Use case · Custom Skills Assessment
Build role-specific assessments from your work
Off-the-shelf tests miss your workflows and tools. Anthropos builds assessments from your documents. Your experts create and own them in Studio without waiting for engineers.
“Your library doesn't test what we actually do.”
Head of L&D, European industrial group
Library vs custom
What changes when you own the assessment
A generic test measures a job title. Your business does not hire job titles.
| An assessment library | A custom assessment in Anthropos | |
|---|---|---|
| What it is built on | Generic questions written for a job title | A real moment from your business |
| Content source | The vendor's catalogue | Your documents, products and processes, up to 40 pages |
| Who authors it | The vendor's content team | Your own subject-matter experts, as Knowledge Owners |
| What it measures | Answers to standard questions | Behaviour across voice, chat, code and document work |
| Scoring | A vendor rubric you cannot open | Your criteria, must-pass checks, 0-5 competency |
| Changing it | A change request, a fee and a wait | An edit your role owner makes the same day |
What it is built for
When off-the-shelf does not fit
Six situations where a generic test measures the wrong thing, and the assessment has to be yours.
Start from the role
Every build starts from a role, not a blank page
Pick the role and Studio starts from the skills that role is meant to have, drawn from the taxonomy. Pick the mission and it tells you which of those skills this one will actually test, before you build anything.
From there, describe it yourself in advanced mode or let Studio draft it and refine what it wrote. Both routes end at the same simulation, scored against the same role.
Engineering Manager, ten core skills, and a mission that tests three of them, named before the build starts. Anthropos Studio.
Source material
Your documents shape each simulation
Anthropos uses your SOPs, product sheets, call scripts, policies and code, with up to 40 pages per simulation. AI characters reference your documents, use your terminology and behave as they would in real situations. Describe a business moment and play it that day.
A scenario built from a customer's own documents. Cervato Systems, our demo org.
Scoring
Your criteria determine assessment scores
The scoring model provides rigour. Each assessment scores named taxonomy skills against your criteria, using binary must-pass checks and the same 0-to-5 competency scale as every other assessment. You can inspect criteria, see what they reward and defend results to assessed people.
A scoring criterion, open and editable. Cervato Systems, our demo org.
FS Group chose Anthropos for the ability to personalise everything: taxonomy, simulations and evaluation criteria, with its own knowledge owners authoring for their roles. 500+ people in two weeks.
Ownership
Knowledge Owners build and update assessments
Anthropos trains one subject-matter expert per role as Knowledge Owner. Each builds assessments in Studio and trains their department. They swap documents and re-publish after changes, without vendor change requests or releases. FS Group piloted 12 roles with individual owners.
The catalogue a company ends up owning, built and kept current by its own Knowledge Owners. Cervato Systems, our demo org.
Beyond the assessment
The same builder makes more than a test
Your experts describe a real moment once. Studio turns it into whichever exercise the situation needs, against the role it belongs to.
Technical roles
Technical assessments are built the same way
Custom does not stop at conversation. The same builder produces a coding task with your own starting repository and your own tests, inside the same scenario as the call, so an engineer is judged on the work and on how they explained it.
One session, both halves of the job: the call on the left, the coding task and its passing tests on the right. Anthropos AI Simulation.
AI Labs
And the Lab runs on your repository
When a coding task is not enough, the same ownership extends to an AI Lab. You set the starting repository, the task and the criteria it is marked against, and the person works in a full editor with a coding agent, exactly as your engineers already do.
The session is scored by the same engine as every other assessment, so a Lab result sits next to a simulation result on one profile, on the same scale.
A Lab in progress: the repository open, the agent working in the terminal, everything recorded for the score. Anthropos AI Labs.
Your material
What happens to the documents you upload
The whole point is putting your own material in. That makes where it goes the first question worth answering.
Uploaded documents are used only in your organisation's assessments, either as task documents the person works with or as background only the AI characters can see. Up to 40 pages per simulation.
Nothing you upload, and nothing anyone produces inside an assessment, is used to train, retrain or fine-tune any model, ours or a sub-processor's. That is contractual in the DPA.
The scenarios your Knowledge Owners build are yours. They are not added to a shared catalogue and no other customer sees them.
ISO 27001 and GDPR, with the EU AI Act assessment available to read. Every scoring criterion is open, so a result can be explained to the person assessed.
Rollout
What it takes to start
Describe the moment you want to assess, attach a document, and preview a playable custom simulation.
Your subject-matter experts are trained as Knowledge Owners and build the first assessments for their own roles.
Knowledge Owners have trained their departments, and the catalogue you own covers the roles that matter.
IT involvement is close to none: business teams upload documents and build in Studio themselves, with IT needed only for single sign-on and the optional HRIS connection.
Objections
What buyers ask before a pilot
Can we build custom simulations for our specific roles and scenarios?
Yes. Custom AI Simulations are built in Anthropos Studio from a plain-language description of the scenario plus your own documents. You control the company context, the characters, the task sequence and the evaluation criteria. Any role you can describe can be assessed, including roles no assessment catalogue covers.
Do we need an instructional designer to build a custom assessment?
No. Anthropos Studio is built for HR, L&D and subject-matter experts rather than content specialists. You describe the situation and attach a document; Studio drafts the scenario, characters, tasks and score checks, and you edit anything you disagree with. The judgment about what good performance looks like stays with your expert.
Can we customise an existing library simulation instead of starting from scratch?
Yes, and most organisations do both. Anthropos has 300+ ready-made simulations across technical, commercial, operations and HR roles. Import any of them into Studio and rewrite the company context, terminology, character personalities and evaluation criteria so the scenario reflects your business rather than a generic one.
What happens to the documents we upload into an assessment?
Uploaded documents are used inside your own organisation's simulations only, either as task documents the person works with or as background knowledge only the AI characters see. The limit is 40 pages per simulation. Anthropos is GDPR compliant and your content is not shared into the public library.
Can a custom assessment run in more than one language?
Yes. Assessments are authored once and delivered in the languages your teams work in, so a custom scenario written by one Knowledge Owner can be run by colleagues in other countries. Scoring uses the same criteria and the same 0-5 competency scale regardless of the language it was played in.
Does a custom assessment feed the same skills data as the rest of the platform?
Yes. Skills in a custom Anthropos assessment are selected from the same 4,000+ skill taxonomy across 700+ roles. Results update each person's verified skill profile and flow into workforce intelligence, gap analysis, AI Readiness and development recommendations. No manual mapping is needed to connect a custom build to the rest of your data.
Who in our organisation can see and edit the assessments our experts build?
Custom assessments live in your organisation's Studio catalogue, visible only inside your tenant. HR and people ops manage the full catalogue and assign content to roles, while Knowledge Owners maintain the assessments for the roles they own. Nothing you build is exposed to other Anthropos customers.
Keep reading
Where to go next
Build one on your own work
Bring a document and a role. We turn a real moment from your business into a playable assessment and score it against your criteria.