Philosophy

Every Eduriti product runs on a constrained multi-agent pipeline. Each agent in the pipeline operates within a defined inferencing boundary — it receives a specific input, executes a specific function, and produces a specified output. No agent has open-ended generative latitude. The constraint is not a limitation. It is the design. Intelligence — human or artificial — requires containing structure or it will run away with itself.

Human oversight is not a checkbox. It is encoded into the pipeline structure. Every Eduriti product has mandatory human intervention points — stages where the AI's output is presented to the user for review before the pipeline proceeds. The AI produces a start point. The human determines whether to continue. AI output is never deployed to a learner, client, or investor without a human having reviewed and approved it.

Every product is built backwards from a required professional output — not from what AI can do. The storyboard an instructional designer would approve. The business plan an investor would read. The report a compliance officer would sign off on. The professional standard is the benchmark. The pipeline is the path to it. This means Eduriti products do not generate generically — they generate to a specification.

Eduriti classifies its products as Minimal Risk AI systems under the EU AI Act. The architecture meets the Act's core obligations: disclosure (users are informed they are working with an AI-assisted system at every stage), human oversight (mandatory review gates prevent unsupervised AI output deployment), data minimisation (uploaded content is processed and deleted — not stored, not indexed, not reused), and constraint architecture (agents operate within defined boundaries, not open-ended generation). Eduriti's EU AI Act compliance assessment is available on request.

Eduriti collects only what is necessary to deliver the service: an email address, a transaction reference, and the content you submit for processing. Uploaded files are processed and deleted. They are not retained, archived, or reused. Generated outputs are stored for 30 days and then permanently deleted. No user profiling. No behavioural tracking. The system processes what you give it and returns what you need.

The name encodes the operating principle. Edu meets Riti — रीति, Sanskrit for method, practice, and established order. A shortcut gets you somewhere faster. A method gets you somewhere right. Eduriti does not automate away professional judgement. It automates the process that enables professional judgement to operate at a higher level.