Cogniti and Ulern, side by side
Cogniti lets teachers build their own AI agents and put them inside the learning management system, and its 2026 symposium has talks from educators at 24 institutions. If your school already uses it, the question is fair: what would Ulern add? We build Ulern, so read this as a vendor’s comparison. Everything said about Cogniti here comes from its own website and documentation, read on 13 November 2026.
What Cogniti is
Cogniti is built by a team at the University of Sydney, by educators for educators. A teacher creates an AI agent and steers it in plain language: be a Socratic tutor, do not give the answer away, take a part in a role-play. The teacher gives the agent resources, such as files and web pages, so its answers stay close to the material. Agents can be embedded in Canvas, Moodle, D2L Brightspace and Blackboard, so students meet them inside the LMS they already use.
It goes further than a chat window. Teachers can build interactive mini apps and set up sandboxes where students build agents of their own, and educators on its home page describe simulations, feedback coaches, diagnostic tools, role-plays and games.
The teacher who made an agent can read a de-identified list of the conversations students have had with it, and ask AI to summarise them, to spot misconceptions and check the quality of the answers.
Cogniti runs on Microsoft Azure and can use models from OpenAI, Anthropic, Google, Mistral and others, switched on or off per institution. A school can take it as a service hosted by the University of Sydney in Azure’s Australia region, or install it in its own Azure environment through Microsoft Marketplace and choose the region. Prices are not published: a tiered fee goes to the University of Sydney, Azure usage is added on the Marketplace route, and small free pilots may be available.
Its strength is range. A teacher with an idea for an AI activity can usually build it in Cogniti without waiting for anyone.
What Ulern is
Ulern is the applied-learning platform for business schools and executive education. Participants learn from your material with an AI tutor, then make a real deliverable under your method, and a supervisor decides at every gate. Two parts, on one path.
The learning. Your material becomes a knowledge base of topics and notes, and each note cites the passage it was written from. A supervisor can open any note, follow it to its source, correct it or add to it, and choose to keep the AI tutor strictly inside those notes. The AI tutor teaches by having the participant do things. It explains an idea once, then asks them to use it, often in a situation like their own. Some questions ask them to choose or sort, others to write an answer out, which the AI tutor reads and replies to. Every answer counts as evidence, and a topic moves up only when the evidence says so (how that works).
The making. The supervisor sets the method once: the steps of the capstone, thesis or applied project, what each step must contain, and where a decision is needed. The participant writes every step. The AI tutor reads each draft against the criteria, says what is missing and asks for a revision. It never writes the work; when the finished deliverable is assembled, any joins it smooths are shown to the participant to approve. The supervisor sees who is waiting at which step, then approves the step or returns it with a reason. The deliverable exports in Word, PDF or Markdown, with the record of every draft, review and decision behind it (a capstone, walked through).
Ulern is hosted in the EU, and its price is published: €6,000 a year covers up to 100 people, €12,000 up to 250, €20,000 up to 500, larger quoted. A 90-day pilot for one programme and up to 30 people is €2,000, credited in full when you continue within 30 days of its end.
Side by side
| Cogniti | Ulern | |
|---|---|---|
| Built around | AI agents and activities a teacher designs | A project with steps, criteria and a supervisor’s decision at each |
| Who shapes the AI | The teacher, in plain-language instructions to each agent | The supervisor, through the method, the criteria and the knowledge base |
| What the AI works from | Files and web pages, searched for relevant passages or read whole | A knowledge base of topics and notes, each note citing its source passage |
| What the participant does | Talks with agents and uses the mini apps a teacher built | Practises each idea, then writes a real deliverable step by step |
| What remains afterwards | The conversations, de-identified for the teacher | The deliverable, the record of how it was made, and evidence per topic |
| Hosting | Microsoft Azure: Australia as a service, or your own Azure | The EU |
| Price | Not published | Published, one price a year |
Where each fits
Cogniti fits a teacher who wants to build an AI activity of their own: a Socratic tutor for a module, a role-play, a feedback coach for an assignment. It lives inside the LMS the school already runs, and the teacher writes every instruction.
Ulern fits a programme that ends in work someone has to supervise: a capstone, a thesis, an applied project in the participant’s own company, assignments between modules, or a custom programme for a corporate client. There the teaching and the making have to be one path, every lesson has to trace back to its source, and the supervisor has to see the work and decide.
Neither comes free beyond a pilot. A good AI tutor costs money in licences, hosting, AI usage and the time it takes to build and keep up, whichever way a school goes. With Ulern, the AI tutor, the knowledge base built from your material, the projects and the supervision come in one price a year. Many schools will run both: Cogniti for the AI activities teachers build into their modules, Ulern for the programmes that end in a project.
What to ask any AI tutor
- Where do its answers come from? Can a supervisor open the source behind a lesson, and correct it?
- Does it make the participant use the idea, or mostly explain it?
- Who decides when a piece of work is good enough, and is that decision recorded?
- What is left at the end that a supervisor, an accreditor or a corporate client can read?
- What does it cost all in, including hosting and AI usage, and where is it hosted?
Where Ulern stands
We build Ulern, the applied-learning platform for business schools and executive education, so read this section knowing that.
We see Cogniti as a good tool for teachers who want to build their own AI activities, and we would not ask a school to give it up. Ulern is for the part it was not built for: a programme’s project, from the first lesson to the supervisor’s last signature, on your material and your method. If anything about Cogniti here is out of date, write to us and we will correct it.