Work.
In the field.

A modular manufacturer was experiencing systemic design-to-fabrication gaps — engineering intent lost in translation between the parametric BIM environment and shop floor execution. Rework and quality escapes were eroding margins on every production run. The team had strong parametric design capability and a capable manufacturing operation, but nothing connecting the two with any reliability.

Unauthorized modifications to approved parametric models were also creating audit trail gaps and compliance exposure, with no systematic way to detect, document, or prevent them.

We built the AI intelligence layer connecting parametric design outputs to shop floor execution — structured as a crawl / walk / run adoption framework to de-risk the deployment and earn stakeholder confidence at each phase. The Compliance Agent was deployed first, ingesting internal SOPs, fabrication standards, and quality gate requirements, then monitoring every shop drawing and fabrication output for deviations. The Modification Guard agent was layered on top, enforcing change control and creating immutable audit trails for every parametric model change. The Digital Thread agent was deployed last, connecting BIM outputs to MES actuals for the first time.

A major enterprise retail operator needed to transition 300+ facilities from static BIM files to operational digital twins. Manual facility onboarding was taking weeks per building, BIM model quality was inconsistent, and there was no governance enforcing the naming conventions and metadata standards that operational digital twins actually require to function at scale.

We architected and built the fully automated ACC → Tandem → Crystal pipeline using n8n agentic automation — handling OAuth authentication, facility creation, sequential model imports, and bi-directional asset linking. A custom naming convention schema was designed for the portfolio, and BIM standards compliance enforcement was implemented at the pipeline ingestion boundary, meaning non-compliant models were flagged and returned before entering the digital twin environment.

Construction draw review is one of the most document-intensive processes in real estate finance — inspectors, lien waivers, AIA G702/G703 forms, budget reconciliation, and compliance against dozens of project-specific covenants. Doing it manually at scale consumes underwriting bandwidth and creates inconsistent risk exposure. The platform needed an AI agent that could handle the compliance layer systematically.

We translated the platform's construction lending SOPs into structured agentic workflows — designing the AI Draw Agent to ingest inspection reports, validate document completeness against loan covenants, flag non-compliant draws, and escalate to human review only when genuinely ambiguous. The Compliance Agent was adapted to treat loan covenants as the constraint document set.