Reporting dashboards, AI workflow automation, software setup, and web design. Real work, built for small businesses
This client operates a portfolio of insurance brands that serve diverse communities across the U.S., and one of those brands generates a constant stream of new policy paperwork that has to be checked for completeness before it turns into a liability. We built a pipeline that pulls new policy documents in every morning, reconciles each one against the underlying transaction records, and runs a scalable, multi-rule AI compliance audit on every packet, verifying thousands of documents in minutes instead of hours, all before 9:30 AM with no one touching it. A second AI model checks every score independently, so anything the two disagree on gets flagged for a person instead of quietly slipping through. It runs on a single cloud server sized correctly for the job, one that spins up, works through the day's batch, and shuts itself down, and the whole system was built and put into production in 30 days. On day one, before the system had learned a single pattern, it already surfaced over $44,000 in premium tied to policies with zero compliance documentation on file.
Client identity confidential. Part of the same insurance holding company engagement featured further down this page
A local small business was drowning in emails and falling behind on client onboarding. We built a Google Apps Script system that runs daily: an AI agent scans the inbox, prioritizes urgency, drafts replies, and logs everything to Google Sheets. When a new client contract is detected, a full onboarding workflow fires automatically.
Full HoneyBook setup for a service business: client portal, branded proposal and contract templates, automated follow-up sequences, AI-assisted note-taking during calls, and scheduler integration. New clients get onboarded automatically.
Built for a local agency administering residential energy grants across two state programs. Tracks project status, funding utilization, and reimbursement pipelines, with filterable views, a live county-level choropleth map, and real-time breakdowns by status and income level.
Georgia's leading South Asian wedding planning company needed more than a website. They needed a full business operation built from scratch. We designed and launched the site, then built every system behind it: HoneyBook configured end-to-end for bookings, contracts, and client communications; Google Drive and Docs organized for client-facing document management; 1099 payroll set up for their team; and AI automations that calculate hourly pay automatically, eliminating manual tracking entirely.
A live session display built for a Johns Creek, GA nonprofit. It replaces manual bulletin boards with a self-updating schedule that counts down to the next event automatically. No staff input, no daily maintenance, no one responsible for keeping it current. Also configured Wave Apps to automate payroll processing and W-2 generation, eliminating manual year-end prep for the organization's staff.
Client identity masked. Standard practice for nonprofit and community organization engagements
A financial services holding company that runs a portfolio of insurance brands serving diverse communities across the U.S., pairing local market expertise with national operational scale. The organization provides the infrastructure, technology, and strategic leadership required to grow insurance service businesses at scale: acquiring and investing in agencies, delivering shared services and compliance, and enabling scalable growth without compromising brand identity.
Client identity confidential. Standard practice for enterprise and financial services engagements
Every time this client's contract gets signed, an email arrives, and everything downstream happens without anyone touching it. A folder gets created to file the correspondence, a matching folder gets created in the shared drive, and a full set of client documents gets copied from a template, customized, renamed, and sent out. The system encodes the essential facts, the client's name and the engagement date, directly into a structured ID, so the automation can pull them back out and use them everywhere they're needed: a bookkeeping entry for the contract amount and the next payment due, a scheduling entry so a manager can assign the right team, and a hold placed on the calendar so nothing else can ever get booked over it. AI only gets involved when the automation hits something it cannot confidently parse on its own, a name or a pattern that does not fit the rule, and even then it asks one narrow question instead of guessing. Once an engagement wraps, the same system closes out the bookkeeping and scheduling entries on its own, so the records stay current without anyone going back to clean them up.
Client identity confidential. Standard practice for active client engagements