AI for construction planning and infrastructure projects

AI for construction planning and infrastructure projects

Sacyr, one of the largest construction companies in Spain, launched an AI-assisted planning pilot with Crata AI that interprets the project's technical documentation and generates the construction schedule automatically. The system extracts activities with 95% accuracy and produces the schedule in minutes, where the manual process took weeks. The pilot ran on residential building projects.

Planning at the pace the work changes

Sacyr is one of the largest construction companies in Spain, with four decades of experience, more than 14,000 professionals and a presence in over 15 countries across four continents. Its work spans civil engineering, building and infrastructure, across projects of very different scales and kinds.

Building a construction plan could take weeks, and every time the technical specifications changed it had to be redone by hand. With planning tied to manual work that had to be repeated at every change, the process became slow and hard to scale to the pace of the projects.

The larger and more changeable the project, the heavier that rigidity grows. The time spent redoing schedules competed with the rest of the technical work, and adapting the schedule to each revision of the specifications became a bottleneck right when flexibility mattered most.

An AI-assisted planner that interprets the documentation and builds the schedule

With Crata AI, Sacyr launched an AI-assisted planning pilot that interprets the project's technical documentation, generates the construction schedule automatically and updates it when the specifications change.

The system reads and structures the technical documentation, identifies the activities that make up the work and builds the schedule from them, ready to export to the formats the team already uses. Where each revision once meant rebuilding it by hand, the system now prepares the bulk of the work and leaves the team to handle the final validation and adjustments.

The pilot ran on residential building projects and covers the full planning cycle:

  • Reads and structures the project's technical documentation, such as specifications and project reports.
  • Extracts the activities that make up the work.
  • Flags missing activities and scheduling risks.
  • Generates the construction schedule automatically.
  • Updates it when the specifications change.
  • Exports the schedules to formats compatible with the team's tools.
Turn your technical processes into AI agents

Long, repetitive technical tasks slow your team down and get redone at every change. At Crata AI we build AI agents that read, interpret and handle the heavy lifting so your team can focus on deciding.

View this solution →
95%
Extraction accuracy
when extracting the activities that make up the construction schedule
<1%
Of the manual planning time
of the time it used to take to build the plan by hand
+90%
Of the initial groundwork
of the initial planning work is prepared by the AI

The technology behind the planner

The pilot combines several layers of technology that work together to interpret the technical documentation, reason over it and build the schedule.

  • Language models (LLM): interpret the project technical documentation.
  • RAG (retrieval-augmented generation): connects the models to the project documentation so the schedule draws on its actual content.
  • Document processing: reads and structures the technical documents so the rest of the system can work with them.
  • AI agent chains: coordinate the steps of reading, planning and updating the schedule from start to finish.
  • Processing with Python: runs the logic that builds the schedules and exports them to compatible formats.

If your team is exploring AI for planning and agents for technical processes, Crata AI can help you design it.

Book a free diagnostic session to talk through your case: Talk to our experts

Contact us at info@crata-ai.com

Share this case study