Yet Another AI That Ships Code?
The line between developer and AI is officially blurred.
Ona, formerly Gitpod, has built an autonomous AI software engineer that does more than just suggest code – it runs the entire development loop.
Ona, the evolution of the cloud development tool Gitpod, is introducing autonomous AI software engineering agents. This isn't another autocomplete tool. These agents operate independently within secure, cloud-based environments to pick up GitHub issues, write code, run tests, and submit pull requests. It is the full development loop, handled by an AI.
This is where the line between a tool and a teammate blurs. Internally, Ona claims its AI agents co-authored 60% of pull requests and contributed 72% of merged code on their own main branch. The real story here isn't about replacing developers, but fundamentally changing their role into that of an architect and reviewer for a team of tireless AI agents. We are moving from writing code to defining the problems the AI should solve.
This shift has massive implications for productivity and how software teams operate. The focus moves from manual coding to high-level problem-solving and quality control. For any dev team or leader, the question is no longer *if* you will adopt AI, but how you will integrate and manage AI agents as part of your core workflow.
The No-Code AI App Rush
The race is on to let non-developers build their own AI-powered tools, moving beyond generic solutions.
CREAO: Build the exact AI app you need with plain English
CREAO lets you build custom AI-powered software using just natural language. This is about creating hyper-specific tools that fit your exact workflow, not forcing your team into a generic SaaS box.
Empromptu: Go from idea to full-stack AI app in minutes
Empromptu focuses on creating full-stack, production-ready AI apps without code. It aims to kill the prototype phase by letting you build and deploy robust tools that can handle real work from day one.
YouMind: The AI-powered studio for your brain dump
This is an all-in-one AI studio for turning scattered ideas and research into polished content. YouMind acts as an IDE for creators, integrating research, writing, and multimedia into one project-based workflow.
Building The Boring AI Plumbing
The AI magic trick is getting old, and now the hard work is happening to make it reliable and genuinely useful.
Gram by Speakeasy: Teaching LLMs to finally use your API correctly
Gram solves the unglamorous but critical problem of making LLMs reliably use your APIs. It translates messy REST APIs into a clean format AI agents can actually understand and execute, which is essential for real automation.
GoSearch Free: Your personal search engine for work
GoSearch connects to over 100 work apps to create one unified, searchable brain. It's a practical solution to digital chaos, letting you find information and trigger actions without switching tabs constantly.
SAMMY Guides: The invisible AI guide for any software
This 'invisible AI agent' provides real-time, interactive software walkthroughs directly on your screen. SAMMY tackles the universal pain of user onboarding, aiming to make complex software intuitive from the first click.
Quick hits
Agilepitch: Your HubSpot CRM just got an AI brain
An AI sales coach plugs into HubSpot to analyse your pipeline and tell reps exactly what to do next to close deals.
The AI Code Conundrum: That perfect code snippet might be an AI hallucination
Developers are finding AI-generated answers on Stack Overflow, forcing a new level of scrutiny and validating the age-old advice: trust, but verify.
From Junior Dev to Tech Lead: It is not just about the code you write
A developer's five-year journey to tech lead reveals that architectural vision and communication skills, not just coding, are the keys to career progression.
My takeaway
Focus on higher-level thinking: system design, user experience, and strategic product decisions. The tools that win will be the ones that integrate seamlessly into this new human-AI collaborative workflow. The big question is how we adapt our skills to manage agents, not just write lines.
Are we training ourselves for this new reality, or are we still focused on optimising for the old one?
Drop me a reply. Till next time, this is Louis, and you are reading Louis.log().