MedTech

An LMS for Medical Device Companies

A medical device company trains the people whose work affects product quality, and it has to show who was trained on which procedure, and when.

Syllabium is an enterprise learning management system (LMS) for regulated industries, made by Athantic AI LLC. It keeps courseware in versioned libraries, pins every entry of a curriculum at a version, holds safety-critical changes until a second person approves them, and keeps tamper-evident training records that an administrator exports for an auditor. In Syllabium, MedTech is the field for clinical training, and its AI prompts and its reflection risk screen are written for it.

What does Syllabium do for medical device companies?

Syllabium gives a medical device company one system for its courseware, its learners and its training records, with every change to content passing one publishing gate and every AI action recorded. Authors build courses in Syllabium Studio, instructors run cohorts in Syllabium Cohorts, learners study and are assessed in Syllabium Learn and Syllabium Assess, and Syllabium Evidence keeps the record.

What the company needsWhat Syllabium does
Train people on the current version of each procedureContent is authored once in a library with releases, and each curriculum pins every entry at a version and is told when a newer one exists.
Change content without losing its historyEvery change to library content passes one publishing gate, which publishes, holds or refuses it and records the decision.
Keep AI from changing safety-critical content unseenOn content flagged safety-critical, an AI agent’s work waits for a person, and no setting changes that.
Show that training was completed and passedEach assessment attempt keeps the questions as asked and the pass mark applied.
Hand records to an auditorAn administrator exports one archive for a period, holding each record, a verification report and a list of file fingerprints.

How does Syllabium serve ISO 13485 and Part 820?

ISO 13485 and 21 CFR Part 820 set a medical device manufacturer’s own quality-system duties, and Syllabium does not claim conformance to either. ISO 13485 is the international standard for a medical device quality management system. 21 CFR Part 820, in Title 21 of the United States Code of Federal Regulations, is the quality system regulation of the United States Food and Drug Administration (FDA), and it now incorporates ISO 13485 by reference. Both ask a manufacturer to train the people whose work affects quality, to judge whether the training worked, to keep records of it, and to control the documents that training teaches. Syllabium’s versioned courseware and tamper-evident training records serve that training, while the procedures, the decisions and the quality system stay the company’s own.

What a quality system asks of trainingWhat Syllabium keeps
Teach the current, approved version of a documentA curriculum pins each library entry at a version, and learners stay on their version until the next one is published.
Review and approve a change before it is issuedA change to safety-critical content stays hidden from learners until a second person approves it, and the person who submitted it cannot approve it.
Keep assessments in step with the contentA verified question goes stale, and is no longer drawn, when a slide it cites changes.
Record who was trained, on what and whenLearning activity is kept as a tamper-evident chain of records, and a learner’s lessons, attempts and surveys export as a CSV file.
Judge whether the training workedPass marks are set by policy, and each attempt keeps the questions as asked and the pass mark applied.
Keep records retrievable, with any change visibleEntries are added, never edited, and changing, deleting or reordering one makes every later fingerprint stop matching.
Keep records for a set periodRetention periods are rules: most records are kept seven years, and a nightly purge records what it removes.

What is planned for Part 11 and Annex 11?

21 CFR Part 11, the FDA’s rule for electronic records and electronic signatures, and EU Annex 11, the European Union’s guidance on computerized systems, are planned targets on Syllabium’s roadmap. The ten standards Syllabium is built to are listed on Standards.

Syllabium’s own records are tamper-evident: each entry carries a fingerprint of itself and of the entry before it, so a change, a deletion or a reordering shows. Every chain is recomputed nightly and on demand, and once a day the newest fingerprints are written to write-once storage outside the database, locked for seven years. The application cannot touch these records directly, only through fixed functions. Evidence describes these records in full.

Which Version Does Each Curriculum Teach?

Each curriculum pins every library entry at a version and is told when a newer one exists, and learners stay on their curriculum’s version until the next one is published. When a library releases a newer version of an entry, each curriculum that draws on it is told, and the curriculum’s owner decides what happens next.

ChoiceWhat happens
Accept the updateThe curriculum takes the newer release.
Defer itThe curriculum keeps the version it has.
Compare versionsThe two releases are compared block by block.
Take a local copyThe curriculum keeps its own copy, which can later be reconnected or promoted into the library.
Freeze the curriculumThe curriculum stays as it is, and overriding the freeze takes a reason.
Start its next versionThe next version is prepared while learners stay on theirs until it is published.

