March 12, 2026 — Built on eSmart Systems’ patent-pending Adaptive AI, the new platform lets utilities and technology companies build, deploy, and scale custom AI models in minutes - without machine learning expertise or retraining cycles, directly accessible for AI Agents.

eSmart Systems, the market leader in computer vision for electric utility infrastructure, today announced the public launch of AI Studio by eSmart Systems: a full-stack AI development and deployment platform that makes the company's proprietary computer vision intelligence available as modular, API-first services for the first time.

AI Studio enables utility teams, technology companies, and inspection professionals to build, tune, and deploy custom AI models aligned to their own operational standards and asset definitions - without requiring an in-house machine learning team.

At the core of AI Studio is eSmart Systems' patent-pending Adaptive AI technology: a methodology that enhances performance from few-shot learning to generate working detectors and classifiers from a small number of images in minutes, adapting in real time to new definitions and edge cases without traditional retraining cycles. The result is up to 35 percentage points accuracy ahead of competing few-shot technologies and ~99% accuracy across every benchmark, with zero retraining.

Solving a Decade-Long Barrier to Utility AI Adoption

For utilities, the promise of AI-powered grid inspection has long been limited by a fundamental challenge: generic computer vision models are not built for the diversity of real-world infrastructure assets, local inspection standards, or the operational requirements of transmission, distribution, and substation workflows. Customizing AI to meet those standards has historically required significant internal machine learning expertise - a resource most utilities and inspection teams do not have.

AI Studio removes that barrier entirely.

"We’ve spent over a decade building infrastructure-grade AI for utilities. AI Studio opens that intelligence as a platform." said Henrik Bache, CEO of eSmart Systems. "It allows teams to customize and scale without friction, while preserving the governance and reliability required in critical infrastructure."

Three Capabilities Powering an End-to-end Model Development Platform

AI Studio is built around three core capabilities, accessible through both a web-hosted interface at ai.esmartsystems.com and fully documented APIs:

• Model Builder: Create new custom models from a handful of example images using Adaptive AI. No annotation pipelines, no training runs, no ML expertise required. Models continuously adapt in real time as definitions evolve or edge cases emerge.
• Model Garden: Access to eSmart Systems' library of production-ready detectors, classifiers, and pipelines, built on millions of power utility images across transmission, distribution, and substation domains. Continuously refined from real-world deployment across 75+ utilities globally.
• Pipeline Builder: Compose multi-step vision workflows visually using a drag-and-drop editor, for example, detection → classification → measurement — with instant API deployment and full input/output transparency for audit and compliance purposes.

Every model and pipeline in AI Studio exposes a dedicated API endpoint, designed for direct consumption by AI agents and integration into existing inspection platforms, engineering systems, drone ecosystems, and enterprise workflows.

Built for the Agentic AI Era

AI Studio has been designed from the ground up as an Agent first, API-based platform. It reflects eSmart Systems' belief that enterprise software is shifting away from pre-built user interfaces toward dynamically assembled workflows powered by AI agents.

"What makes Adaptive AI genuinely different is that it breaks the dependency between model performance and retraining cycles. When a utility's definition of a defect changes - and it always does - the model updates in minutes, not weeks," said Erik Åsberg, CTO of eSmart Systems. "That changes the economics of deploying AI in the field entirely."

A Defensible Platform Built on Proprietary Intelligence

eSmart Systems' competitive position rests on three pillars that no generic computer vision or foundation model vendor can replicate: more than a decade of proprietary, contextualized utility imagery and asset data built across 75+ utilities worldwide; patent-pending Adaptive AI that eliminates the customization barrier without retraining; and a platform architecture designed to absorb and leverage improvements in frontier AI models over time.

The result is an AI platform that enables immediate value and continuous improvement through rapid iterations, with performance previously unseen in these types of computer vision applications. An AI platform that is ready for the Agentic AI reality we are now in.