Avolve AI
Avolve AI
Avolve AI: Building an AI-Powered Document Intelligence Platform for Architecture & Engineering
A production-grade, microservices-based AI platform that automates the classification, validation, and intelligent analysis of architectural plans and technical documents, built by Folder IT’s nearshore AI engineering team for a US and UK market leader in permitting and plan review software.
Avolve Software is a US and UK-based technology company specializing in permitting, licensing, and land management software for government agencies and enterprise clients. As demand for AI-powered document automation grew across the architecture, engineering, and construction (AEC) sector, Avolve launched an ambitious initiative to build Avolve AI — an intelligent platform capable of automating the reading, classification, and validation of architectural plans and technical documents at scale. The project required a specialized nearshore AI engineering team with deep expertise in generative AI, computer vision, and cloud-native architecture — operating in US business hours and embedded directly into Avolve’s product and engineering organization. Folder IT was selected to lead the technical integration, AI model calibration, and architecture of the platform from the ground up.
Avolve AI
The Challenge
Architectural plan review is one of the most document-intensive processes in government and enterprise operations. Plans arrive in high volumes, in varied formats, and must be validated against strict normative and jurisdictional requirements — a process traditionally dependent on manual review by specialized engineers.
Avolve needed to automate this process at scale, addressing four core technical challenges:
- Intelligent document classification — distinguishing architectural plans from support documents, and segmenting pages by discipline (architecture, electrical, structural, plumbing, and others).
- Structured data extraction — automatically detecting critical elements such as north arrows, scales, engineering stamps, and approval whitespace areas.
- Normative compliance validation — verifying that submitted plans meet ICC building code requirements without manual intervention.
- Scalable AI inference — processing large document volumes asynchronously, without interruptions, across a distributed microservices architecture.
0
Average AI Recall
0
Months Roadmap
0
AI Modules Delivered
The Solution
Folder IT deployed a dedicated nearshore AI engineering team — operating in US business hours — to design, build, and validate the full Avolve AI platform, from inference architecture to AI model development and continuous improvement.
The team led technical integration meetings with Avolve’s strategic technology partners — including Microsoft, ESRI, and Ultralytics — ensuring platform compatibility and scalability across every technical front.
The platform was built on a modern microservices architecture (.NET 9.0, RabbitMQ, Docker, MongoDB) with asynchronous processing, enabling high-volume document handling without performance degradation. All AI inference endpoints are standardized under OpenAPI 3.0, ensuring traceability and reproducibility across every task.
Development is managed under Agile/Scrum methodology, with a structured Jira board organized in epics and sprints — coordinating AI, engineering, and product teams across a 12-month, four-phase roadmap designed and documented entirely by Folder IT.
The team led technical integration meetings with Avolve’s strategic technology partners — including Microsoft, ESRI, and Ultralytics — ensuring platform compatibility and scalability across every technical front.
The platform was built on a modern microservices architecture (.NET 9.0, RabbitMQ, Docker, MongoDB) with asynchronous processing, enabling high-volume document handling without performance degradation. All AI inference endpoints are standardized under OpenAPI 3.0, ensuring traceability and reproducibility across every task.
Development is managed under Agile/Scrum methodology, with a structured Jira board organized in epics and sprints — coordinating AI, engineering, and product teams across a 12-month, four-phase roadmap designed and documented entirely by Folder IT.
Team Composition
- 1 VP & Chief Technical Officer
- 2 Senior ML Engineers
- 2 Senior AI Engineers
- 1 Senior DevOps Engineer
Key Capabilities
Delivered Document Classification Automatic distinction between architectural plans and support documents, with page-level segmentationResults
- GPT-5-Mini achieved 88% recall in scale detection and 85% in north arrow detection across real architectural production documents.
- Full microservices inference architecture deployed — enabling asynchronous, high-volume document processing without performance degradation.
- Successful geospatial PoC validating AI-based parcel and street detection in subdivision plans using segmentation models.
- End-to-end QA framework implemented with smoke, regression, and scalability testing — plus automated performance metrics per task type.
- Complete 12-month product roadmap designed and documented by Folder IT, currently guiding Avolve’s technical development and resource planning.
- ACIP program active — continuous model retraining and feedback loop ensuring sustained accuracy across production environments.
Technologies
Backend & Architecture
AI & Machine Learning
Image Processing & Geospatia
Devops & Methodology
