The right capability for the task
Interpret requests and events, then select a suitable micro-agent based on intent, context, audience and policy.

SXP.ai brings intent orchestration, purpose-built micro-agents and institutional governance together in one platform. Every capability has a defined scope, with security, privacy and oversight designed into the architecture.
SXP.ai is hosted entirely on Amazon Web Services and is an AWS Validated Partner, giving universities enterprise-grade resilience, security and regional data residency from day one.
Independently assessed against the AWS Foundational Technical Review, covering security architecture, operational practices and platform resilience.
Customer data stays in-region: AWS London for the UK and EU, Sydney for Australia and N. Virginia for the US.
We align with all 14 NCSC Cloud Security Principles. Detailed documentation is available on request.
SXP.ai connects experiences and automation to the systems, information and AI agents your university already uses. Through existing APIs, MCP and agent-to-agent (A2A) communication, SXP.ai can retrieve and update information, initiate workflows and call capabilities provided by other systems or their agents.
As university systems introduce their own AI agents, SXP.ai can orchestrate those capabilities alongside its own micro-agents, using MCP or A2A where appropriate and available.
Open-ended agentic AI is powerful - and unpredictable. SXP.ai deliberately takes a different approach. Every micro-agent is built for a defined purpose, with clear boundaries around what it can say, what information it can access, and what actions it can take.
Interpret requests and events, then select a suitable micro-agent based on intent, context, audience and policy.
Scope institutional websites, documents and knowledge bases by capability and audience, keeping staff-only material separate from student-facing content.
Connect to systems and external agents through approved interfaces, including REST APIs, webhooks, MCP and A2A.
We choose the best AI model for each task, and evolve those choices as the technology improves. Universities get access to new capabilities without redesigning their experience or automation workflows.
Different micro-agents can use different models, and a single micro-agent can combine models where that produces a better result. SXP.ai manages model selection for capability, performance, cost and data residency.
Our standard approach uses Anthropic Claude through Amazon Bedrock, with direct Anthropic and OpenAI connections where appropriate.
The EU AI Act is the world's first comprehensive AI regulation, setting legally binding requirements for AI systems operating in or serving the EU market. It can apply to universities outside the EU too, for example where AI is used by students based in the EU. Rather than leave each institution to work out how every use case is classified, SXP.ai proactively applies high-risk controls across all personalised use cases, providing a consistent standard of AI governance however the platform is used.
EU AI Act high-risk controls are integrated directly into our ISO 27001 and ISO 27701 management systems and platform.
Request our AI governance and security documentation, including EU AI Act controls and NCSC Cloud Security Principles alignment.
No. Customer data is never used to train, fine-tune or improve any AI model.
No. Each micro-agent is limited to its configured tasks, content, permissions and actions. Actions are logged and auditable, with human escalation where appropriate.
Customer data stays in-region: UK and EU data in AWS London, Australian data in AWS Sydney, and US data in AWS N. Virginia.
Yes. ISO 27001, ISO 27701 and Cyber Essentials certificates, together with NCSC Cloud Security Principles and EU AI Act governance documentation, are available on request.
It may. The Act can apply to universities outside the EU where AI systems are used by, or produce outputs for, people in the EU, including EU-based online students. Your legal team should assess your circumstances. We apply high-risk controls across personalised use cases as standard.
Yes. Our platform and website are designed to meet WCAG 2.2 Level AA, with high contrast, reduced motion, dyslexia-friendly fonts and adjustable text sizing.
It depends on the use case. The Act identifies specific high-risk education uses, including admissions, learning-outcome evaluation and exam monitoring. We apply high-risk controls across personalised use cases regardless of classification.
No. Different micro-agents can use different models, or combine models within one micro-agent. We manage selection and can adopt new models as capabilities develop without disrupting your deployment.
Our standard approach uses Anthropic Claude through Amazon Bedrock. We can also connect directly to Anthropic and OpenAI where a use case requires it.
See how ONE/SPACE and AGENT connect intelligent experiences with governed automation.
Talk to us ↗