We build AI tools that work with you.

01

Client Chatbot

Secure document-grounded assistant with cost auditing

A document-grounded chat assistant reached through encrypted, time-limited URL tokens. Users select authoritative source files and get Markdown-rendered answers constrained to that material — while IP-restricted admin surfaces manage the files and every exchange is logged with token counts and computed cost.

AI inside: a file-context retrieval pattern where the selected documents are injected directly into the prompt, so the model answers strictly from the approved material and cites which file each answer came from — with per-message cost tracking for billing visibility.

03

AI Documentation Generator

From source code to polished docs in one command

A command-line tool that turns a JavaScript source file into a single, self-contained HTML documentation page — with live search, filters, deep links and source snippets — plus watch and serve modes with automatic browser reload.

AI inside: an optional LLM enrichment layer writes summaries, descriptions, examples and per-parameter docs for every extracted API item; hand-written JSDoc always takes priority, and deterministic heuristics guarantee documentation is produced even with no AI available.

05

Safe LLM

Frontier-model power without exposing sensitive data

Use powerful external LLMs without your sensitive data ever leaving your control. Before any text or document goes out, a local anonymization pipeline detects PII — names, institutions, IBANs, IDs and more — and replaces each with a consistent, realistic substitute. Responses are seamlessly rehydrated with the real values on the way back.

AI inside: a trusted local LLM (via Ollama) performs anonymization, complemented by Microsoft Presidio and NER/regex multi-pass PII detection; an encrypted per-project mapping vault guarantees consistent substitution across documents, with leak validation on every response.

07

Advanced RAG Chatbot

A complete platform for building secure, citation-backed RAG chatbots.

Organizations can create internal AI assistants that understand and converse with their own documents. The platform handles document ingestion, knowledge organization, retrieval, chatbot creation, and access control in one place.

Every answer includes inline citations that link directly back to the original source material.

AI inside: a seven-stage LangGraph retrieval pipeline — query rewriting, multi-query expansion, HyDE, hybrid vector + BM25 search with Reciprocal Rank Fusion, CRAG relevance checking, LLM reranking and cited answer generation with GPT-4o.

09

EON FM

A fully autonomous AI radio station

A radio station with no human on the mic. An AI agent plans the weekly and daily schedule, writes and reads the news, picks and downloads music, runs listener quizzes and reacts to listener messages live — streamed as a wall-clock-synchronized broadcast, so every listener hears the same thing at the same moment.

AI inside: a LangChain/LangGraph deep agent orchestrating specialized subagents (weekly planner, daily programmer, listener-message handler), OpenAI LLMs for news and DJ patter, OpenAI or local Qwen3-TTS for the station voice — with deterministic fallbacks so the broadcast never stops.