Saval vs Cogniti AI: Comparing AI-Powered Learning Platforms

As AI-powered platforms multiply in higher education, it's worth being precise about what each one is actually built to do — because two tools that both say "AI" and "learning" can be solving genuinely different problems.

What Cogniti does well

Cogniti, developed at the University of Sydney, lets educators build course-specific AI agents that act as subject tutors, offering personalised feedback on student work while simulating scenarios and guiding students through complex problem-solving steps. It's been used for genuinely inventive applications — from a marketing agent modelled on an industry professional to psychology students practicing risk-assessment conversations with a simulated patient. Educators shape each agent's instructions and source material themselves, and the platform is built to keep student data within institutional control, which has made it attractive to universities wary of routing student conversations through third-party consumer AI tools.

In short: Cogniti is a toolkit for building customised AI tutoring and simulation agents, put directly in the hands of individual educators.

What Saval does well

Saval starts from a different question — not "how do we let educators build an AI agent," but "how do we structure formative, gamified, authentic assessment so it holds up as generative AI reshapes what a valid assessment task looks like." That means Saval is built around:

• Gamified formative assessment designed to sustain engagement through genuinely challenging, reasoning-based tasks
• Scenario-based authentic assessment across disciplines like engineering, business, and healthcare
• Learning analytics that surface why a student is struggling, not just whether an answer was correct
• Direct alignment with the UNESCO AI Competency Frameworks, built into the assessment design itself

Where they diverge

Cogniti's unit of work is the agent — an educator configures a tutor, simulation, or coach, and students interact with it conversationally. Saval's unit of work is the assessment task — a structured, gamified, often scenario-based activity with built-in formative feedback and analytics designed for tracking progress over time. Cogniti is closer to a flexible AI-tutoring layer that sits alongside teaching; Saval is closer to an assessment and analytics layer with gamification built in from the ground up.

Cogniti's simulation strength is particularly notable — it's been used for high-stakes practice scenarios like a deteriorating patient simulation where a nursing student must correctly diagnose the problem and request appropriate intervention, with strong student reception. Saval's strength is different: turning everyday coursework into structured, replayable, formatively assessed experiences that generate ongoing analytics an institution can act on.

Which one does your institution need?

If the priority is giving individual educators a low-code way to build custom AI tutors and realistic simulations for their own courses, Cogniti's agent-building model is purpose-built for that. If the priority is embedding gamified, authentic, UNESCO-aligned formative assessment across a program — with the analytics to back instructional decisions — that's the problem Saval was designed to solve.

The two aren't mutually exclusive. Increasingly, institutions are finding that the more interesting question isn't "which one," but how an AI-tutoring layer and a gamified formative-assessment layer can complement each other across a student's learning journey.