AI Academic Document Generation Platform with Custom Engines

This platform turns a university assignment brief into a near-finished draft package: a Word document, a slide deck, a defense speech and A1 engineering drawings. The user does the final pass; the system does the other 99%.

It is not an API wrapper. The core is a custom MCP server driving a three-wave parallel agent swarm, wrapped in an enforcement layer that physically prevents low-quality output from shipping. Twelve automated validators check citation coverage, block-formula preservation, numeric consistency and even physical plausibility of calculated magnitudes. Stage flags are sha256-signed, so the model cannot raise its own page limit to sneak an oversized document through, and builder markers stop it substituting a hand-rolled script for the official document builder.

Two engines were written from scratch: a presentation engine with a constraint solver (18 layouts, 12 themes, WCAG contrast validation) and a CAD engine that emits DXF/DWG drawings with a GOST 2.104 title block generated in code after 18 vectoriser iterations.

Live in production on a dedicated server with roughly 2,960 unique visitors a month from pure organic search and zero ad spend, the largest share of requests coming from the United States. 33 orders were delivered end to end, 867 MB of finished work on disk, which proves the pipeline works rather than just demos.

Pre-revenue and stated as such: all five payment integrations are written and tested but no live merchant keys are attached, so the buyer connects their own accounts. The pricing engine is already calibrated across 24 work types and 100+ disciplines, so the commercial model is ready the moment a merchant account is plugged in.