CogniGraph, IronFlow, and IronCrew are Rust. TinyCrew and this website are Bun/TypeScript. TranscriptIntel's analysis pipeline is Python. When we recommend a stack for your project, it's because we ship production software in it daily — not because it was on a slide.
Six production languages — each chosen for what it does best, each proven in shipped products.
Performance-critical systems and tools
CogniGraph, IronFlow, IronCrew, Cognitive-OCR, CogniGraph Chunker — all async Tokio/Axum, all single-binary deployments. We choose Rust when performance, memory safety, and zero-dependency deployment matter.
Web applications and APIs
This website runs on Bun with React 19 SSR. TinyCrew is our Bun-native multi-agent framework. PLU Finder is TypeScript end-to-end. We build with React, Fastify, Hono, and Nest.js.
AI/ML pipelines and data processing
TranscriptIntel's analysis pipeline, BehaviorSync's extraction engine, and our ML training workflows. FastAPI for APIs, Streamlit for data apps, Pydantic for validation.
Embedded scripting and edge computing
The scripting layer in IronFlow (96-node workflows), IronCrew (multi-agent crews), and CogniGraph (server-side graph operations). Also used with OpenResty for high-performance API gateways and Redis scripting.
Enterprise Windows and cross-platform
.NET Core for enterprise services and cross-platform applications. WinForms for Windows-native desktop tools. SDKs and integrations for the Microsoft ecosystem.
Enterprise microservices
RESTEasy on embedded Undertow for lightweight microservices that ship as a single executable JAR — no external servlet container. Jakarta REST with Jackson JSON, used for enterprise service integrations.
Modern reactive interfaces — we're opinionated but pragmatic.
In-house — Bold, structured design language
Our own design system inspired by Dutch graphic design traditions — sharp geometry, bold typography, high contrast, grid discipline. 30+ components including data tables, modals, steppers, file uploads, and canvas signature pads. Three theme variants (Orange, Purple, Crimson). Works with vanilla HTML/CSS, Tailwind, React, and Vue.
We match the database to the problem, not the other way around.
Our primary database
Graph + document + vector in one database. AQL for complex queries, Foxx Microservices for server-side logic, native clustering. CogniGraph uses it as a pluggable backend. We've run ArangoDB in production for 10+ years.
Qdrant for pure vector performance. LanceDB for embedded deployments. CozoDB as CogniGraph's embedded backend. Weaviate and AstraDB when the infrastructure calls for them.
PostgreSQL for relational workloads. SQLite via IronFlow's db_query node. MongoDB for document stores. Redis for caching and session management. CockroachDB for distributed SQL.
AWS S3 and S3 Vectors for RAG pipelines (IronFlow has native S3 nodes). Cloudflare R2 for edge storage. Minio for S3-compatible on-premise.
Practices that ship reliable software, not theoretical best practices.
Claude Code for agentic coding assistance. AI-powered code review and test generation. Our AI Orchestrator methodology delivers 10-30x faster than traditional approaches — proven across every product we've shipped.
TypeScript strict mode everywhere. Vitest and Jest for testing. ESLint and Prettier for consistency. CogniGraph has 127 unit tests. IronFlow ships with 20 example categories. We test what we ship.
Pluggable backends (CogniGraph's GraphBackend trait). Sandboxed scripting (Lua in IronFlow and IronCrew). Clean HTTP APIs (Axum, Fastify, FastAPI). Single-binary deployments for Rust tools. Docker when containerization adds value.
Products and platforms in production — not abandoned prototypes.
Contact us to discuss your project. We'll recommend a stack we actually use — and show you the products we built in it.
Get in Touch