Dynamic orchestration

Dynamic Orchestration Powered by Machine Learning

Leveraging advanced machine learning and our intelligent Mapping Engine, PolySaaS automatically discovers, normalizes, and routes data between disparate business applications.

Deploy powerful cross-app workflows (like Dolibarr → Odoo customer sync) in minutes — fully configurable from the database, no code changes required.

Model the world

Represent your environment in structured data—for example, devices and relationships across a large org (e.g. PolySysMon-style monitoring)—so downstream analytics and models have a consistent picture.

Capture & store

Scanning and telemetry can feed search- and analytics-ready stores (Elasticsearch, BigQuery, or sinks you choose). Clean pipelines reduce export/import round-trips.

Train & serve

Connect frameworks such as TensorFlow, PyTorch, or scikit-learn via endpoints and orchestration rules you control—parameters and runs stay tied to tenant configuration.

Prompts & agents

Design and version prompts, chain workflows, and route to your models or optional external LLMs only when you explicitly allow—ownership, compliance, and cost stay yours.

Why this matters

Full loop in one place: configuration, data, orchestration, and execution aligned with DOSE—not a bolt-on “AI feature” with opaque limits.
Framework-agnostic core: first-class tables and APIs are not tied to a single vendor model.
Enterprise posture: sensitive workloads can stay on infrastructure you operate; external AI is opt-in.

In the product today (technical)

Tenant-scoped models: MLEngine, MLTaxonomy, MLDataset, MLPrompt (Django / PostgreSQL). MLPrompt stores key, description, and prompt_text; align matchingEventKey on the other ML rows with the same key string when you want them tied together.
REST API (authenticated): /dose/api/mlengines/, /dose/api/mltaxonomies/, /dose/api/mldatasets/, /dose/api/mlprompts/.
matchingEventKey fields link rows to orchestration and events (contract documented in repo).
Related configuration: parameters.Parameter with matchingKey for ordered, keyed parameter sets.

ML Engine selector (preview)

Pick an engine profile that PolySaaS can orchestrate for your tenant-scoped ML workflows.

Product screenshots

Captures from the live admin and tenant UI (PolySaaS / DOSE). Sources live in the repo under documentation/website/assets/chat-uploads-for-wp/; below they load from GitHub main for convenience. After upload to WordPress Media Library, swap each src to your wp-content/uploads/… URL if needed.

Django admin dashboard listing Dose Tenant Management models including ML prompts — Admin dashboard (Jazzmin): tenant-scoped models, including **ML prompts** alongside engines, taxonomies, and datasets.

Django admin change form for an ML dataset — **ML dataset** — tenant, keys, JSON payload, and publish metadata in admin.

Django admin form for an ML engine — **ML engine** — name, endpoint, optional `matchingEventKey`, description.

Django admin form for an ML prompt — **ML prompt** — tenant-scoped prompt key, description, and prompt text.

Django admin form for ML taxonomy — **ML taxonomy** — same pattern: tenant, event key, structured `content_json`.

Tenant user dashboard with PolySaaS branding — Tenant **end-user** view (example): workspace home and navigation context.

Jazzmin admin with theme customizer sidebar — Admin chrome (earlier layout): theme / UI builder — streamlined in current builds; shown for context.

Additional captures (full set) are in the same repo folder; add more <figure> blocks here or link to a gallery page as you prefer.

Internal / draft: This page is for refining messaging before it appears in main navigation or the public feature list. Claims here should stay aligned with documentation/product/ML-PLATFORM-CONCEPT-AND-TABLES.md in the PolySaaS repository.