AI does the work it can. People own the judgment.

Clients and accreditation bodies ask how Speeki uses AI in audits. The AHA Framework is our answer. It sets, for every task in an audit, what AI carries out, what a person reviews and what only a person decides.

What we are asked

Three questions about AI in an audit.

Speeki developed the AHA Framework, and applies it in its own assurance work, to answer them in the same way every time.

What does AI do?

AI works on defined tasks: gathering and checking evidence, comparing documents with a standard and drafting findings. Each task has a set place in the framework.

Who is accountable?

A qualified person. Every task has a named practitioner, and AI does not make the judgments that an audit conclusion rests on.

How is it controlled?

The framework decides the level of human involvement before the work starts. It does not leave that to chance or to the individual auditor.

The AHA Framework

Three zones for every audit task.

AHA stands for ESG assurance human-AI task allocation. Each task falls in one of three zones. The zone sets how much AI does and how much a person does.

Zone 1

AI-executable

Nicole AI carries out the task. A practitioner oversees the process and handles exceptions.

Examples

  • Collecting and organising evidence
  • Checking documents against explicit criteria
  • Cross-checking figures across documents
Nicole AI
Carries out the task
Person
Oversees and handles exceptions

Zone 2

Human-supervised AI

Nicole AI prepares the work. A practitioner reviews it, tests it and decides what stands.

Examples

  • Identifying gaps between a report and the standard
  • Reviewing anomalies in reported data
  • Drafting findings for review
Nicole AI
Prepares the work
Person
Reviews and decides

Zone 3

Human-irreplaceable

A practitioner leads the task. AI may support it, but does not decide.

Examples

  • Setting materiality and scope
  • Judging whether a disclosure fairly represents performance
  • Forming the assurance conclusion
Nicole AI
May support, does not decide
Person
Leads and decides

How a task is placed

Two tests decide the zone.

What the task needs

We assess each task against four characteristics.

  • Professional judgment. How far the task depends on an expert's assessment.
  • Ethical exposure. How much harm a wrong call could cause.
  • Relational sensitivity. How much the client relationship shapes the task.
  • Novelty. How unfamiliar the situation is.

What clients and accreditation bodies expect

Some tasks stay with a person because the people who rely on the audit expect it, even where AI could do the work.

We call this the confidence threshold. We set it by what clients and accreditation bodies need to trust the result, not by what AI is capable of.

Accountability

A named person is responsible for every judgment.

Speeki audits with Nicole AI, its AI team member, and qualified practitioners. The framework keeps their roles separate and on record.

Read how Speeki governs its AI
  1. Nicole AI

    AI does not conclude

    Nicole AI does not form an audit conclusion or make a certification decision.

  2. Lead auditor and Certification Committee

    People decide

    The lead auditor makes a recommendation. Speeki's Certification Committee makes the certification decision.

  3. Audit record

    The work is recorded

    For each task, we record the zone, the practitioner responsible and the judgment applied.

What this means for you

Clear answers for the people who rely on us.

For clients

  • You can see which parts of your audit AI carries out and which a person decides.
  • A qualified practitioner stands behind every conclusion.
  • AI gives broader coverage of evidence without taking judgment away from people.

For accreditation bodies

  • Human responsibility for audit conclusions is defined in advance, not case by case.
  • The framework allocates tasks against the expectations of clients and accreditation bodies.
  • Each task shows who was responsible for it.

FAQs

Questions about how Speeki uses AI.

No. Nicole AI supports the work. Practitioners form the conclusions, and Speeki's Certification Committee makes the certification decision.

Speeki uses Nicole AI, its AI team member, for defined audit tasks. The AHA Framework sets which tasks it carries out and which a person leads.

Yes. The AHA Framework is how Speeki allocates tasks between Nicole AI and practitioners in its assurance work.

No. Speeki does not provide consulting of any kind. AI is used to carry out and support audit tasks, not to advise clients.

See how Speeki governs its AI, or contact us with a specific question.

Questions about AI in your audit?

Ask us how AI and people share the work.

Talk to Nicole AI, or speak to the team about how the AHA Framework applies to your engagement.