Agentic AI Meets CCS : Building a Smarter Project Screening Tool /Decisions

By: Tariq Siddiqui: April 2026

Fore Thoughts

Over the past few weeks, I’ve been experimenting at the intersection of my core domain—carbon capture and storage (CCS)—and a rapidly evolving concept in AI: agentic systems. The idea was simple: can we move beyond static spreadsheets and single-model analysis toward a system where multiple specialized “agents” collaborate to evaluate a CCS project—much like a multidisciplinary team would?

From Domain Experience to Systems Design

With decades of experience in subsurface CCS, subsurface engineering and project development, i wanted to explore how this knowledge could be translated into a repeatable, scalable workflow. I built a lightweight prototype that brings together four distinct perspectives:

• Subsurface feasibility

• Infrastructure

• Market and regulatory context

• Financial viability

  
  

Each of these is handled by a dedicated AI agent, coordinated through an orchestration layer. The output is not just a narrative, but a structured assessment—similar to how early-stage project screening is done in practice and how the risks are identified and mitigated in each area.

Dual-layer Decision Framework

To complement this, I added a deterministic scoring layer. While the agents provide AI-Driven qualitative reasoning, the Deterministic scoring model applies consistent, auditable logic across key variables such as CO₂ source type, transport distance, reservoir characteristics, and data quality. This creates an interesting and valuable tension: the AI may lean optimistic, while the scoring model enforces discipline. What’s emerging is not just another AI demo, but the outline of a decision-support workflow:

• Multi-agent reasoning for breadth of insight

• Deterministic scoring for consistency

• Structured outputs for decision-making

Towards Agentic Decision-Support Systems

While still an early prototype, this work demonstrate how domain expertise can be embedded into an agentic AI framework to support complex engineering decisions. For CCS—where uncertainty, capital intensity, and long project cycles dominate—this kind of hybrid approach could significantly improve early screening and prioritization.

Next Steps

The next steps will be:

1. Structuring output for consistency

2. Integrating domain knowledge more deeply

3. Introduce Retrevable-Augmented generation (RAG) to add domain proprietary information

Closing Thought

AI is most powerful not when it replaces engineering judgment, but when it augments and challenges it.

Curious to hear how others are thinking about applying agentic AI in complex, multidisciplinary domains like energy transition.

---

The UEPA : Navigating CCS Complexity for Client Success

For your next CCUS project, trust the expertise of UEPA. With our in-depth knowledge of CCS project development, we offer precise and efficient support for your project. Our comprehensive services cover the full lifecycle of project development and reservoir characterization, ensuring successful Class VI permit applications. Partner with UEPA to ensure your CCUS project is executed smoothly and sustainably.