Architects of Trust: Fireside Chat with Industry Leaders
Architects of Trust: Fireside Chat with Industry Leaders
Confidential Computing Summit 2024
Mark Bower, VP of Product at Anjuna, sits down with leaders from the financial services, blockchain, and security sectors to discuss their adoption and utilization of Confidential Computing technology.
Confidential Computing Summit 2024 - Fireside Chat With Industry Leaders
27:36
Left click drag to rotate. Two-finger drag / mouse wheel to zoom. Right click drag / Command (or Control) drag to pan.
AI Agents Don’t Just Assist. They Act
AI is evolving fast. Beyond copilots, Agentic AI empowers autonomous agents to act on behalf of humans. Writing code, making decisions, and triggering tasks across business systems. These agents interact with APIs, data, and tools with unprecedented privilege and autonomy.
Privilege Without Protection is Risky
Agentic AI holds credentials, processes sensitive data, and triggers actions. But without strong verification, these autonomous actors become high-value targets for:
Memory poisoning: manipulating agent state
Tool misuse: abusing integrated APIs
Identity spoofing: impersonating trusted agents
Rogue agents: acting outside of governance
How Agentic AI Proves Itself: Trust Rooted in Hardware Attestation
Anjuna makes Agentic AI safe by isolating agents, data, and MCP servers in Trusted Execution Environments (TEEs). This hardware-enforced foundation ensures:
Immutable identity: agents prove exactly what they are before access
Runtime integrity: only authorized code executes
Always-on confidentiality: data and credentials remain encrypted
Policy-based control: access is granted only after attestation
How Anjuna Enables Trusted Agentic AI
Anjuna Seaglass
Provides hardware-rooted identity, runtime attestation, and encrypted execution. Seaglass ensures agents and MCP servers run only trusted code inside TEEs, with data and credentials protected even from cloud admins.
Secures collaboration, and enables multi-party AI data fusion in a confidential environment. Northstar governs MCP access, isolates sensitive data, and prevents rogue or poisoned agents from breaking trust.
Establishes a trusted supervisor that governs autonomous agents in real time. The Manager for Agentic AI verifies every request through hardware-rooted attestation, enforcing policies and controlling tool or data access based on context.
Trusted Agentic AI in Action
Anjuna makes Agentic AI safe for sensitive industries, enabling trusted agents to process critical data, act with proven integrity, and collaborate without risk.
Financial Services: autonomous reconciliation agents with verified access
Payments: fraud agents working across providers without exposing data
Healthcare: AI agents analyzing patient data securely across institutions
Defense: mission systems and AI copilots running in TEEs
Left click drag to rotate. Two-finger drag / mouse wheel to zoom. Right click drag / Command (or Control) drag to pan.