ForestHub

What ForestHub does

ForestHub is a deep-tech company building the infrastructure for autonomous AI agents on embedded systems. We enable intelligent agents to run directly on microcontrollers and industrial edge hardware — fully on-device, without cloud dependency. Our platform bridges classical embedded development with modern agent-based AI. Using a visual development environment, developers can design, configure, and deploy AI agents that analyze signals, detect anomalies, support decision-making, and operate within real-time and resource constraints. ForestHub brings intelligence where it matters most: on the machine, in the field, and at the edge. By eliminating cloud dependency, we enable low-latency decision-making, full data sovereignty, and resilient system architectures — especially relevant for manufacturing, automation, and mission-critical environments. ForestHub helps industrial teams bring AI onto their own devices, from the first feasibility check to production rollout on the plant floor. The team covers feasibility, architecture, prototyping, integration and training, works hardware agnostic and vendor independent, and builds every solution edge first so data stays on site.

Products

• ForestHub Orchestration PlatformAn Edge AI and Agents Orchestration Platform where workflows are graphs and the AI is one node among many. Built for industrial control where every decision must be inspectable, replayable, and bounded. It covers the full lifecycle: design, deploy, run, observe, from visual workflow to running engine on your Linux edge device.
• Workflow BuilderA visual canvas for industrial agent workflows. Author the sense-reason-act loop as a graph, wire sensors, deterministic operations, LLM agents, and actuators into a single workflow, then deploy as a versioned artifact to any registered device.
• Engine RuntimeA Go binary in a Docker image (distroless, ~10–15 MB, linux/amd64 + linux/arm64) that loads `workflow.json` and interprets the graph at runtime. No code generation, no recompile per device. It runs standalone and offline.
• LLM-proxyPart of the open-source runtime, enabling interaction with language models.
• Workflow contractPart of the open-source runtime, defining the structure and interaction of workflows.

Key capabilities

• Edge AI and Agents Orchestration Platform
• Visual workflow builder
• On-device execution without cloud dependency
• Support for various intelligence tiers (rule-based logic, classical ML, SLMs, mid-sized open-weight LLMs, frontier LLMs via cloud)
• Inspectable, replayable, auditable, and bounded decisions
• Low-latency decision-making
• Full data sovereignty
• Resilient system architectures
• Open source runtime (AGPL-3.0)
• Managed platform with hosted visual builder and device management
• Team collaboration and governance
• Managed hosting and updates
• Commercial licensing options
• OEM and integration engineering services
• Hardware agnostic and vendor independent solutions
• GDPR compliant solutions by design
• Knowledge transfer and long-term enablement

Use cases

• Analyzing signals
• Detecting anomalies
• Supporting decision-making
• Operating within real-time and resource constraints
• Smart buildings
• Industrial control
• Vibration-based machine monitoring
• Line-side image recognition
• Predictive maintenance
• Classification
• Short diagnostic dialogs with technicians
• Document RAG (Retrieval Augmented Generation)
• Demanding reasoning on-premise
• Complex customer-facing dialog
• Multilingual reasoning
• Protecting uptime and cutting cooling costs in data centers
• Lowering energy use and equipment downtime in retail
• Reducing operating costs and holding comfort/ESG targets in commercial real estate
• Cutting energy bills in hotels without sacrificing guest comfort
• Early detection of falls and quiet risks in senior living
• Routing critical alarms and trimming energy in hospitals
• Optimizing heating and cooling with less energy and lower peak load for HVAC manufacturers
• Shipping real intelligence on OEM hardware in smart homes
• Embedding local decision layers for building automation OEMs

AI approach

ForestHub provides an Edge AI and Agents Orchestration Platform that allows developers to design, configure, and deploy AI agents on embedded systems. It orchestrates various intelligence tiers, from rule-based logic and classical machine learning to small language models (SLMs) on-device, mid-sized open-weight language models on-premise, and frontier language models via cloud. The platform emphasizes on-device processing, data sovereignty, and auditable, bounded AI decisions, particularly for industrial control environments.

Tech named: XGBoost, random forests, compact neural nets, Phi, Gemma, Llama, Qwen, Claude, GPT, Gemini, Docker, Go, NVIDIA Jetson

Industries served

• Embedded Software Products
• Manufacturing
• Automation
• Mission-critical environments
• Industrial control
• Data centers
• Retail
• Commercial real estate
• Hotels
• Senior living
• Hospitals
• HVAC manufacturers
• Smart home OEMs
• Building automation OEMs
• Machine building
• IoT

What it says sets it apart

• Infrastructure for autonomous AI agents on embedded systems
• Fully on-device, without cloud dependency
• Bridges classical embedded development with modern agent-based AI
• Visual development environment for designing, configuring, and deploying AI agents
• Intelligence orchestration: right intelligence at the right layer (rule-based, classical ML, SLMs, mid-sized LLMs, frontier LLMs)
• Graph-first approach: the graph is the program, AI is a node
• Every decision is inspectable, replayable, auditable, and bounded
• Data sovereignty: data stays local, GDPR compliant by design
• Vendor independence: no cloud requirement, no vendor lock-in
• Engineering substance: code built to hold up under industrial conditions
• Made in Germany: German engineering standards, long-term product care, reliability
• Grown out of years of applied research at Hahn-Schickard Institute
• Open source runtime available under AGPL-3.0