We've been building our dream teammate Charlie since early '25. Nice to meet you!
The small and mighty team of seasoned experts building (and increasingly assisted by) Charlie.
Riley sets Charlie’s direction and leads the architecture behind its agent runtime. He keeps the product thesis and technical core in the same loop, connecting models, tools, devboxes, reliability, and customer evidence. Before Charlie, he engineered at Dropbox, cofounded Flight, which was acquired by Figma, and founded Dexa AI.
Jerrod closes the loop between customer learning and company decisions. He connects trust, economics, positioning, and rollout to design partnerships, customer success, go-to-market, and operations. Before Charlie, he led Codecov through its growth from open-source roots to its acquisition by Sentry and worked in venture investing and company-building at FundersClub.
Ryan makes the hard parts of agent work coordinated, testable, and improvable. He turns ambiguous behavior and recurring production failures into systems for routing, durable tasks and mailboxes, evaluation, diagnostics, and optimization. Before Charlie, he worked on Google Cloud Storage, prototyped a code-generation tool with early access to Google’s internal language models, and led LLM applications to chip-design verification at ChipStack.
Bart turns Charlie’s asynchronous capabilities into a product teams can see, understand, and control. He gives that work a clear path into the customer experience, from onboarding and integrations to Activity and the dashboard. Before Charlie, he worked in product engineering at Modulo and Kraken and on Caesars Digital’s online casino product.
A few proud milestones along the way:
The goal is not to add another inbox or require people to manage a fleet of bots. Good engineering leverage lets a team turn direction into progress with less status-chasing, handoff work, and coordination tax.
A useful system should leave behind artifacts people can inspect, discuss, approve, revise, and learn from. Transient chat is useful for interaction; it is not a sufficient record of consequential engineering work.
Capability without context is unreliable. Charlie should work from the systems where the relevant truth lives, verify what it can, make its actions observable, and stay inside explicit boundaries.
Charlie can carry out bounded work and surface useful evidence. People decide what matters, set constraints, review important outcomes, and remain accountable for decisions that carry meaningful product, technical, or organizational consequences.
We prefer small, legible, iterative systems that do a few important things well. Strong defaults should make the safe and useful path easy without hiding how the system works.