PROJECT DESCRIPTION
WorkenAI is an enterprise AI platform that brings any AI model, a company’s own documents and full control of cost and access into one governed workspace. It is built for teams that must answer for where company data goes, what AI costs and who used which model. Everything is processed in the EU, with an on-premise option where nothing leaves the network, no training on customer data, and a full audit trail behind every request.
Solution
We designed and developed WorkenAI end to end — the model gateway, the retrieval layer over a company’s own documents, and the governance that sits over both. We built: - A unified gateway to Anthropic, OpenAI, Google, xAI and others through a single interface, with a managed key or the organisation’s own (BYOK) - Retrieval over internal documents, with connectors to Google Drive, SharePoint and OneDrive, so answers are grounded in the company’s own sources - Budgets at company, team and individual level, enforced automatically before every call - Guardrails that keep inappropriate content out and sensitive data in - Usage reporting down to who, what, which model, how many tokens and at what cost, with CSV export - Roles, teams and projects, so access follows the shape of the organisation Models can be compared side by side, so teams choose the right one for each task on evidence rather than habit.
Technology
Technology: - Web App: Next.js, React - Back-end: NestJS (Node.js) - Database: PostgreSQL with pgvector, Drizzle ORM - Cache and queues: Redis - AI models: All major providers through a single gateway - Infrastructure: Docker, deployable on-premise or in the EU
SummAry
WorkenAI gives an organisation one place to use AI without losing sight of cost, access or data. Instead of scattered tools and personal accounts, teams get a single governed workspace — with the controls and the audit trail that make AI defensible under the EU AI Act and GDPR.
