2025 to now · Wells Fargo, COO AI Product
Prompt Studio
Enterprise prompt operations for Global Operations: governed design, QA, chaining, and execution across lines of business. ~380,000 hours / year by 2027. Patent pending, co-inventor.
Scope the job with the business before anyone writes a prompt.
How it was built
I scoped each use case with the line of business, then designed the prompt pack, the policy bounds, the maker-checker path, and the admin model that puts a team on the studio.
How it scaled
The same studio now runs across businesses. I ran onboarding and workshops for 100+ analysts. The patent-pending no-code workflow is the path to the ~380,000 hour projection.
Key features
Policy-bound prompt packs, chaining with tool calls, maker-checker review, usage and quality analytics, and enterprise onboarding.
- Scope the job with the business before anyone writes a prompt.
- Lock which models, tools, and data the pack may use.
- Write the prompt chain and any tool calls.
- A second person checks the output before it ships.
- Provision access and walk the team through first use.
- The team runs the pack inside the studio.
- Track quality and usage on a modular dashboard.
- Repeat the same studio in the next line of business.
- Train analysts so they can run their own cases.
- The no-code workflow generates, manages, and runs prompts.
- The hours model is the 2027 projection from that workflow.