Technology
Stack
The systems layer we design and ship against.
For operators and engineers who want the actual surface area — models, languages, runtimes, data, cloud, agent harnesses, and the enterprise systems we connect into. Not a vendor checklist. The stack we audit, architect, and execute with.
Spec surface
Depth for people who already speak the stack.
Categories below are how we usually cut the architecture conversation — model plane, language plane, product surface, data, infra, tooling, and systems of record.
01 — Models & agents
Multi-model by design. Harnessed for production.
We architect against model APIs and open weights — routing, evaluation, memory, tool use, and human-in-the-loop controls. Agent harnesses, Hermes-style runtimes, and pipeline orchestration so intelligence compounds instead of becoming another brittle demo.
- OpenAI GPT family
- Anthropic Claude
- Google Gemini
- xAI Grok
- Hugging Face / open weights
- Agent harnesses & tool routers
- Hermes & orchestration runtimes
- Eval, tracing, and guardrails
02 — Languages & runtimes
Systems languages where it matters. Scripting where it ships.
TypeScript and Python for product and agent surfaces. Rust and C++ when latency, safety, or native edges demand it. Swift for Apple clients. Bash and Linux for the glue that keeps pipelines honest.
- TypeScript / JavaScript
- Python
- Rust
- C++
- Swift
- Bash / shell
- Node.js
- Linux environments
03 — Product surfaces
Modern web stacks, not framework religion.
React and Next.js as the default commercial surface. Svelte or Angular when the existing estate requires it. Tailwind and HTML5 for interfaces that stay fast and maintainable under real operating load.
- React
- Next.js
- Svelte
- Angular
- HTML5
- Tailwind CSS
- Node services
04 — Data & backends
Operational truth lives in the data plane.
Postgres as the backbone. Redis for hot paths. Supabase and Neon when we want managed velocity without giving up SQL. Prisma and GraphQL where the access layer needs structure.
- PostgreSQL
- Redis
- Supabase
- Neon
- Prisma
- MongoDB
- GraphQL
- FastAPI / Django
05 — Cloud, CI, infra
Shipable infrastructure with explicit failure modes.
Vercel and Cloudflare for edge/product delivery. AWS when the estate is already there. Docker and Kubernetes for portable runtime. Terraform, GitHub Actions, and Nginx for the path from commit to production.
- Vercel
- AWS
- Cloudflare
- Docker
- Kubernetes
- Terraform
- GitHub / Actions
- Nginx
06 — Tooling & delivery
IDEs, agents, and the pipes between them.
Cursor and VS Code for human+agent development loops. GitHub as source of truth. Pipelines for build, eval, and deploy. We design the harness around how teams actually work — not how slide decks claim they work.
- Cursor
- VS Code
- GitHub
- CI pipelines
- Agent tooling
- Figma handoff
07 — Enterprise systems
Intelligence that attaches to the systems of record.
CRM, ERP, comms, and knowledge layers are usually the constraint. We map Salesforce, SAP, Slack, Notion, and adjacent suites into the architecture so agents and workflows have real context — with permissions and auditability intact.
- Salesforce
- SAP
- Slack
- Notion
- Custom CRM / ERP
- SSO / identity boundaries
Next
Have a stack already? We map it before we prescribe.
Bring the repo, the CRM, the model bills, and the friction. We'll tell you what to keep, cut, and rebuild.