform IT and HR operations across the enterprise. It enables organizations to deploy and manage both Aisera and third-party agents that autonomously coordinate decisions, orchestrate processes, and execute tasks at scale.
Theta Lake was named in the AIFinTech100 for the fifth consecutive year. This list recognizes the most innovative and impactful companies leveraging AI to revolutionize the financial services sector. Theta Lake delivers practical AI for real-world compliance across voice, video, and chat, and is trusted by top financial firms.
Customer Experience
Fiddler emphasized the importance of agentic observability in developing reliable multi-agent AI systems. They introduced a hierarchical monitoring framework that provides visibility into agent interactions, decision-making processes, and coordination patterns, enabling teams to diagnose issues and ensure system transparency from development through production.
Observability
Harness outlined new MCP tools to enhance its AI-powered chaos engineering capabilities. They allow users to initiate resilience testing through natural language prompts, simplifying the process of discovering, executing, and analyzing chaos experiments across enterprise systems.
Quantum
Aliro published a research paper demonstrating the use of Aliro Simulator to validate the BBM92 quantum key generation protocol with high physical realism. It confirmed that the simulator accurately aligns with both experimental results and theoretical models.
Security
AppOmni highlighted its presence at Black Hat 2025, emphasizing the growing risks of SaaS and AI integration in enterprise environments. They stressed the need for proactive SaaS Security Posture Management to address challenges like identity sprawl, misconfigurations, and third-party app risks.
Expel detailed its approach to integrating AI and automation within its Managed Detection and Response (MDR) service. They emphasized reducing alert fatigue by correlating signals across diverse security tools, enabling analysts to focus on high-priority threats. This strategy has led to tangible benefits for customers, including faster response times and more efficient security operations.
Island demonstrated its approach to safeguarding sensitive data within browser memory. They developed a two-pronged solution combining targeted memory deallocation hooks with an automated validation framework to prevent residual plaintext data from persisting after deletion. This strategy enhances protection against memory dump tools, securing assets like cookies and authentication tokens.
Mitiga released Helios AI, a cloud-native security solution designed to automate threat detection and response across SaaS and multi-cloud environments. Its first feature, AI Insights, dynamically adjusts risk levels and executes contextual triage tests to distinguish genuine threats from routine activities. By combining advanced AI engines with a full-fidelity data lake, it enhances SecOps efficiency and scalability.
Securiti launched its Rules Matrix to streamline global direct marketing compliance. It automatically classifies consent statuses across 58 jurisdictions, integrating with CRMs to ensure marketing activities align with regional regulations. This tool helps organizations proactively manage consent, reducing the risk of non-compliance and enhancing consumer trust.
SGNL advised CISOs on strategies to securely implement agentic AI while accelerating enterprise transformation. They emphasized the importance of enforcing Zero Standing Privilege and real-time, context-aware authorization to prevent AI agents from inadvertently accessing sensitive data. This approach positions security leaders to enable innovation without compromising organizational safety.
TrustCloud issued a practical guide for implementing responsible AI governance aligned with ISO 42001 and the NIST AI Risk Management Framework. They provided actionable steps, from team setup to continuous monitoring, to help organizations operationalize these standards and ensure trustworthy AI development.