Intelligent agents and models, built, run and governed.
AI Foundry is where enterprise AI is built and operated. Agents that decide and act, conversational and custom agents, ML and forecasting models, vision and orchestration are created, deployed and governed as one fleet, grounded in enterprise context.
Business processes that run themselves, end to end.
Agents make complex decisions, integrate across systems and execute multi-step processes autonomously through domain-specific agentic flows, at parallel scale.
Autonomous multi-step execution
Complex decision trees
Cross-system integration
Domain-specific agentic flows
Parallel execution at scale
Completion, cost and time metrics
Conversational Agents
Chat and voice agents that resolve, human-like and low-cost.
Chat, voice and flow-driven agents hold natural, context-aware dialogue and complete tasks across channels, built without code and measured on success, cost and reach.
Chat, voice and flow-driven agents
Natural-language, context-aware dialogue
Tasks completed end to end
Multi-channel deployment
Success, cost and reach metrics
No-code agent builder
Custom Agents
Describe the automation in plain English, get a governed working agent.
Agents are created from a plain-language description, with event monitoring, business rules and custom work instructions, through a governed draft-to-approved lifecycle.
Natural-language agent creation
Described in words, built by the platform
Event monitoring and business rules
Custom work instructions
Governed lifecycle (draft to approved)
Manage, edit and export
Agent Studio
Mission control for the AI agent fleet.
The whole agent fleet is built, deployed, operated and governed from one console, with fleet-wide success, cost and reach and operational governance at scale.
Build, deploy and operate agents
Chat, voice and flow-driven agents
Fleet-wide success, cost and reach
One operational console
Operational governance at scale
Generative-AI management for the organisation
Flow Orchestrator
Multi-agent workflows, modelled visually and orchestrated end to end.
A visual modeler orchestrates multiple agents and connected processors for data, AI and integration into end-to-end automated flows, versioned and reusable.
Visual workflow modeler
Multi-agent orchestration
Connected processors (data, AI, integration)
End-to-end automated execution
Versioning and lifecycle (new to approved)
Reusable, shareable flows
ML Studio
From feature to production model, the whole ML lifecycle.
One studio runs projects, datasets and experiments, feature engineering, notebooks and AutoML, with one-click deployment and live model monitoring.
Projects, datasets and experiments
Feature engineering and pipelines
Notebooks and AutoML
Experiment tracking
One-click model deployment (API)
Live monitoring (latency, health, volume)
Forecaster
Time-series forecasting for any metric, accuracy tracked.
Forecasting projects auto-select the best algorithm across ingested sources, tracking forecast metrics and accuracy through guided, no-code wizards.
Time-series forecasting projects
AutoML algorithm selection (ARIMA, Prophet)
Multi-source data ingestion (CSV, DB, API)
Forecast metrics and KPIs
Accuracy tracking (weighted MAPE)
Guided wizards, no code
Anomaly Detector
Anomalies caught the moment they appear.
Metrics are monitored continuously against learned baselines, with alerts, anomaly forecasts and automated actions, triaged system-wide.
Automatic anomaly detection
Continuous metric monitoring
Learned baselines (no manual thresholds)
Alerts and anomaly forecasts
Automated actions on anomalies
System-wide triage and watchlist
Vision AI
Images and video into structured data, built as flows.
Vision processors for object, text, face and motion compose into no-code visual flows over images and video streams, deployed in real time.
Vision processors (object, text, face, motion)
Annotators for human-readable overlays
Visual flows (no-code composition)
Image and video-stream processing
Real-time deployment
Component playbook (48 components)
Prompt Analyzer
Prompts as strategic assets, tested and certified.
A central prompt library tests, certifies and versions prompts with success-rate and execution metrics, keeping quality consistent across agents and flows.
Central prompt library
Prompt testing and certification
Release management and versioning
Success-rate and execution metrics
Consistency and quality control
Linked to agents and agentic flows
Omnis
An AI researcher that works the long task and delivers the report.
Goal-driven research runs long-horizon, multi-step tasks across enterprise data, cross-referencing and reasoning to report-grade deliverables, with resumable history.
Goal-driven AI research
Long-horizon, multi-step tasks
Integrated across enterprise data
Cross-referencing and reasoning
Report-grade deliverables
Task history and resumption
How it works
From build to production, one fleet.
Agents and models are built from enterprise context, orchestrated into flows, deployed and governed from one console, and measured on success, cost and accuracy in production.
Enterprise AI, built to run.
AI Foundry builds, runs and governs the agents and models that make the enterprise intelligent.