Want to go deeper? Join our Patreon community for exclusive, not publicly available content and support the future of architecture, engineering, and construction.

INDUSTRY INSIGHTS
AI Agents: The Definitive Guide

Let’s face it. No matter how hard you try, you cannot avoid the conversation on AI.

In recent weeks, we have been lucky enough to sit down with a number of experts including:

• Rajitha Charapala - VP of Product, Data & AI

• Thiago Da Costa - CEO Founder at Datagrid

• Alan Mosca - CTO at nPlan

• James Garner - Global Head of Data at Gleeds

The recurring theme in all of these conversations? 

Agents!

As someone extremely curious, yet not quite understanding of the impact/use cases and potential of AI Agents, these conversations were extremely insightful and educational in understanding the biggest hype of 2025.

So with that said, we felt it timely to share these experts insights on AI Agents.

AI Agents in Construction: A Comprehensive Guide

Understanding AI Agents

AI agents represent a fundamental shift in how artificial intelligence operates within construction. Unlike traditional AI systems that simply respond to specific prompts or queries, AI agents possess autonomous capabilities that allow them to take action, learn from outcomes, and make decisions independently. These agents can observe their environment, process information, and execute complex sequences of tasks without constant human intervention.

At their core, AI agents combine intelligence with agency - the ability to act on their own initiative. As Thiago Da Costa, founder of DataGrid, explains:

This capability transforms them from passive tools into active participants in construction processes.

What truly sets agents apart is their ability to chain multiple tasks together and understand context. When presented with a challenge, an agent can break it down into steps, execute each one in sequence, and adjust its approach based on the results. This mirrors how human professionals work, but with the added advantages of tireless operation and consistent performance.

/

Current Applications in Construction

The construction industry is seeing AI agents deployed across numerous specific applications. Here's a few ways on how they're being used today, based on real implementations described by industry leaders:

Document Management and Administration

• RFI Management: Agents can review incoming RFIs, cross-reference them with existing documentation, and even suggest responses based on historical data and project specifications

• Submittal Processing: Automated review of submittals against specifications, checking for compliance and flagging discrepancies

• Standard Operating Procedures: As described by Thiago Da Costa, "We built an SOP generator agent that actually generates the SOPs based on how companies have worked in the past... you can connect in your past projects, your past bids, and you can ask the agent to generate the SOPs"

• Contract Analysis: Agents can review contract documents to identify potential risks, inconsistencies, or missing information

• Documentation Updates: Automated tracking and updating of project documentation as new information becomes available

Field Operations

• Installation Guidance: Field teams can photograph equipment or materials and receive immediate installation instructions and specifications

• Progress Tracking: Agents analyze site photos to quantify work completed and compare against schedule

• Quality Control: As described by Da Costa, "If you ask an AI to quantify something, it should actually check if you have any other drawings that have that same thing with more detail, or maybe there's a schedule somewhere else"

• Asset Location: Teams can quickly locate equipment and materials by querying agents about inventory and placement

• Field Queries: Immediate responses to field questions by accessing and synthesizing project documentation

Design and Engineering

• Drawing Review: Agents can compare different versions of drawings to identify changes and potential conflicts

• Model Comparison: Agents can look at entire models and compare them to other models and compare the things that are wrong with themodel without any instructions

• Specification Compliance: Automated checking of designs against project specifications and building codes

• Quantity Extraction: Automated take-offs and quantity calculations from drawings and models

• Schedule Generation: Agents can generate full works schedules and programmes, optimising for critical path, risk and workforce planning

Risk Management and Safety

• Water Intrusion Prevention: Agents analyze site photos and weather data to identify potential water intrusion risks

• Safety Hazard Detection: Automated analysis of site photos to identify safety violations or potential hazards

• Schedule Risk Analysis: Agents analyze project data to identify potential delays and their knock-on effects

• Financial Risk Assessment: Monitoring of cost data and progress to flag potential budget overruns

• Environmental Monitoring: Integration of weather data with site conditions to predict and prevent environmental risks

Project Controls

• Cost Tracking: Agents monitor actual costs against budgets, flagging variances and potential issues

• Schedule Monitoring: Automated analysis of progress and potential impacts on project timelines

• Resource Optimization: Agents can suggest optimal resource allocation based on project needs and constraints

