aime™ migrates to the Loom™ Workflow Engine™ — deterministic, observable orchestration for educational AI

aime today announced the completion of a foundational re-architecture of its orchestration layer. The platform has migrated from a Redis Stream–based chain-of-agents model to the aime Loom Workflow Engine — a novel orchestration engine designed, built and owned by aime — replacing event-driven coordination with deterministic, action-driven workflows across timetable updates, lesson generation, content enrichment and every other multi-step AI process in the aime platform.

Loom is not a fork, wrapper or rebrand of an existing workflow engine. It is an original aime creation, purpose-built for the requirements of educational AI: small, single-purpose workflows; deterministic replay of every run; strict typed contracts between activities; and an execution model tuned for the long-running, multi-step pipelines that lesson generation, curriculum alignment and classroom-scale orchestration demand.

The previous architecture relied on event-passing between a central orchestrator and a fleet of agents. As workflows grew in scope, that model produced deep conditional branching, limited parallelism, no deterministic replay, and a debugging surface that was increasingly difficult to reason about. The new Loom-based architecture replaces that 'orchestration forest' with small, single-purpose workflows whose execution is fully defined, fully observable and fully replayable.

What changes with Loom

• Action-driven execution — workflows are invoked directly, not coordinated through emitted events.
• Deterministic workflow definitions — each workflow has explicit control flow, with no implicit transitions.
• Agents become Loom activities — stateless, with strict typed input and output contracts.
• Built-in observability — step-level execution history, workflow state, failure points and duration metrics, with no custom logging layer.
• Deterministic replay from the failed step — no manual re-trigger, no state corruption, guaranteed consistency.
• High concurrency on modest hardware, with no queueing bottlenecks.

A cleaner separation of responsibilities

Loom also clarifies the contract between backend and AI engineering. Backend engineers now own workflow triggering, validated input, transaction boundaries and persistence. AI engineers own deterministic activity logic against strictly typed schemas, with no raw database access or external state mutation inside the AI layer. The platform's internal repository structure has been consolidated to match, giving aime a unified codebase and a simpler deployment pipeline.

Why this matters

Educational AI workloads are long-running, multi-step and unforgiving of silent failure. A lesson pack that half-generates is not a lesson pack. Determinism, observability and replay are not engineering luxuries in this domain — they are the difference between a system a teacher can trust and a system that produces occasional, unexplained gaps. Loom turns those properties into platform defaults.

“The previous architecture got us to a working platform. Loom is what gets us to a dependable one. We have moved from coordinating events and hoping the graph executes correctly, to defining workflows that execute the same way every time and can be replayed step-by-step when something goes wrong. That is the foundation educational AI needs.”

“We deliberately broke the 'orchestration forest' into small, single-purpose workflows. Each one is observable, replayable and easy to reason about — which means our engineers spend their time on pedagogy and product, not on chasing distributed state.”

“Loom is how aime scales without throwing infrastructure at the problem. The engine's internal design, concurrency model and execution substrate are aime's proprietary IP.”

Availability

The Loom Workflow Engine is in production across the aime platform as of March 10, 2026, and underpins orchestration for aimeCLOUD™, aime's lesson generation pipeline and every multi-step AI process in the aime stack. Legacy event-stream orchestration has been fully phased out.

About aime

aime builds the operating system for educational intelligence — the foundational infrastructure layer that future education systems will run on. aime's stack combines structured curriculum knowledge, pedagogy-aware reasoning, compact education-tuned models, agentic orchestration and offline-capable deployment, and is designed for ministries of education, universities and national school systems.