I am an Italian tech founder and software engineer, currently living in Lisbon. I love building products that people use every day.
Currently, I am working on Typefully, an AI-powered writing app for social media with hundreds of thousands of users worldwide.
In the past I also worked on various other projects:
- Mailbrew - Personal email digests (excited!)
- Unreadit - Semi-automated newsletters based on Reddit
- Boxy Suite - Expanded Boxy into a suite with subscription
- Boxy - Mac web wrapper for Gmail Inbox
- Libretto - Grades tracking app
I have a background in engineering and computer science and was employee number 30 at Bending Spoons, before leaving to do my own startups.
Writing
With so many agents poking at your repos and injecting everything into context, it's more important than ever to protect passwords and secret keys. Something that works really well: instruct agents to use `op` (the 1Password CLI) to read secrets directly from your vaults. No more .env files lying around for the next AI agent tool you try next week to grab. Plus, authorizing sensitive operations with your fingerprint? Chef's kiss. Using this for all my scripts, want to transition the big repos soon to a similar approach. Gist with sample script: https://gist.github.com/frankdilo/398c54fe082076cbee45a0f6a6260226
I am probably late to the party, but after this talk, I want to give Amp Code a proper try. 1/ Sub-agent architecture where a coordinating agent picks the right model for you. Curious to see if this works in practice and removes the overhead of picking models. 2/ Next level TUI. 3/ You can share conversations with teammates.
Flew less this year, which has been a great life improvement if you ask me.
