What Our Clients Say

Don helped us tackle a backlog of over 90,000 images from our camera trap network — a review process that would normally require multiple passes by volunteers and staff. He evaluated leading AI wildlife identification platforms, tested them against our ‘truth set,’ and identified the best options for accuracy, scalability, and ease of use. His work reduced the amount of manual review required while keeping quality and accuracy high, turning what felt like an impossible task into a process that allowed us to catch up on images for the first time ever — likely saving us weeks worth of work. We were so happy with Don’s solution that we have shared it in presentations to highlight how much more work we were able to accomplish.

Don took what started as a fuzzy scope and turned it into decisions we could actually stand behind. He ran discovery sessions with our team, customers, and even our customers' customers to figure out what we really needed. Then he documented everything as high-level user stories and helped us prioritize using MoSCoW scoring. When it came to evaluating platforms, Don didn't just create another vendor comparison spreadsheet. He built a scoring model that made sense, ran pilots with three different vendors, and did his own technical testing to prove what could (and couldn't) work with our existing platform. To align our team on requirements that were agnostic of any specific platform, Don utilized an AI prototyping tool (Onlook) to generate two prototypes: one for the end-user experience and another for the business-user portal. This provided everyone with a working, end-to-end view of the product before the build. We ultimately chose to build in-house, reducing risk and churn, because the trade-offs were explicit and the MVP roadmap was clear.

For our ERP launch, we had to audit 20K+ images from 10 years of artwork, a job that would have taken weeks of manual work. Don built an image matching system that cross-referenced our files against our artwork list and flagged only what needed review. The result: 94% of our files matched automatically, with the human-review app used only for edge cases. It turned a tedious, weeks-long slog into a process we could finish in days.