AI and automation / 2026
Most technical learning material explains a mechanism in prose or video but gives the learner no controlled way to change assumptions, inspect state transitions or test whether they actually understand the system.
Simulation Learning Lab now spans evidence intake, deterministic media preprocessing, OCR and optional transcription, multimodal mechanism generation, structured authoring, a validated simulation contract, compilation into a portable runtime, eight visual grammars, provenance, fidelity tracking, active recall and local mastery state.
My part
Decisions
A lesson should describe what changes, why it changes and what state follows. Visual grammar is then selected as a renderer rather than becoming the source of truth for the lesson itself.
Validated lesson specifications compile into a reusable runtime so a new explanation does not require another handcrafted application before it can become interactive.
Multimodal generation can accelerate mechanism authoring, but evidence preparation, manual authoring, validation, compilation and learning remain usable without a provider key.
The runtime keeps evidence references and fidelity choices visible so a learner can distinguish observed source material from assumptions, scaling and illustrative teaching decisions.
Screens
SYSTEM ARCHITECTURE
Deterministic preprocessing records media metadata, representative video frames and OCR or transcript evidence before semantic interpretation begins.
A structured simulation specification is the boundary between authoring and rendering, keeping mechanism state independent from presentation grammar.
The compiler validates and normalises the specification, applies grammar defaults and emits a portable runtime bundle.
The runtime supports guided stage traversal, scenario changes, state inspection, retrieval prompts, provenance and fidelity evidence from the same causal model.
Sessions remain file-based and local-first so a hosted storage adapter can be introduced later without changing the compiler contract.
Built with
More work
Gazelle's KitchenA WhatsApp business gatekeeper that qualifies genuine orders, enforces business and safety rules, contains off-topic or abusive conversations, and only surfaces the small set of cases that genuinely need the owner.PilotSellavie AIAn independent project: multi-tenant conversational-commerce software that connects social messaging with business knowledge, guarded AI, human handoff, orders, billing, analytics and operations.MaintainedAI Automation Control PlaneA runnable multi-tenant control plane for AI-assisted business workflows, combining FastAPI, Next.js, PostgreSQL, Redis, n8n, provider routing, human approval, idempotency, replay, cancellation, audit traces, and deterministic local demonstrations.Maintained