AI x Energy Systems

Autonomous agents for resilient, explainable, and carbon-aware energy infrastructure.

We are building autonomous agentis for energy systems: a research-driven venture turning multimodal, modular LLM agents into trustworthy copilots for forecasting, optimisation, and real-time control. Our research spans energy system stability, transportation electrification, and human-in-the-loop autonomy.

Lead

Dr Zekun Guo

Lecturer (Assistant Professor) in AI for Engineering

Research roots University of Hull • DAIM

Research Streams

Autonomous LLM agents
Planning, tool-use, and evaluation loops tailored to power systems.

Forecasting & optimisation
Battery dispatch, hydrogen infrastructure, and airport-scale energy design.

Human-in-the-loop safety
Explainability, oversight, and resilient control for critical assets.

Research Aims

The group’s research program bridges frontier AI and deployable engineering. We prototype agentic control stacks, validate them on real-world energy assets, and open-source methods that accelerate the transition to net-zero operations.

Aim 01

Agent-native energy orchestration

Design modular agent architectures for dispatch, scheduling, and anomaly response in batteries, EV fleets, and microgrids.

Aim 02

Trustworthy autonomy

Fuse model-based optimisation with LLM reasoning to guarantee stability, interpretability, and controllable risk boundaries.

Aim 03

Human-centered deployment

Build operator-first UX with transparent explanations, audit trails, and collaborative overrides for mission-critical decisions.

Aim 04

Open science & reproducibility

Release benchmarks, reference agents, and evaluation pipelines that let academia and industry compare approaches fairly.

Focus Pillars

The lab integrates AI agents with power engineering to unlock resilient and low-carbon operations.

Transportation Electrification

Agent-led charging coordination for EV depots and airport ground fleets; robust to renewable volatility and grid constraints.

Power System Stability

Fast contingency response using multi-agent coordination, uncertainty-aware scheduling, and adaptive safety envelopes.

LLM Multi-Agent Systems

Multi-modal perception, domain toolchains, and evaluator agents that close the loop between simulation and field data.

Team & Collaborators

We pair academic rigor with venture execution. Interested in collaborating or joining? Reach out below.

Group Leader

Dr Zekun Guo

Lecturer in AI for Engineering (University of Hull, DAIM); Founder of Energentic AI; seconded researcher at University of Birmingham.

Personal site →

Co-lead

Dr Julius Sechang Mboli

Lecturer (University of Hull, DAIM); expert in responsible AI, IoT, and Industry 4.0.

Personal site →

Researcher

Xingxing Ren

AI agent researcher at Coral Protocol; PhD candidate at Brunel University of London focusing on multi-agent LLMs.

PhD Student

Kayode Fashedemi

PhD researcher in Data Science (University of Hull), designing and applying AI methods to real-world energy decisions through rigorous workflows.

PhD Student

Germain Kofi Acka Aidoo

PhD candidate in Data Science (University of Hull). His research advances governance-embedded multi-agent systems for hybrid renewable microgrids.

Research Master

Muktar (Kolade) Arubyi

MSc by Research student in Data Science (University of Hull). Bridging engineering and analytics to enhance energy performance with data-driven methods.

Researcher

Salar Setorg

Researcher at University of Hull focusing on renewable energy and intelligent systems, and PGR Representative at the School of Engineering.

We are growing

Open to PhD researchers, visiting scholars, and industry pilots in energy systems, optimisation, and agentic AI.

News

Updates from the lab and venture side.

Dec 2025

Innovate UK ICURe Exploit

The Energentic team has been recommended for the spin-out route and is moving forward to the Exploit stage, running Jan-Mar 2026 with up to £15,000 funding support.

Dec 2025

NeurIPS 2025 workshops

Anemoi presented at the LAW workshop; SimuGen featured at the SEA workshop—showcasing governance-aware agents and multimodal simulation pipelines.

Sep 2025

Innovate UK ICURe Discover & Explore

Dr Zekun Guo leads the team progressing from ICURe Discover to ICURe Explore, securing £2,500 (Discover) and up to £40,000 to explore the market.

Mar 2025

IGNITE Network + Flexible Fund

Funding to prototype hydrogen infrastructure planning agents for airports, co-led with University of Birmingham and University of Glasgow.

Feb 2025

HI-ACT Secondment Fund

Funding to support a secondment at the University of Birmingham to research modelling hydrogen and electric demand at airports for UK decarbonised aviation.

Jan 2025

MSc AI for Engineering Programme

Dr Zekun Guo leading the first cohort of the MSc Artificial Intelligence for Engineering variant programme and the AI-driven Optimisation and Control module. Programme link

Blogs & Talks

Notes from our public writing and invited sessions.

Blog · Dec 2025

HI-ACT secondment: hydrogen aviation

Reflections on a HI-ACT-funded secondment at Birmingham Energy Institute—building hydrogen aviation models and deepening collaboration networks.

Read the blog →

Update · Dec 2025

SimuGen & Anemoi

Multi-modal and multi-agent frameworks for simulation code generation and agent-to-agent coordination; releasing open benchmarks.

SimuGen · Anemoi

Talk · Apr 2025

Agentic AI Energy Management

LLM-enhanced rolling-horizon decisions for battery systems, presented at SuperAIRE ECR Workshop (Sheffield, 2025).

View slides →

Blog · Feb 2025

Human-in-the-loop agents

Bridging human expertise and AI autonomy in multi-agent systems; co-authored with the CAMEL-AI community.

Read the piece →

Selected Publications

Research that underpins the group’s roadmap.

2025 · arXiv:2502.10557

Can Large Language Model Agents Balance Energy Systems?

Integrates LLM agents with scenario-based SUC to improve grid efficiency and resilience under wind uncertainty.

arXiv · GitHub

2025 · arXiv:2506.15695 · Presented at NeurIPS 2025 Workshop on Scaling Environments for Agents

SimuGen: Multi-modal Agentic Framework for Constructing Block Diagram-Based Simulation Models

Multi-modal agent pipeline producing accurate, interpretable Simulink code by combining visual diagrams with expert priors.

arXiv · OpenReview · GitHub

2025 · arXiv:2508.17068 · Presented at NeurIPS 2025 Workshop on Bridging Language, Agent, and World Models for Reasoning and Planning

Anemoi: Semi-Centralized Multi-Agent Systems with Agent-to-Agent Communication

Agent-to-agent MCP server enabling coordinated, real-time collaboration; tops GAIA benchmark with GPT-4.1-mini as planner.

arXiv · OpenReview · GitHub

Collaborate

We welcome joint projects, student researchers, and industry pilots. Share your use case and the asset you want to control.

Z.Guo2@hull.ac.uk
J.Mboli@hull.ac.uk

Fastest way to start a collaboration proposal or student project idea.

Links

University of Hull • DAIM

Connect

LinkedIn
GitHub