about
About WonderTwin.
Closing the dev loop that AI agents broke.
section 01
How this started
WonderTwin emerged from a problem that followed Tela through his career — Bloomberg Tax, Avalara, Moz, and BearTax. The same wall every time: mocks and sandboxes that didn’t behave like production, tests that flaked on someone else’s schedule, a dev loop that broke whenever a third-party service had a bad day.
Now, with AI agents hammering those sandboxes at machine speed, the problem has gone from chronic annoyance to critical infrastructure gap. This is the problem WonderTwin solves.
For years, Joshua and Tela have been thinking about the same class of problems: how a system comes to know the world through its tools, and how it closes the gap between observation and action more efficiently and accurately than previous generations of tools. When they realized they could twin the world an agent reaches for — every external service, local and behavioral, running at machine speed — they built WonderTwin to close the loop.
section 02
How we work
Five things we keep coming back to.
-
Principle 01
Open source first .
The full twin catalog ships under MIT license. Always will. Pro extends the catalog with adaptive behavior — drift detection, certified SDK compatibility, known-issue intelligence — but the foundation stays free.
-
Principle 02
Architecture over policy .
Privacy is built into the architecture. Telemetry is sanitized on your machine before transmission. What reaches our servers is anonymous by construction.
-
Principle 03
Adaptive twins .
A twin is a running model of the real service. Pro twins adapt and calibrate as the service evolves. OSS twins are static snapshots.
-
Principle 04
Customer friction is the build queue .
Early-customer friction surfaces as roadmap. We treat every error report from a trial customer as a higher-priority signal about what to build next.
-
Principle 05
Founder-personal until it isn’t .
While we’re small, every customer trial gets one of us in the inbox within 24 hours. We’ll keep it up for as long as it’s the right answer.
section 03
Max & Dot
Two characters emerged from our development of adaptive twins. Every agent reaches into the world through its tools, and what it does next depends on how well it sees and understands what comes back. Max moves first. Dot watches and remembers. Neither works right without the other. Now they live in the CLI, in the error messages, in the site you’re reading, in the conversations we have about the work. They embody closing the loop.
section 04
How to reach us
GitHub
github.com/wondertwin-ai
Stars, issues, PRs — the loop is open.
hello@wondertwin.ai
For anything else.
Contact paths
- Customer support: Open an issue on GitHub or hit reply to any WonderTwin email you’ve received from us.
- General, press, or partnership inquiries: hello@wondertwin.ai
- Security disclosures: security@wondertwin.ai (See our Privacy Policy for response SLA.)