Information AuditAssess

From a portfolio of pilots to the gaps that stop them adding up.

A four to six week enterprise AI readiness assessment. We classify the AI portfolio, test the cases that matter against what handing work to an agentic worker would require, and name the small number of structural gaps in the way. Then we put our name under the diagnosis.

Discuss an audit How the three phases fit together

Duration

4 to 6 weeks

Engagement Type

AssessmentFirst phase of Assess, Define, Embed

Typical Client

CIO, CDO, CTO, AI programme leadEnterprises with a year or more of AI initiatives behind them

Output

DiagnosisNamed gaps, classified portfolio, sequence of decisions

The Challenge

The pilots work. The programme does not add up.

An AI programme that has been running for a year has assistants in use, models in production and a list of initiatives that keeps growing.

What it usually does not have is a way to say why none of it compounds.

Every initiative is tagged productivity

Minutes saved across many tasks, reabsorbed into the working day. Adoption is high and there is no line in the accounts to point to.

Decisions are taken per pilot

Each initiative chooses its own stack, its own way of holding rules, its own idea of what a customer or a supplier is. Nothing accumulates.

The context lives inside the pilots

Definitions, rules and process knowledge are baked into individual solutions. They cannot be shared, versioned or withdrawn.

A platform decision is pending and the requirement is not written

The question has arrived as "which platform". Vendors have been invited. Nobody has yet said what the organisation is asking for.

Deliverables

What you walk away with

A diagnosis, not a score. Each deliverable is a decision the organisation can take.

Classified portfolio

Every initiative placed three ways: by horizon (Inform, Act, Run), by investment type (pilot, option, improvement, venture), by economic effect (productivity, capacity, operating model).

Readiness of the cases that matter

For the initiatives that could act: what would have to be true for the organisation to hand the work over, tested against the five conditions of an agentic worker: a scope of work, authority, rules, an owner, a record.

Named gaps

A small number of structural gaps, each with what it blocks and who would have to own it. A role nobody holds, a layer nobody owns, a foundation that does not exist yet.

Sequence

The order in which the gaps close, since they close in an order. Not quick wins and strategic bets: dependencies.

Requirement, where a platform decision is pending

The requirement written before any vendor is considered, ready to become a request for proposal in Define.

Executive readout

The diagnosis presented to the executive team and defended in the room, with the Define phase scoped if the organisation wants to continue.

Process

Four to six weeks. Four steps.

Week 1

Scope

Interviews with the programme lead, architecture, the platform owners and the business owners of the cases that matter. Boundaries agreed.

Weeks 2 to 3

Map

Portfolio classified. Platforms, information sources and roles around each initiative mapped. The cases that could act tested against the five conditions.

Week 4

Name

The gaps named and sequenced. Where a platform decision is pending, the requirement drafted.

Weeks 5 to 6

Decide

Executive readout. The organisation decides whether to move into Define, and on what.

Who It's For

Enterprises with a year or more of AI initiatives behind them

CIO / CTO

Has been asked for a view on agentic AI that does not come from a vendor.

CDO / Head of Data & AI

Runs the programme and can see the initiatives are not compounding.

AI programme lead

Has to explain to the board why adoption is high and the return is hard to find.

Head of architecture

Suspects the role for AI is vacant and wants an outside view that says so.

FAQ

Common questions about the Information Audit

What is an Information Audit?

An enterprise AI readiness assessment that names what structurally blocks an organisation from moving AI from Inform to Act. It classifies the AI portfolio, tests the cases that could act against what acting requires, and delivers a small number of named gaps in sequence. It is not a maturity scorecard and does not benchmark against peers.

How is this different from a maturity assessment?

A maturity assessment scores capabilities. This names gaps. A capability score tells you where you sit; a named gap tells you what has to exist before an AI agent can be trusted with work as an agentic worker, and who has to own it.

What do you mean by Inform and Act?

Inform is AI that answers, summarises and drafts while a person does the work. Act is work handed to an agentic worker: an AI agent with a scope of work, authority, rules, an owner and a record. The audit tests which initiatives could cross that line and what is in the way.

How long does it take?

Four to six weeks. Most stakeholder time is in interviews during the first two weeks.

What happens after the audit?

The organisation decides. The readout scopes the Define phase if you want one: holding the enterprise architecture role for agentic AI on an interim basis and writing the foundational architecture. Many organisations also use the diagnosis on its own to reset the programme.

Do you sell or resell software?

No. Ardinois is vendor-independent. Where the audit finds a pending platform decision, it writes the requirement before any vendor is considered.

Discuss your audit

Not sure whether the programme is adding up? That is what the audit answers.

Tell us where you are. Pieter or Sander will reach out within one business day to scope a 30-minute call.

Independent. No software to sell.
4 to 6 weeks. Named gaps, in sequence.
Confidential.

Sending a formal RFP? Use our RFP submission page instead.

Our partners previously led Data & AI at