Product and architecture
Strengthen the system without discarding the useful work already done.
Proof is in the system behind the screen. See the workflow, safeguards, and operating reality.
AI engineering case studies from Smicolon: the architecture, workflow controls, and safeguards behind the products and internal systems we build and run.
- Architecture before scale
- Tested before release
- Monitored in production
- Owned by your team
Read the system, not just the surface
Every study shows the architecture, the controls, and the decisions behind the screen — the parts that decide whether a product survives contact with real users.
Quality and testing
Define what good behavior means and catch regressions before users do.
Security and data
Make permissions, sensitive paths, tenancy, and ownership explicit.
Deployment and operations
Trace what happened, release without holding your breath, recover fast, and keep the system maintainable.
Product engineering
BrandBuddy
BrandBuddy helps small Danish businesses plan, write and send email campaigns. Rasmus built the first prototype with Lovable. Smicolon supported the next stage with technical guidance, code review and UX/UI design as the product reached its first paying customers.
Prototype to ProductionAI Product DevelopmentAI Product Experience Design
Edge runtime
In-worker Mastra runtime
Agents and workflows execute inside the dashboard API Worker.
Durable state
D1 records workflow state and snapshots beyond a single request.
Queue boundaries
Consumers, retries, and dead-letter paths separate request handling from long-running work.
Traced execution
Every workflow step emits telemetry; Mastra and Langfuse turn it into a readable trace.
Internal operating system
Edge-Native AI Orchestration
A technical deep dive into durable Mastra workflows running inside a Cloudflare Worker with D1 state, queues, model routing, telemetry, and Payload publication.
LLM Application ScalingPrototype to Production
AI Applications
Digibate
Digibate is an AI-powered social media content and marketing platform that helps founders and marketers create, plan, and publish — with AI that assists, and a user who's always in control.
AI Product Development
Proposal flow
Opportunity intake
Webhook-driven records create a consistent starting point for review.
Configurable workflow
Application rules and agent-assisted stages are explicit rather than hidden in personal routines.
Reviewable state
Applications are edited, approved, submitted, and tracked through every status change.
Commercial context
Economics remain visible alongside the application decision.
Internal operating system
Opportunity and Proposal Operations
How Smicolon turned incoming marketplace opportunities into a configurable, reviewable commercial decision workflow with application state and economics in view.
Prototype to Production
