From your own material to training: onboarding, guidelines and policies as six-minute lessons
Bring your onboarding folder, guidelines and policies to the Materials page, read the analysis and make a course in the Studio the same day. AI writes the draft, you check it.
· 6 min read
In brief
An organisation's own guidelines are training material nobody reads: the onboarding folder, the security policy, the customer service guide, the procurement instruction. In CoreSkills they are brought to the Materials page, analysed and turned in the Studio into six-minute lessons with questions and, if wanted, a video. AI writes the draft from your own material, the quality check finds the errors, and you check and publish. A course is made within the same day, and it shows to the staff under your organisation's name. Material is not given to the AI without permission and never used to train models.
The Materials page: documents in at once
The organisation's own documents are brought to the Materials page by dropping files or a whole folder onto it, or by choosing them. A zip archive is unpacked. Readable formats are PDF, Word, PowerPoint and text files. Images and tables are skipped and named, and a scanned PDF whose text is an image is reported as having no text found. The same document does not enter the list twice.
At upload you decide whether the material may be given to the AI. The permission covers the whole batch, and it can be changed from the list one document at a time. A document without permission shows in the list but ends up in no course, no analysis, no search and no staff Ask the content conversation. Clicking a document's name shows the text as the AI reads it.
Permission to give material to the AI is not the same as permission to copy it. A purchased report or one behind a paywall is not given to the AI unless its licence expressly allows it, and each source is described for what it is: your own guideline, a statute or a licensed publication. The AI writes the course only from material it may be given.
Analysis before courses
Before making courses, the material can be analysed. The analysis reads the documents one by one and takes from a few minutes to half an hour depending on the amount, and the report is written for the administrator who decides which courses to make.
The report gives the material's condition document by document: in order, needs work or out of date, and why, for example a reference to a system that no longer exists or a guideline that names no owner. It says which subjects the material covers and which of them a course already teaches, and what is missing against the sector's obligations, with the basis of each obligation. It finds overlaps and contradictions, that is what two documents say differently. And it recommends courses in order of importance: a course of your own from the material, when the documents hold rules the staff must know and no course yet teaches them, or a library learning path, when the material touches a subject the CoreSkills library already teaches properly.
The analysis changes nothing. It makes no courses and assigns no paths; the decisions are yours. The report is a good starting point, not an audit: if a guideline is up to date although the report doubts it, that is exactly the knowledge the AI does not have.
The Studio: six steps to a course
In the Studio a course is made in six steps.
- Subject. Say what the course teaches and for whom, and choose the course's language, sector and format: a single video, a video with questions, or a module. The sector steers the examples: in a care course the example comes from care work, not from an office.
- Source material. Choose documents from the materials or add web pages, files or provisions of statutes. The course refers to a document without copying it, so the same guideline can underlie several courses.
- Outline. The AI proposes an outline of lectures, which can be edited.
- Content. The AI writes the lectures, the questions and the videos' narration scripts.
- Check. The content is gone through before videos are made, and every step can be edited. The quality check finds errors and ambiguities and proposes corrections, accepted one by one. It also shows the passages where the text repeats a source verbatim: such a course cannot be published until the passages are rewritten in your own words, except when the source is a statute or your own material. In the same place you mark which skills the course accrues to.
- Publication. The course is published at once to your own organisation only, and it shows to the staff on the Courses page under your organisation's name, with its sources at the end as they were added.
A lecture can carry extra material from outside the service, for example an authority's YouTube video or a guidance page, and an external video can also be added as a step of its own that counts towards completion. Unfinished work is saved and can be continued later.
Several at once: request courses
When the analysis has recommended several courses, they can be requested together. CoreSkills checks the request, writes the courses from the named documents and runs them through the quality check. The finished courses arrive on your Approvals page, where they are approved or sent back with a request for changes, and the request says what has to change.
The organisation's profile has a setting, Publish as AI-checked. When it is on, a course produced for you is published as soon as the automated check passes, and the course says AI-checked; once a person has later read and approved it, the label changes. By default the setting is off, and a course waits for your approval before publication.
What an own course becomes
Own courses show to the staff at the top of the Courses page as a group of their own, and they accrue to the organisation's own learning paths. From them you can build a path of your own, certifying or not, and assign it to a job role with a due date just like a library path. Onboarding that used to be a folder is then a path that every new employee gets at their first sign-in and whose completion shows in the report.
The same material serves the staff in a second way. In the Ask the content conversation an employee can ask in their own words what the onboarding guide says about remote work or whom to report a security breach to, and the AI answers only from the permitted documents, the courses in use and the statute library, with numbered sources. If the material says nothing on the matter, it says so rather than guess.
Where the material is and where it goes
The material is in CoreSkills' own cloud service in Finland, and it is visible only to your own organisation. The AI is given only the text of permitted documents as it is, without the administrator's name or email, and it is not used to train models. A document's text is not anonymised, so a document containing personal data is best left without permission or cleaned before upload. If the organisation has its own Azure OpenAI, course production and the analysis run through it, and the cost shows on your own Azure bill. How the data is protected is covered in Where your data is.
Frequently asked questions
What material can be brought into the Studio?
PDF, Word, PowerPoint and text files, one at a time, as a folder or as a zip archive. Images and tables are skipped and named, and a scanned PDF whose text is an image is reported as having no text found. During the trial the materials hold 20 documents, on a subscription 300.
What does the materials analysis tell you?
Document by document, whether it is in order, needs work or is out of date, and why; which subjects the material covers and which of them a course already teaches; what is missing against the sector's obligations; what two documents say differently; and recommended courses in order of importance. The analysis changes nothing and makes no courses.
Is our material used to train AI models?
No. Material is not used to train models, and it does not go to a service that would store it. A document is given to the AI only if permission was given at upload, and the permission can be changed document by document. If the organisation has its own Azure OpenAI, course production and the analysis run through it.
Does someone check a course before publication?
Yes. Every step can be edited, and the quality check finds errors and ambiguities and shows the passages where the text repeats a source verbatim. Such a course cannot be published until the passages are rewritten in your own words, except when the source is a statute or the organisation's own material. Courses CoreSkills writes wait for your approval on the Approvals page, unless you have switched on publishing as AI-checked.