• Change Order Management: Automated analysis of change orders for impact on schedule and cost

• Performance Analytics: Continuous monitoring and analysis of project performance metrics

Communication and Collaboration

• Meeting Documentation: Agents can join meetings, take notes, and automatically categorize and distribute information

• Information Distribution: Automated routing of project information to relevant team members

• Query Resolution: Agents can answer project-related questions by synthesizing information from multiple sources

• Stakeholder Updates: Automated generation of progress reports and stakeholder communications

• Cross-team Coordination: Agents help ensure all team members have access to current and relevant information

Each of these applications represents a real implementation of AI agents in construction today. What's particularly powerful is how these agents can work together. As Rajitha Chaparala explains, "You can have multiple agents, like what's getting triggered after one is done thinking about the next step." This interconnected approach allows for comprehensive automation of complex workflows while maintaining human oversight at critical decision points.

BEFORE WE DIVE BACK IN…
Innovating the Future: Robotics and the Revolution in Construction

Get the full story. Purchase "Innovating the Future: Robotics and the Revolution in Construction" now on our Patreon.

OR

Get the report for FREE by sharing this newsletter with someone else.

Getting Started with AI Agents

For organizations looking to explore AI agents, the key is to start small and focus on well-defined use cases. Begin by identifying repetitive tasks that require multiple steps or decision points. These might include document review processes, safety checks, or progress monitoring workflows. The goal is to find areas where agents can provide immediate value while limiting complexity.

Many construction technology providers now offer agent-based solutions that can serve as entry points. These platforms often come with pre-built agents designed for specific construction tasks, allowing teams to experiment without significant technical investment. Some providers also offer development environments where organizations can customize agents to their specific needs.

The experimentation phase should focus on learning and adaptation. Start with a single workflow, measure results carefully, and gather feedback from users. This approach allows teams to understand both the capabilities and limitations of agents while building confidence in the technology.

Implementation Guide - 10 Steps to AI-Agent Readiness

Successful implementation of AI agents requires a structured approach that combines technical readiness with organizational change. Here's a comprehensive list based on insights from industry leaders:

1. Assess Data Readiness

Start by evaluating your organization's data infrastructure. As Thiago Da Costa explains,

This assessment should include:

• Audit of existing data sources and formats

• Evaluation of data quality and completeness

• Review of data access and integration capabilities

• Assessment of data security protocols

2. Document Core Processes

Before implementing agents, map out your workflows in detail. This documentation should capture:

• Step-by-step process flows

• Decision points and required inputs

• Expected outputs and success criteria

• Current pain points and inefficiencies

3. Select Initial Use Cases

Choose your first implementation carefully. As James Garner advises,

Look for:

• High-volume, repetitive tasks

• Workflows with clear inputs and outputs

• Processes with existing digital documentation

• Areas where mistakes are costly but easily preventable

4. Build Your Technology Stack

Develop the technical infrastructure needed to support AI agents:

• Integration platforms for connecting different systems

• APIs for data exchange

• Monitoring and logging capabilities

• Security and compliance controls

5. Establish Governance Framework

Create clear guidelines for agent operation and oversight. This should include:

• Protocols for human review and intervention

• Performance monitoring metrics

• Error handling procedures

• Compliance and audit requirements

6. Manage Change and Training

Change management is crucial for successful implementation. As Rajitha Chaparala emphasizes:

Your change management plan should include:

• Comprehensive training programs

• Clear communication about agent capabilities and limitations

• Regular feedback sessions with users

• Support systems for troubleshooting

7. Start Small and Scale

Begin with a pilot program in a controlled environment. Thiago Da Costa recommends:

Your scaling strategy should include:

• Clear success metrics for the pilot

• Regular evaluation points

• Plan for expanding to additional workflows

• Resource allocation for scaling

8. Monitor and Optimize

Implement continuous monitoring and improvement processes:

• Track agent performance metrics

• Gather user feedback

• Identify areas for optimization

• Update processes based on learnings

9. Plan for Integration

Consider how your AI agents will work together as you expand implementation. Rajitha Chaparala notes:

Think about:

• Workflow dependencies

• Data sharing between agents

• Coordination mechanisms

• Overall system architecture

10. Future-Proof Your Implementation

Ensure your implementation can evolve with technology. As Alan Mosca suggests:

Consider:

• Flexible architecture that can accommodate new capabilities

• Regular technology assessment and updates

• Ongoing training and development

• Innovation pipeline for new use cases

Remember that successful implementation is an iterative process. Start with a solid foundation in data and process management, then build up gradually while maintaining strong human oversight and continuous improvement practices. The goal is not to replace human expertise but to augment it with powerful tools that can handle routine tasks while freeing up people for more strategic work.

