Specialty Retail / Small BusinessAI Implementation
Custom AI Customer Service Agent for Wood Thumb, a San Francisco Woodshop
Wood Thumb, a San Francisco woodshop and design studio, was evaluating off-the-shelf customer service chatbots running $79-$165/month, tools that weren't purpose-built for their business and required ongoing SaaS subscriptions. Our team designed and built a custom AI agent integrated with their website and Gmail, using the Anthropic API with a FastAPI backend. The agent handles customer inquiries, appointment context from their scheduling system, and common questions about services and lead times: all trained on the shop's actual voice and offerings. The result: a purpose-built solution for roughly $5-20/month in API costs, with full ownership of the tool and no vendor dependency.
Annual savings against the commercial alternatives
Custom to their voice, workflows, and data
Vendor lock-in, since they own the code
Mechanical Engineering / Plumbing DesignAI Implementation
Plumbing Takeoff Tool for a Mechanical Engineering Firm
A mechanical engineering firm was taking off plumbing by hand: tracing each sheet for pipe runs, reading sizes off the callouts, and tallying lengths by system into a spreadsheet, a job that ran for weeks across a project. We built a takeoff tool that reads the plan PDF, extracts every pipe segment as a vector with its length and size, reads the callouts, and totals the runs by system, such as sanitary waste, grease waste, and hot water. The extracted lines sit over the original drawing, the engineer clicks any pipe, callout, or junction to inspect it, and segments the tool cannot classify are held out for review rather than rolled into a total. The engineer's time on each sheet now goes to reviewing the extraction rather than tracing it.
Pipe segments extracted from a single sheet, with lengths and sizes
Totals for sanitary waste, grease waste, and hot water, by pipe size
Inspectable and reclassifiable on the drawing, with unclassified runs held for review
AEC Technology / Design IntelligenceAI Product Strategy
Aané: AI-Powered Architectural Detail and Material Search
Partnered with Aané (aane.io), an early-stage AEC software company building an AI-powered architectural detail and material search platform. The core problem: architecture firms were losing institutional knowledge when project teams turned over, and spending hours re-researching details and materials they'd already solved. We contributed to product strategy and ICP positioning: helping the team articulate why 'reduce research time by 80%' is the right lead metric, and how to sequence onboarding to demonstrate value before asking for firm-wide adoption. The platform now serves architects, interior designers, and engineering firms as a searchable knowledge repository with AI-enabled design intelligence.
Visit aane.ioResearch time reduction, the core product claim
Architects, interior design firms, engineering teams
Time to value as the positioning anchor
AEC Industry IntelligenceData Engineering + Live Dashboards
AEC Hub Data Intelligence Platform
Built and maintain 7 live data dashboards pulling from BLS, FRED, Census Bureau, FHWA, and USAspending.gov, covering construction employment, material prices, building permits, bridge conditions, federal infrastructure spending, site risk, and AEC salaries across 44 occupations and 393 metro areas. Each dashboard is validated against source data, updated automatically, and built to serve real decisions, not just display numbers. The platform now attracts AEC professionals, researchers, and firms using it as a primary reference for market intelligence.
Salary records across 44 roles and 393 metros
Federal and public data sources integrated