Questions in the question bank cite slide text word for word, each with an objective, a thinking level and a difficulty. Libraries are released, cloned, frozen, archived and rolled back, and every closed decision on a held change downloads as CSV, Markdown or PDF. Courseware describes the libraries and the studio in full.

How Is Safety-Critical AI Content Signed Off?

A person flags content safety-critical, and from then on a change to it stays hidden from learners until a second person approves it; an AI agent’s work there always waits for a person. No setting gives an AI agent a plain “allow” on safety-critical content, and the rule is enforced in the code and in the database. An administrator can switch the second person off for people’s own changes, which itself needs a second administrator, and the switch never applies to AI. Held changes are compared word by word, and each is approved, rejected with a reason, withdrawn or revised. In the studio, the content tree marks safety-critical content, AI activity and other people’s changes.

The AI assistant beside the editors works in the mode each author chooses: Off, Ask, Suggest or Act, with Suggest as the default. It works on text only, and it does not publish content. Each request is a defined action on a named piece of content, never a free chat, and every model call is logged with its model, tokens and cost. No person’s identity is sent to a model: a database rule refuses identity data for every provider. Syllabium uses Claude models, through Amazon Bedrock and Anthropic's API. An organization can also bring its own model (BYOM). AI review before publish, Governed AI and Your own models describe each part.

What Can an Auditor Export?

An administrator exports one archive for a period, holding each record, a verification report and a list of file fingerprints, and every export is itself logged.

RecordWhat it holds
Learning activityA tamper-evident chain of each learner’s activity
Assessment attemptsThe questions as asked and the pass mark applied
Publishing decisionsEach decision of the publishing gate on a change to library content
AI actionsEvery attempt with the check that decided it, and every model call with its model, tokens and cost
ConsentThe versioned terms, privacy notice and AI notice, whose published words cannot change, and each acceptance
Sign-insEvery sign-in, failure, lockout, sign-out and password change
Readings of a person’s recordEach reading or export by someone else, shown or refused
AI access reviewsEach review of AI agents’ permissions, due every ninety days

Instructors also export a learner’s transcript of lessons, attempts and surveys as a CSV file, and each export is recorded. Retention periods are rules: most records are kept seven years, and the sign-in and read records two. A person’s words can be erased while every fingerprint still verifies, and the erasure is itself recorded. Training records for audits explains what an auditor asks for, and Evidence how the records are kept.

How Are Organizations Kept Apart?

Syllabium has three levels: the platform, the tenants on it, and the organizations inside each tenant. A tenant is the owning company, licensed for a number of organizations. The database itself lets a session read only the platform’s shared rows and its own tenant’s and organization’s rows, and write only at its own level. Each organization owns its content and works in its own field, set when the organization is created, and a field grants no access. Institutions that send people through an organization’s training, such as hospitals and clinics, have client accounts inside it. A library’s published releases can be shared with another organization, read-only or read-and-copy, and each share is recorded and revocable. Any read across tenants by Athantic AI LLC is read-only, per session, and carries a recorded reason. Platform shows the three levels and the eight products, and Regulated industries the other fields.

What Do Learners and Instructors Use?

WhoWhat Syllabium gives them
LearnersA dashboard and lesson reader, lessons released on the cohort’s schedule, assessments, practice questions after a lesson, and reflections
Instructors and supervisorsRosters, schedules, a progress grid and results for the cohorts they staff, and a facilitator guide with a note beside each slide that learners never receive
AuthorsVersioned libraries, a slide canvas, a curriculum composer and a question bank
AdministratorsUsers and roles, capability grants for people and AI agents side by side, publishing workflows and the AI controls

Assessments draw their questions for each attempt, shuffle the options, limit attempts and apply a pass mark set by policy, and a timed assessment saves answers and closes on time, with extra time as an accommodation. Flagging an integrity concern, resetting an attempt or clearing it for credit each takes a recorded reason, and nobody can do it on their own attempt. In MedTech, each submitted reflection is screened against risk patterns written for clinical training, and a match goes by email to the learner’s named supervisor or to escalation staff; the screen matches word patterns and involves no AI model.

Get Started

Enroll your organization online, see the pricing, or talk to us. Syllabium is made and operated by Athantic AI LLC.

Contact us