Two Level 3 qualifications in Future Fluency, specified in full.

Roughly four statements in five carry no canonical event identifier. In the measured window only 19% of rows carried one. The rest are real activity recorded through the front-end path without a registry identifier, so they cannot be attributed to a named event. They are still statements, and they still carry a learner, a verb, an object and a time. They are simply not countable by event.

The two extensions are populated from Zavmo's own identifiers, not from Gateway's. Today the outcome and criteria slots are filled from the learner's own outcome record inside the platform. They do not yet carry references in the form used on this page, such as ZFF-AI-03 3.3. Mapping these units into those slots is work to be done before first registration.

Until it is done, the tie between a statement and a criterion on this page is a mapping we intend rather than one the data already carries. The mechanism is built and proven. The mapping to these six units is not.

A status is a measurement with a date on it. Eight events that were wired in July are not any more, because the feature that emitted them was deliberately removed. Two of those had already fired, so they stay in the observed snapshot permanently. A further six event identifiers have fired in production that are not in the registry at all. No status should be read without its date, and none of those retired or unregistered events is referenced in the unit sections below.

xAPI records the platform, not the job. It has no sight of what a learner did in a spreadsheet, a workflow tool, a meeting or a conversation with their manager. At Level 3 most of the assessed activity is in the job, so for most criteria the platform record is corroboration around workplace evidence rather than the evidence itself. The unit sections say plainly where that line falls.

The scope rule in section 3 is stated absolutely, and three units cross it. Section 3 says that the moment an outcome reaches a colleague, a team or a department, it has left Level 3. But ZFF-DA-03 4.2, 4.4 and 4.5 require the learner to present a recommendation to the person who decides and answer a challenge to it, ZFF-PJ-03 5.4 and 5.5 recommend to the learner's manager, and ZFF-GM-03 4.3 requires feedback from a person who saw the result. Those are right, and the rule as written appears to forbid them.

The rule is really about whose work is being changed, not about who is in the room. As stated it does not say so, and that is the kind of internal contradiction an external quality assurer reads first.

Change proposed Reword the section 3 sentence to separate the two: the work under improvement stays the learner's own, while the audience for a recommendation or a piece of feedback may be another person. No criterion changes.

ZFF-GV-03 gives 60% of its hours to guided learning, against 50% or less everywhere else. The unit shows 30 guided hours against 50 TQT. The comparison across the Diploma is ZFF-AI-03 at 50%, ZFF-GM-03 at 50%, ZFF-DA-03 at 50%, ZFF-AU-03 at 42% and ZFF-PJ-03 at 10%. ZFF-GV-03 is the only unit above half, and it is a unit whose skills outcomes are almost entirely workplace conduct, which argues for fewer guided hours rather than more.

Change proposed Reduce to 25 guided hours, matching the other 50 TQT unit. The knock-on is small and needs to be carried through: the Diploma total falls from 155 to 150 guided hours in the unit index and in section 2.4, and the guided share in section 2.4 moves from 40% to 38%. TQT, credits and the Award are unaffected.

Some unit aims were condensed and lost their trailing scope clause. ZFF-DA-03's aim now ends "Scope is personal." The fuller original ended by saying what the recommendation is not about, namely a department's performance. ZFF-GM-03 and ZFF-PJ-03 carry no scope sentence in their aims at all, while ZFF-AI-03, ZFF-GV-03 and ZFF-AU-03 each carry a full one.

Scope is what holds these units at Level 3. The aim is the first thing a reviewer reads and the last thing a centre reads before writing an assessment plan, so it is the wrong place to be terse.

Change proposed Reinstate the fuller wording in ZFF-DA-03, and add matching clauses to ZFF-GM-03 and ZFF-PJ-03 so that all six aims state the boundary in the same shape.

ZFF-PJ-03's aim uses a verb its own level check rules out. The aim says the learner "redesigns" the process. The same unit's level check states that the verbs stay at level and that design is not among them, and ZFF-AU-03's level check says plainly that design is a Level 6 verb and using it would mis-level the unit. No criterion uses it, so this is confined to the aim, but it is the sentence a reviewer reads before the criteria.

Change proposed Replace "redesigns it" with "changes it" or "sets up an improved version of it". ZFF-AU-03's aim also contains "designed to break it", which describes the test cases rather than an act of the learner; that one is defensible and we suggest leaving it.

Nine criteria presuppose an event that may not happen. They are assessable when the event occurs and unachievable when it does not, and nothing currently tells a centre what to do in the second case.

