Can We Trust AI-Written Code? Why Executable Specs Are the Future of Software ft. Gabriela Moreira @ColorsofTechEntrepreneurship
Can We Trust AI-Written Code? Why Executable Specs Are the Future of Software ft. Gabriela Moreira  @ColorsofTechEntrepreneurship
Uploaded April 2026 | Updated September 2026, 10 hours ago
In this episode, we sit down with Gabriela Moreira, CEO of Quint Co and a passionate advocate for formal methods in software development. Gabriela shares her journey from compiler research and teaching formal methods at university in Brazil to building Quint — a language that makes formal specification accessible to everyday developers. She explains how Quint was born inside Informal Systems to solve a real problem: TLA+ was powerful but too hard for most engineers to read and adopt.

The conversation dives into how AI is changing the economics of formal verification. Writing specs used to require learning a new language and investing significant time upfront, but with LLMs, developers can now generate Quint specifications in hours instead of weeks. At the same time, as AI writes more and more of our code, the need for confidence and correctness has never been higher — making executable specifications more relevant than ever.

Gabriela also walks through real-world use cases, including AWS's use of TLA+, Circle's Arc L1 blockchain built on Quint-verified consensus, and work with Monad's BFT protocol. Whether you're building smart contracts, distributed systems, or any software where bugs can cost real money, this episode will change how you think about the gap between "it works on my machine" and "I know it's correct."

⌛ Timestamps:
00:00:00 Episode trailer
00:00:36 Show & guest introduction
00:02:04 From compilers and type systems to TLA+ and formal methods
00:06:23 Gabriela's first encounter with AI and data science
00:08:32 Gabriela’s hobbies
00:09:54 The origin story of Quint and Informal Systems
00:13:29 Deep dive into Quint and formal specs
00:19:20 Real-world use cases: Malachite, Circle's Arc, Monad BFT
00:27:42 Biggest misconceptions engineers have about formal specs
00:33:35 Bugs that traditional testing misses but Quint catches
00:38:22 Formal verification for smart contracts and Web3 security
00:41:02 Where to follow Gabriela and Quint
00:42:02 Episode outro and next episode preview

Follow Quint Co & Gabriela Moreira:
Gabriela's Twitter/X: https://x.com/bugarela
Quint Twitter/X: https://x.com/quint_lang
Quint Github: github.com/informalsystems
Gabriela's LinkedIn: linkedin.com/in/bugarela?originalSubdomain=br
Gabriela's Github: github.com/bugarela

Our social media:
Crypto card guide: cryptocardguide.com
Twitter/X: x.com/ColorsofWeb3pod
LinkedIn: linkedin.com/company/colors-of-web3-entrepreneurship

If you enjoy this episode, please subscribe to stay updated on our upcoming episodes with industry experts and innovators in the Web3 and Entrepreneurship spaces. Thank you for listening!
Can We Trust AI-Written Code? Why Executable Specs Are the Future of Software ft. Gabriela MoreiraKanishks method to onboard 1 billion users into Web3Chess game on the blockchain (using SNARK)Organic Web3 Recruiting: Building Authentic Connections feat. Kornel Kot from W3CONN3CTThe need for specialized data labeling in AIHow Emily first learned about the Web3 space (thanks to Bitcoin)EtherFi Cash: The On-Chain Card Taking On Banks | feat. Slater Heil - Head of MarketingThe fundamental UX problem with blockchainWhat does a KYC (Know Your Customer, or permissioned) blockchain mean #web3 #avalanche #kycAbhijats main advice for Web3 foundersThe harsh crypto environment in which Fyde Treasury carried out their fundraising2 brief stories of Bitcoin exactly 1 decade apart - 2014 vs. 2024
Colors of Tech & Entrepreneurship |

Can We Trust AI-Written Code? Why Executable Specs Are the Future of Software ft. Gabriela Moreira

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER