Unmanaged AI: Activating the AI you already own

Level 3: Activating the AI hiding inside your existing systems

Only once governance and a secure baseline are in place does Lexel turn to activating the AI hiding inside your existing systems.

At this level, organisations:

• Inventory AI features in current line‑of‑business, collaboration and productivity apps.
• Run a security and permissions health‑check (particularly across SharePoint and file shares) to ensure AI doesn’t inadvertently surface sensitive content to the wrong people.
• Prioritise a short list of business‑ready use cases by department – for example, finance variance analysis, customer service summarisation, HR documentation, or sales proposal support.

Level 3 is where AI stops being a “technology topic” and becomes part of how core business processes run, using tools and licenses most organisations are already paying for.

A later blog in this series will break down how Lexel approaches this step in more detail.

Critical precursor: security and permissions health‑check

Before any of this embedded AI is turned up, Lexel insists on one critical step: a security and permissions health‑check, particularly across SharePoint and file shares.

The reason is simple. When AI tools are connected to your content and data:

• Any existing permission errors or oversharing are amplified.
• AI can surface documents and records that users technically “have access to” but were never meant to see (for example, salary spreadsheets, performance reviews, board minutes).

Lexel’s AI 130 assessment focuses on:

• Identifying high‑risk libraries and shares where sensitive information may be exposed.
• Validating permissions and inheritance models.
• Re‑architecting information structures where needed so AI only sees what it should.

This work aligns closely with Lexel’s Cybersecurity Posture Framework, which focuses on tightening identity, access and data controls before new capabilities are enabled.

Prioritising business‑ready use cases

Once security is under control, Lexel helps you prioritise business‑ready use cases by department:

• Finance: variance analysis, narrative generation for monthly reports, invoice coding assistance.
• Customer service: automatic summarisation of tickets and calls, suggested responses, next‑best‑action prompts.
• HR: drafting policy updates, contracts and communications based on templates and historical documents.
• Sales and marketing: proposal drafting, content personalisation, opportunity summarisation and call notes.

The emphasis is on quick wins that use existing licences and data structures, with clear owners and metrics.

At this point, AI stops being a theoretical strategy and becomes a concrete set of improvements inside the systems people already live in every day.

How HP AI devices support Level 3

As AI capabilities become more tightly integrated with line‑of‑business tools, the endpoint experience starts to matter more:

• Some AI features run partly on the device (for example, local transcription, summarisation or on‑device models) and benefit from the dedicated NPUs and performance in HP AI PCs.
• HP’s security stack helps ensure that even when rich data and AI outputs are cached locally, the device remains hardened against malware and tampering.
• Standardised HP fleets also reduce variability in performance and support, which is critical when AI features become part of core workflows rather than optional add‑ons.

In practice, that means fewer support calls about “AI features not working”, more predictable user experience, and a more secure posture as sensitive insights move closer to the edge.

If you recognise your organisation in any of the challenges above, the quickest next step is simple: