Unmanaged AI: Getting to a defendable AI position

Level 1: Foundation and Risk Mitigation

The first level is about acknowledging reality and making it safe.

Lexel starts by helping organisations:

• Publish a simple, organisation‑wide AI policy that sets non‑negotiables (such as anonymising data before sending it to external AI tools), defines what is in‑ and out‑of‑bounds, and establishes who owns AI governance.
• Run a staff survey to uncover where AI is already being used, which tools are in play, and where the biggest risks and best ideas are hiding.

Within a few weeks, most organisations move from “we think staff are using AI but we don’t know how” to “we have visibility of usage, a baseline policy, and an agreed risk posture”. This alone is a major step forward in terms of legal defensibility and board confidence.

Level 1 gives you:

• AI policy you can stand behind – a concise, organisation‑wide AI policy that sets non‑negotiables (for example, anonymising data before sending it to external tools), clarifies what is in and out of bounds, and names who owns AI governance.
• AI usage and risk survey – a structured survey to uncover which AI tools are already in use, where confidential data may be leaving the organisation, and where your best internal AI ideas are hiding among early adopters.

Where AI devices fit at Level 1

Policy alone is not enough; it must tie back to the devices people actually use. When you know that confidential data is being accessed from unmanaged or legacy endpoints, the risk profile looks very different from a fleet of modern, encrypted, policy‑enforced devices.

This is where HP AI‑enabled PCs start to add value even at Level 1:

• Modern HP devices support strong identity, encryption and endpoint protection out of the box, which makes it easier to enforce the AI policy you just wrote.
• Features like hardware‑based security and secure boot reduce the chance that sensitive data copied into AI tools is being handled from a compromised device.

The initial outcome is clear: once you understand who is using AI and from where, you can prioritise moving high‑risk users (for example, finance, HR and executives) onto secure, well‑managed HP endpoints first.

Level 2: Secure general productivity – standardising on safe, sanctioned AI

Once you have the basics in place, the second level focuses on getting a safe, standard AI assistant into everyone’s hands, rather than leaving staff to self‑select tools.

Here Lexel works with IT and business leaders to:

• Deploy secure, tenant‑based AI chat (such as Microsoft Copilot) across the organisation, so everyday tasks like drafting, summarising and analysing can be done inside your own environment.
• Align device, Windows and security posture so these tools run reliably and securely – including where AI‑capable PCs and modern endpoints make a difference.
• Deliver targeted training and prompt‑engineering guidance so productivity gains extend beyond a few power users, especially to people who may be skeptical of AI.

The goal for Level 2 is simple: replace shadow AI with a secure, organisation‑wide assistant that your legal, security and compliance teams can stand behind.

Why HP AI devices matter more at Level 2

At Level 2, HP AI devices move from “nice‑to‑have hardware” to a practical enabler of your AI rollout:

• Performance for AI assistants: New HP AI PCs incorporate dedicated NPUs for on‑device AI, improving responsiveness and reducing the load on shared infrastructure when users rely heavily on Copilot‑style tools.
• Security at the endpoint: HP’s hardware‑based security stack helps protect below, in and above the OS, which complements your M365 security and reduces the attack surface of AI‑driven work.
• Manageability at scale: Standardising on a small number of HP device families makes it easier for Lexel’s Managed Services teams to manage updates, policies and support. This becomes critical once AI tools are deployed widely.

In short: Levels 1 and 2 set the governance and platform foundations. HP AI devices and Lexel’s managed services ensure those foundations are strong enough to carry AI into everyday work.