ZFF-GM-03 4.3 needs a person willing to give feedback. ZFF-DA-03 5.1 needs a decision to be made and communicated. ZFF-GV-03 3.5 needs a situation where the rules did not clearly apply, 4.4 needs a response to a reported risk, and 5.2 is worded so that the learner's practice must have fallen short. ZFF-AU-03 4.2 and 4.3 need the automation to produce a wrong result. ZFF-PJ-03 4.1 needs a problem to arise and 4.5 needs one to remain unresolved.

Two of the unit aims lean into this, saying the learner runs the automation until it is shown reliable and runs the new process until it fails at least once. A qualification should not depend on a failure arriving to order.

Guidance only The criteria are sound. Add a nil case to the assessment guidance for each: what the learner records instead, and how the assessor tests the same underlying skill. For ZFF-AU-03 the fallback is already present in the unit, since 3.2 requires test cases expected to fail.

ZFF-AI-03 4.1 is a knowledge criterion under a skills outcome. LO4 reads "Be able to interpret and evaluate the output of an AI tool", and 4.1 asks the learner to describe what critical thinking means when applied to an AI output. Describing a concept is knowledge. The other six criteria in LO4 are all acts on a real output.

It is a small thing and it is exactly the kind of small thing an external quality assurer picks up, because the convention that "be able to" outcomes carry skills criteria is well established.

Change proposed Move it into LO1 as a seventh criterion, where the other knowledge sits. The alternative, leaving it and noting the reason in the assessment guidance, is defensible but weaker.

ZFF-GM-03 LO2's stem verb is not discharged by any of its criteria. The outcome is "Be able to compare own current skills with the skills own role will require", and its four criteria ask the learner to identify strengths, identify missing skills, select one and identify sources of help. None of them asks for a comparison. The comparing happens at 5.1, under LO5.

The same pattern appears in a milder form at ZFF-PJ-03 LO2, where no criterion says "plan", although 2.1 to 2.6 taken together plainly constitute planning. We would leave that one.

Change proposed Either restem LO2 to match what its criteria do, for example "Be able to identify the skills own role will require", or add a comparison criterion. Restemming is the smaller change and does not touch a criterion.

ZFF-AI-03's level check claims more than its criteria support. It maps the descriptor clause on being aware of different perspectives or approaches to "1.1 to 1.3, and 1.6". Criteria 1.1 to 1.3 describe how a language model produces answers and why it can be wrong. That is the nature of the area, not a set of perspectives, and 1.1 and 1.2 are already claimed against that clause in the row above. Only 1.6 genuinely answers it.

Change proposed Narrow the mapping to 1.6, and add 2.2 if a second reference is wanted, since identifying a task that should not be given to an AI tool and giving reasons carries a competing approach. The claim is weaker and it is true.

ZFF-GM-03's level check maps "interpret and evaluate" to a criterion that reviews. The check cites 5.2, whose wording is "Review how far the actions taken changed own performance of a task in own role". Review is a different descriptor clause, and LO5 is already claimed against it in the same sentence. Nothing in the unit uses the verb evaluate.

This is the same class of problem as the one already known on the Level 5 page, where a verb is used in one place and ruled out in another. Here it is a mapping the wording does not carry.

Change proposed Cite 5.1, which compares skills at the end against those identified at the start, and 4.4, where feedback is weighed and acted on. If neither is thought sufficient, the honest course is to say the clause is met at qualification level rather than in this unit.

ZFF-GM-03's level check does not map "use appropriate investigation to inform actions" at all. Every other unit maps it: ZFF-AI-03 at 4.3 and 4.4, ZFF-DA-03 at 3.2, ZFF-GV-03 at 4.2, ZFF-AU-03 at 4.3 and ZFF-PJ-03 at 4.2. ZFF-GM-03 is one of the two Award units, so between them ZFF-AI-03 and ZFF-GM-03 must carry the descriptor for the Award, and the omission is visible.

Change proposed Cite 2.4, which requires the learner to identify sources of help inside and outside their workplace. That is investigation informing an action, and it needs no criterion change.

ZFF-AU-03's level check does not map "aware of different perspectives or approaches". The unit does contain the material: 1.1 and 1.2 set out what makes a task suitable and unsuitable for automation, and 2.4 requires reasons for the steps selected. The clause is simply not claimed.

A descriptor clause does not have to be met by every unit individually, only by the qualification. If that is the intended reading it should be stated, because at present five units map the clause and one silently does not.

