The stack we build and run on
We’re tool-agnostic and pick what fits the job. These are the technologies we use day to day, and the platforms we integrate with, described honestly as integrations, not partnerships.
Build
- PHP 8.3
- Python
- TypeScript / Node
- React & vanilla JS
- MySQL / PostgreSQL / SQLite
- REST + webhooks
AI & agents
- LLM APIs (provider-agnostic)
- Retrieval / RAG pipelines
- Speech-to-text & TTS for voice
- Prompt + eval harnesses
- Guardrails & handoff logic
- On-site simulated mode when no key
Automation
- n8n / queue workers
- Event-driven triggers
- CRM & sheet sync
- Scheduled jobs + retries
- Run logging & alerting
Growth & data
- Meta & Google Ads APIs
- GA4, Search Console, Clarity
- Server-side + first-party events
- Conversion API
- Looker-style reporting
Infrastructure
- Shared & VPS hosting (Hostinger, etc.)
- HTTPS everywhere
- Backups & migrations
- Staging → production flow
- Secrets in environment vars only
Security
- Prepared statements
- CSRF + output escaping
- Rate-limited auth + hashing
- Role-based access
- MIME-validated uploads
- Error logging, no raw errors in prod
How we choose
If a no-code tool solves it reliably, we use it. If it needs to scale or you need to own it, we build it. Either way you get documentation, access to everything, and no lock-in to us.
Technology questions
Pragmatic, production-grade choices: modern PHP for server-side, a clean front-end, SQLite or PostgreSQL for data, and proven automation platforms where they fit.
Both. If a no-code tool solves it reliably, we use it. If it needs to scale or you need to own it outright, we build it.
We are model-agnostic: a frontier model for reasoning, a smaller cheaper model for classification and extraction. You are never locked to one vendor.
Yes. We build on standard stacks that run anywhere from shared hosting to cloud, with no proprietary runtime.
Yes. Full source in your repository, documented, with a handover session.
Parameterised queries, output escaping, CSRF protection, dependency scanning, server-side secrets and least-privilege access are standard.
Fast load times, semantic markup, schema and clean URLs are requirements of every build, not additions.
Retries with backoff, circuit breakers and queues, so your product keeps working when a third-party provider has a bad day.
Yes. We document the system and build an admin interface so routine changes do not need us.
We weigh reliability, cost, scale and ownership, and recommend the cheaper path when it genuinely serves you.