Make AI useful within your institution
LAMB brings generative AI into the institution’s learning environment. Teachers and students use assistants and activities through the LMS, while the institution manages access, model providers and the conditions under which course data is processed.
It is free software under GNU GPL v3, with source code available for inspection, adaptation and sharing under the license. LTI connects activities to LMS platforms such as Moodle and Canvas. The optional Moodle connector also gives LAMB LEGATUS, the agent for teachers, access to permitted course content and activity through standard web services, without installing a Moodle plugin.
Three ways to begin
Choose the next step that fits your team.
Train your people
For educators, developers and administrators, and on AI governance, from the LAMB team.
About training ↗02 · START YOURSELFInstall LAMB
Install it on your own servers with the step-by-step guide and prepare a teaching pilot.
Installation guide ↗03 · CONSULTANCYPlan your adoption with us
A demonstration, pilot design, deployment planning and AI governance advice. We also give keynotes and seminars.
About consultancy ↗The last mile for your AI
Your institution may already run its own models or use a provider that meets its requirements. LAMB gives each student access to them under rules set by the school and the teacher: teacher allowances, per-student limits, timetables and a personal key whose use is counted. In 0.8 this covers LLM access activities, for students’ own tools. The last mile for your AI →
Governance and provider choice
Decide who can create and use AI activities, which generation and embedding services are approved, and how content and conversations may be accessed. These choices make LAMB part of the institution’s privacy, GDPR and AI-governance work. Regulatory compliance depends on the institution’s policies and deployment, as well as the software configuration.
Choose from supported providers using self-hosted inference, cloud services or a combination appropriate to your policy. Hosting LAMB and hosting the models are separate decisions; make both explicit.
Privacy, GDPR and institutional control →
What an institution can provide
| Need | What LAMB contributes | What your institution decides |
|---|---|---|
| Educators need course-specific assistants | An interface for instructions, materials and testing | Who creates assistants, which sources are allowed and how quality is reviewed |
| Students need access in their course | LMS integration paths | Supported LMS/LTI configuration, enrolment and support |
| Teams need a choice of AI services | Configurable model connections | Local or external providers, approved data, spending and service agreements |
| Students need model access for their own tools | LLM access activities: personal keys with limits, timetables and usage records | Which models, how many tokens per teacher and per student, and when |
| Administrators need an operated service | Organization and user administration | Hosting, backups, monitoring, updates and operational ownership |
| Innovators need a starting point | Public API and educational tool resources | Scope, deployment responsibility and acceptance of each application |
Capabilities differ by installed version. See availability before treating a demonstration as a deployment commitment.
Begin with a bounded pilot
Choose one course, a small teaching team and a clear learning purpose. Name the academic owner, the LMS contact and the system operator. Agree what data can be used, how model costs will be covered and how students will be informed.
Before inviting students, test the complete path: teacher access, document preparation, assistant behaviour, student launch and support. Where an activity returns a grade, test the displayed scale and gradebook result as well as the technical connection.
Evaluate the pilot using questions you can actually answer: Did the assistant help with the selected task? Which responses failed? Could the teacher correct the problem? Was the service supportable? Usage counts alone do not establish learning benefit.
Privacy is a deployment decision
Self-hosting LAMB does not by itself keep every model request on campus. External generation and embedding providers may process submitted content. Conversational access and retention also need an explicit institutional policy.
Questions for your privacy and security review →
Move to the right next step
- LMS team: review the publishing and course-access path.
- System operator: plan operation and deployment.
- Teaching lead: prepare a first assistant.
- Application team: explore the developer route.