Change proposed Either cite 1.1 and 1.2 in the level check, or add a line to section 3 saying the descriptor is evidenced across the qualification rather than unit by unit.

Four criteria repeat in substance across units, and one learner could satisfy them all with one piece of evidence. The "different views held by people in own area of work" criterion appears three times, at ZFF-AI-03 1.6, ZFF-GM-03 1.4 and ZFF-GV-03 2.4. The restricted information criterion appears at ZFF-AI-03 2.3 and again at ZFF-GV-03 1.2. Declaring AI use appears at ZFF-AI-03 5.4 and again at ZFF-GV-03 1.5 and 3.3. What happens to information entered into a tool appears at ZFF-AI-03 1.5 and ZFF-GV-03 2.2.

Each is properly placed in its own unit and the subject differs slightly each time. The risk is not the duplication, it is a portfolio in which the same paragraph is submitted three times and passed three times.

Guidance only Keep the criteria. State in the assessment guidance that the same piece of evidence may not be counted against more than one of these, and say what the distinct focus of each is.

ZFF-GV-03 4.1 and 4.2 do not agree about whether something has happened. 4.1 asks the learner to identify a risk arising from the use of an AI tool, which reads as prospective. 4.2 asks them to use appropriate investigation to establish what happened, which reads as an incident after the fact. A risk that has not materialised has no "what happened", and a learner will be told two different things by the two criteria.

Change proposed Align 4.2 to the prospective reading at the next revision, for example establishing how the risk arises and what it could affect. Until then, guidance should say that either reading is acceptable and the assessor should record which one the learner took.

Two units measure time to different standards. ZFF-AU-03 2.5 and 5.1 both require the learner to state the time a task takes "and how that was established". ZFF-PJ-03 1.3 and 5.1 do the same. ZFF-AI-03 5.2 asks the learner to "estimate the time added or saved" with no basis clause at all, which makes it the one time criterion in the Diploma that an assessor cannot test.

ZFF-AI-03 5.2 also sits in the Award, so it is likely to be the first time criterion any learner meets.

Change proposed Add the same basis clause to ZFF-AI-03 5.2 at the next revision so that all four read alike. Until then, guidance should require the basis of the estimate to be recorded.

Three criteria assess an act that leaves no artefact. ZFF-GV-03 3.1 requires a policy check made before proceeding, and the checking is a moment, not a document. ZFF-GV-03 3.2 requires information to be removed or withheld, and an absence cannot be shown. ZFF-AI-03 1.5 requires the learner to describe what happens to information entered into the AI tools their workplace uses, which many learners cannot establish because the vendor does not say and the employer has not written it down.

These are the right things to assess. They are the three where a centre is most likely to accept a bare assertion.

Guidance only Name the evidence route for each. For 3.1 and 3.2, a contemporaneous record made by the learner plus targeted oral questioning in the professional discussion. For ZFF-AI-03 1.5, the tool's own published terms and the workplace policy, with the learner recording where the answer could not be found and what they did about it.

ZFF-GV-03 5.3 judges the workplace's rules rather than the learner's own work. It asks the learner to evaluate how effective the workplace's rules were in the situations met. The unit's own boundary note already says to watch this one, and the unit aim says the learner's duty ends at reporting.

On balance it holds, because "in the situations met" ties the judgement to cases the learner personally encountered rather than to the policy as a whole. That is a review of the learner's own experience of a rule, which is at level. It is a fine distinction and it is not written down anywhere.

No change Keep 5.3 as worded, and state the reasoning in the assessment guidance so that a centre does not read it as an invitation to appraise organisational policy.

Tested and holding. Three things were looked for across all 145 criteria and not found, and they are worth recording as checked.

No criterion asks an assessor to judge an attitude, a level of confidence or a state of mind. ZFF-GM-03 is the unit most at risk of it and its design note is doing real work. No criterion uses analyse, critically analyse, design or reformulate, and the only appearances of those words are in the level checks that rule them out and in the ZFF-PJ-03 aim at item 4 above. Scope stays personal in every criterion, subject to item 1.

A small cluster of criteria are bare acts and would sit below Level 3 if read alone: ZFF-AI-03 3.2, ZFF-GM-03 3.2 and ZFF-PJ-03 3.1 all simply require the learner to do the thing. Each is the load-bearing step that a later criterion in the same outcome reviews or evaluates, which is how a Level 3 outcome is properly built.

No change Recorded as checked.