AI Security & Governance · Program

Securing AI Agents & LLM Applications

Two hands-on days for teams that build or deploy AI agents and LLM applications, so security ships with the feature.

Request a proposalSee the agenda

Delivered in partnership withAccuroAI

Duration
2 days
Format
On-site or live virtual, lab-heavy
Group
Up to 25 participants
Audience
Developers, DevOps, AppSec, platform teams

Why this matters now

Agents act on their own. Secure them first.

AI agents can read your data, call your APIs and act on their own. That is the value, and the risk. One instruction hidden in a document, email or web page can lead an agent to leak data or take the wrong action. Many engineering teams are shipping agents faster than their security practices are adapting. This program closes the gap, using the OWASP Top 10 for LLM Applications, the OWASP Top 10 for Agentic Applications and hands-on labs.

Who should attend

  • Developers building with LLM APIs, RAG or agent frameworks
  • AppSec and product security engineers
  • DevOps and platform teams deploying AI tools and MCP servers
  • Architects reviewing AI system designs

What your team will be able to do

  • Threat-model an LLM application or AI agent
  • Find and test for direct and indirect prompt injection
  • Apply least privilege to agent tools, APIs and MCP servers
  • Use the OWASP Top 10 for LLM Applications and for Agentic Applications as review checklists
  • Add guardrails, logging and human approval for high-risk actions

Program agenda

What we cover

Every module is tailored to your industry, your approved tools and your policies before delivery.

Request the full agenda

  1. How LLM apps and agents fail

    RAG and agent architectures, trust boundaries, and where attackers get in.

  2. Prompt injection, hands-on

    Direct and indirect injection, jailbreaks and data exfiltration, practised in labs.

  3. The OWASP Top 10 for LLM Applications

    Sensitive information disclosure, supply chain, excessive agency, improper output handling and the rest of the list, applied to real designs.

  4. Agents, tools and MCP

    Risks from the OWASP Top 10 for Agentic Applications: securing tool calls, MCP servers, credentials, memory and permissions, with human approval for high-risk actions.

  5. AI coding assistants and the developer workstation

    Risks from AI coding tools, repository context sharing and local models.

  6. Runtime governance

    Policy-as-code, monitoring and audit trails for agents in production, with an AccuroAI demonstration.

How we deliver

Built around your organisation

On-site or live virtual

At your offices anywhere in the world, or live virtual for distributed teams across time zones.

Tailored to your organisation

Examples, exercises and policy content aligned with your sector and the tools you have approved.

Certificate and report

Certificates for participants and a completion report for HR, compliance and audit.

In partnership with AccuroAI

AI security training, taught on a real AI security platform

Our AI Security & Governance programs are delivered in partnership with AccuroAI, an enterprise AI security and governance platform covering shadow AI discovery, prompt data loss prevention, AI agent security and endpoint AI governance. Your teams learn the controls on a live platform, not on slides.

  • Live demonstrations and hands-on labs on the AccuroAI platform
  • Realistic scenarios: shadow AI, prompt data leakage and AI agent misuse
  • Vendor-neutral skills: no purchase needed to attend, and the methods apply whatever tools you use

Go deeper with AccuroAI: AI agent security and the AI agent governance framework.

Explore the AI Security Academy

FAQ

Frequently asked questions

Can’t find your answer? Email info@prodinnov.co

What background do participants need?

Comfort reading code (Python or JavaScript) and a working knowledge of web APIs. No machine learning background is needed.

Can you use our own application as a case study?

Yes. We can run a threat-modelling session on one of your AI applications as part of the program.

Is it vendor-neutral?

Yes. The principles apply to any model provider or agent framework. AccuroAI is used for the runtime governance demonstration.

How is the program priced?

Each engagement is scoped and quoted after a discovery call. The quote depends on participant numbers, format, location and whether we threat-model one of your own applications. You receive a written proposal first.

Request a proposal

Bring Securing AI Agents & LLM Applications to your teams

Tell us about your teams and goals. We’ll set up a short discovery call, then send a scoped proposal.

Request a proposal

Keep going

Related programs

All programs
With AccuroAI

AI Risk & Strategy for Leaders

A half-day briefing for CXOs and boards: where AI pays back, where it exposes you, and who owns it.

Half dayCXOs and boardView program With AccuroAI

Safe AI at Work

Company-wide training on using ChatGPT, Copilot, Gemini and other AI tools productively, without leaking company or customer data.

90 minAll employeesView program With AccuroAI

Shadow AI & Prompt Data Leakage Defense

Hands-on workshop for security and IT teams: find unapproved AI, stop sensitive data in prompts, and respond to AI incidents.

1 daySecurity and ITView program