Straits Institute for Applied AI
Assessment & Certification · The SIAA Standard

The SIAA Standard tests how well people apply AI to real work.

The SIAA Standard is the assessment system for applied AI skills. Candidates complete realistic work tasks from their own profession with a live AI, and we mark both the results and the human judgement and decisions behind them.

The SIAA Standard candidate screen: a workplace task on the left, a live AI assistant on the right
What a candidate sees: the task and its documents, with a live AI beside them.
What we assess

We mark how people create value with AI, and how they manage its risks.

In every test we mark six things: three that measure whether a person can create economic value using AI, and three risks they must manage.

Creating value · 1 of 6

Acceleration

Can they do the work they know, better and faster?

How we test it

Routine work they know well, as one task or as requests that keep arriving.

Tested from Proficient

How our tests are designed

Each test is real work from the person's own job, done with a live AI.

01

What employers actually look for

Employers want people who create value with AI and use it safely, so we test both. Value means working faster, taking on new kinds of work and running AI agents. Safety means keeping data safe, catching the AI's mistakes and handing over work others can rely on.

02

Real scenarios from the job

Every question is real work from the person's own job. It comes with the documents, names and numbers they would have at work.

03

AI must not replace human judgement

Every question needs the person's own judgement: what to give the AI, what to check and what to send. Before we use a question, we give it to the AI on its own. We use it only if copying the AI's answer scores clearly lower than careful work.

04

We mark the thinking behind the answer

An AI gives a different answer each time it is asked, so comparing final answers alone is not fair. We mark how each person worked: what they asked the AI, what they checked and what they fixed.

Set in each person's own work

Uniquely Versatile: We cover 163 professions and 96 work skills, across 20 industries.

Each assessment is set in the person's own profession and uses the work skills that job needs, so a nurse, a civil engineer and a banker each get questions from their own work.

Questions are built from the profession books of the AI+ Library.

The tests

Tests for hiring, for whole teams and for a certificate.

Who should I interview?

Candidate Screening Test

About 30 minutes · job candidates

Every candidate for a role sits the same 30-minute test, written for your role, industry and work skills. You get a report on each one. It says whether they are likely to meet the Proficient standard, shows the evidence from their own work, and gives questions to ask at interview.

Where do my people stand?

Workforce Assessment

About 60 minutes · staff, civil servants, students

Each person sits a 60-minute test on their own kind of work. You see results for every person, team and department in the six competencies. Test again after training to measure the change.

Can I prove my level?

SIAA Standard Certificate

Proficient, Advanced, Expert

Proof of a level that any employer can check. There are three levels. Each has its own test of real work with a live AI, harder at each level. Safety, Reliability and Handover are required at every level.

An SIAA Standard Test Report: overall result, area profile and area scores for a graded test
Every person gets a report HR can act on. The top of a sample report from a graded test.
The research

Backed by academic research.

Our paper sets out how to test a person's use of AI when the AI could do the task alone. Before anyone sits a question, we give it to the AI on its own, and rewrite any question the AI can pass by itself.

The paper: Measuring the Human Contribution When AI Is Allowed: Psychometric Foundations for Work-Sample Tests of AI Use. Seng Beng Ho, CH Peh, Roger Simões, Shawn Chia, Elijah Ng, Jennifer Dodgson.

3,906
Times we gave our questions to the AI alone, in October and November 2026: pasted whole, in pieces or reworded.
1 in 155
Copied AI answers that reached a pass in our grading tests. The one that passed was on a question since rewritten.

Start with a pilot in one team.

A pilot takes about eight weeks. You see the test working with your own people before you decide on more.