AI Agent Operational Lift for Curis Design in Indianapolis, Indiana

Why AI matters at this scale

Curis Design operates in the mid-market architecture segment (201-500 employees), a size band where project complexity often outpaces process maturity. The firm specializes in healthcare and commercial projects — sectors with intense regulatory scrutiny, demanding client expectations, and thin margins. At this scale, AI isn't about replacing architects; it's about compressing the tedious, error-prone phases of design and documentation that consume 60-70% of project hours. With $45M estimated annual revenue and likely hundreds of active projects, even a 15% efficiency gain translates to millions in recovered billable capacity.

The architecture industry has been slow to adopt AI beyond visualization, but the data foundation already exists. Years of Revit models, BIM 360 coordination logs, and specification libraries represent a proprietary dataset that can train models for code compliance, clash prediction, and generative layout. For a firm founded in 2020, the tech stack is likely modern and cloud-connected, lowering the integration barrier. The key is to start with high-ROI, low-risk applications that don't require perfect data on day one.

Three concrete AI opportunities with ROI framing

1. Automated code compliance review. Healthcare projects must comply with FGI Guidelines, NFPA 101, and local amendments. Manual review by senior architects costs $150-250/hour and often happens late in design, causing expensive rework. An NLP and computer vision pipeline that scans Revit models and PDF submissions against digital codebooks can flag 80% of violations at schematic design. For a $20M hospital wing, catching life safety issues early can save $200K+ in change orders and prevent 4-6 weeks of schedule delay. The ROI is immediate and measurable.

2. Generative design for room and department layouts. Healthcare planning involves complex adjacency requirements, patient flow optimization, and strict dimensional clearances. Generative algorithms can produce and rank thousands of layout options against program requirements, energy performance, and cost metrics in hours instead of weeks. This not only accelerates the design phase but also strengthens fee proposals by demonstrating data-driven optimization to clients. Firms using TestFit and Hypar for similar tasks report 40-50% reduction in test-fit turnaround.

3. AI-assisted specification and RFI management. Writing specs and responding to RFIs during construction administration are high-volume, repetitive tasks. An LLM fine-tuned on the firm's past project manuals and MasterFormat standards can draft Division 01-34 specs from model data, then assist project architects in drafting RFI responses by retrieving relevant details and submittal history. This can cut spec production time by 30% and reduce RFI turnaround from 5 days to 2, improving contractor relationships and reducing liability from delayed responses.

Deployment risks specific to this size band

Mid-market firms face a "valley of death" in AI adoption: too large for ad-hoc experimentation, too small for dedicated data science teams. The primary risk is talent — architects who understand both design and data are rare. Mitigation involves partnering with AEC-focused AI vendors (e.g., Autodesk Forma, UpCodes) rather than building in-house. Data governance is another concern; healthcare projects involve protected health information (PHI) that must never leak into public LLM training sets. All AI tools must run in a private cloud or on-premises environment with strict access controls. Finally, change management is critical: senior designers may resist tools that appear to threaten their expertise. Positioning AI as a "compliance co-pilot" and "design accelerator" rather than an autonomous designer is essential for adoption.