Services

Find out which AI uses are worth testing, and which to drop.

Two fixed-scope offers. The assessment scores each candidate AI use on six criteria and puts a go, fix-first or stop decision in writing. The pilot tests a “go” on your own examples before anyone relies on it.

A “stop” is a valid result. taliQ gives no legal, privacy or cyber-security advice.

Index cards sorted into three stacks on a sandstone desk, the middle stack set slightly forward.

Is this right for you?

  • Staff already use ChatGPT, Copilot or Gemini and nobody has written down what for
  • You have a list of AI ideas and no agreed way to choose between them
  • A vendor is pitching an AI feature and you want an independent view before you buy
  • An AI trial ‘seemed fine’ but nobody can say how often it was wrong
  • You run a construction or trade business and want to know where AI helps with quotes, tenders, site reports or quality paperwork

The problem to solve

Most AI trials fail quietly: nobody agreed beforehand what good output looks like, so nobody can say whether the trial passed. The assessment fixes the first half of that by scoring each use against the same six criteria. The pilot fixes the second half by writing the acceptance criteria down before testing, the way an inspection and test plan works on a construction site.

What the assessment looks at

Five lenses behind the six scores.

  1. People

    Roles, capability, adoption and who carries the change.

  2. Process

    The workflow, decisions, exceptions and sources of friction.

  3. Data

    Availability, quality, sensitivity and permitted use.

  4. Technology

    Fit, integration, reliability and operating ownership.

  5. Governance

    Accountability, boundaries, review and ongoing control.

Fixed-scope offer

AI Use-Case Assessment

Up to eight candidate AI uses scored on one worksheet, each with a written go, fix-first or stop decision.

Duration
2 weeks
Your time
About 4 to 6 hours across the two weeks
Fee
$5,000 + GST

Right for you if

  • Staff already use ChatGPT, Copilot or Gemini and nobody has listed what for
  • You have a list of AI ideas and no agreed way to choose between them
  • A vendor is pitching an AI product and you want an independent view first
  • You run a construction or trade business and want to know where AI helps with quotes, tenders, site reports or quality paperwork

In scope

  • An inventory of the AI tools staff use today and what information goes into them
  • Up to eight candidate use cases, each walked through with the person who does the work
  • Scoring of each use case on value, frequency, data readiness, consequence of error, ease of checking and information sensitivity
  • A written decision for each: go to pilot, fix a named prerequisite first, or stop

Not included

  • Building or configuring any AI tool
  • Legal, privacy or cyber-security advice (named as a referral where needed)
  • Vendor selection beyond a short list of options for each go decision

What you provide

  • A named owner who can make decisions about AI use
  • Access to the people who do each candidate task (30 to 45 minutes each)
  • De-identified examples of the work involved; no customer records are needed
  • A copy of any existing IT, privacy or acceptable-use policy

Deliverables

DeliverableFormatWhat it contains
AI-use registerSpreadsheetEvery AI tool in use: who uses it, for what, which information goes in and whether that is allowed.
Use-case assessment worksheetsOne page per use caseThe workflow step, the six scores, the hold points that apply and the evidence each score rests on.
Decision noteFour to six pagesRanked use cases, the go, fix-first or stop decision for each, and the reason in plain English.
Read-out meeting60 minutesWalk through the decisions with the owner and anyone accountable for the affected work.

The outcome

You know which AI uses are worth a pilot, which are blocked and by what, and which to drop, with the evidence written down.

Book a free discovery call about this offer

Fixed-scope offer

AI Pilot Inspection & Test Plan

One AI use tested against written acceptance criteria on your own examples before anyone relies on it.

Duration
4 weeks
Your time
About 6 to 8 hours across the four weeks
Fee
$6,000 + GST

Right for you if

  • A use case scored “go” and you want proof it works before rollout
  • You are about to switch on an AI feature in software you already use
  • A previous AI trial ‘seemed fine’ but nobody can say how well it performed

In scope

  • One use case, one tool or configuration
  • Acceptance criteria agreed in writing before any testing
  • A test set of 20 to 50 de-identified examples drawn from your real work
  • A witnessed evaluation run, recorded result by result

Not included

  • Building custom software (quoted separately if the pilot needs it)
  • Ongoing monitoring after the release decision
  • Testing more than one tool or configuration

What you provide

  • A completed use-case assessment (ours or your own, using the published worksheet)
  • 20 to 50 de-identified examples of the task, with the answer a competent person would give
  • One staff member to review outputs during the evaluation run (about 3 hours)
  • Access to the tool, or a trial account

Deliverables

DeliverableFormatWhat it contains
Inspection and test plan (ITP)Two to three pagesWhat is checked, the acceptance criterion for each check, how many examples, who checks and who signs.
Evaluation recordSpreadsheetEvery test example, the AI output, the reviewer’s verdict and the reason.
Release record or NCROne pageEither the signed hold-point release with conditions of use, or a non-conformance report stating what failed and what would need to change.
Operating notesOne to two pagesHow staff use it, what they must check every time, and when to stop and log an NCR.
90-day records check45 minutesReview the NCR log and spot-check results against the release conditions.

The outcome

A yes or no on one AI use, backed by test evidence, with the conditions staff must follow if it goes ahead.

Book a free discovery call about this offer

Inspect the method first

Common questions

Do we need an AI strategy first?

No. The assessment starts from the tasks your staff actually do and produces the shortlist a strategy would otherwise guess at.

Will you recommend specific tools?

For each “go” decision the note lists the realistic options, including tools you already pay for.

What if nothing scores well?

Then the decision note says so, and why. A written “stop” that saves you a subscription and a failed rollout is a useful result.

Can we use the worksheet ourselves?

Yes. The worksheet and the online scorer are published free. The paid assessment adds an independent assessor, the walk-throughs with your staff and the written evidence.

What would you like to work on?

Bring the question you have now. The first call is free.