The Future Landscape

The future of AI agents in construction promises fundamental changes in how we interact with technology and manage projects. Here's what we can expect to see in the coming years:

Evolution of Interfaces

Traditional software interfaces - with their menus, buttons, and forms - will likely give way to more natural interactions. Construction professionals will increasingly communicate with AI agents through voice commands, text conversations, and even gestures. This shift will make technology more accessible to field workers and reduce the learning curve for new tools.

Proactive Project Management

Future AI agents will move beyond reactive problem-solving to proactive project management. By continuously analyzing project data, these systems will:

• Anticipate potential delays before they impact schedules

• Identify safety risks before incidents occur

• Predict budget overruns while there's still time to correct course

• Suggest resource reallocation based on emerging project needs

• Automatically generate contingency plans for potential issues

Multi-Agent Ecosystems

The next generation of AI agents will work together in sophisticated ecosystems. Instead of single agents handling isolated tasks, we'll see networks of specialized agents collaborating across project phases:

• Design agents working with cost estimation agents to optimize value engineering

• Safety agents coordinating with scheduling agents to manage site logistics

• Quality control agents feeding information to procurement agents for future specifications

• Documentation agents supporting field agents with real-time information updates

Integration Across Organizations

AI agents will increasingly operate across organizational boundaries, facilitating better collaboration between:

• General contractors and subcontractors

• Designers and builders

• Owners and project teams

• Suppliers and contractors

Enhanced Decision Support

Future agents will provide increasingly sophisticated decision support by:

• Simulating multiple project scenarios

• Analyzing complex trade-offs between time, cost, and quality

• Providing real-time impact assessments of proposed changes

• Offering data-driven recommendations for risk mitigation

Impact on Roles and Skills

This evolution will reshape many construction roles:

• Project managers will become orchestrators of AI agent teams

• Field supervisors will leverage agents for real-time decision support

• Document controllers will shift focus from processing to strategy

• Engineers and designers will collaborate with AI for optimization

The future of AI agents in construction isn't just about automation - it's about transformation. These technologies will fundamentally change how projects are planned, executed, and delivered. Organizations that prepare for this shift now will be best positioned to leverage these capabilities as they emerge.

Key Insights and Action Items

The successful implementation of AI agents requires careful planning and execution. Here are the essential takeaways for construction organizations:

First, understand that AI agents represent a fundamental shift in how work gets done. They're not simply another technology tool but a new approach to managing construction processes. This requires rethinking workflows and responsibilities rather than simply automating existing processes.

Second, data quality and accessibility are fundamental to success with AI agents. Organizations need to invest in their data infrastructure and standardization efforts before they can fully leverage agent capabilities.

Third, human oversight remains crucial. While agents can handle many tasks autonomously, they work best when paired with human expertise and judgment. Establishing clear protocols for human-agent interaction ensures optimal outcomes.

For immediate action, you should:

1. Assess current data infrastructure and identify gaps that need addressing.

2. Document key workflows and processes, focusing on areas where agents could add immediate value.

3. Start small with pilot programs in well-defined areas, measuring results and gathering user feedback.

4. Develop clear protocols for human oversight and intervention in agent-managed processes.

5. Invest in training and change management to ensure successful adoption.

The rise of AI agents marks a significant evolution in construction technology. Organizations that approach this transformation thoughtfully and systematically will be best positioned to leverage its benefits while managing associated risks and challenges.

This technology represents not just an opportunity for efficiency gains but a chance to fundamentally improve how construction projects are delivered. The key is to move forward deliberately, with clear objectives and a commitment to maintaining the critical balance between artificial intelligence and human expertise.

WEEKLY MUSINGS
Patagonia Code, AI Zones, Booms and Busts

The secret language of investors

UK's AI action plan

The Katerra effect and its aftermath

OUR SPONSORS

BuildVision — streamlining the construction supply chain with a unified platform for contractors, manufacturers, and stakeholders.