🧑💻 Developer-First #157 - AI is coming for legacy code recurring revenue
What Morgan Stanley just did with AI should terrify IT services firms
Hello friend,
Morgan Stanley has developed an AI agent that translates legacy code into precise functional specifications in natural language. In an interview with the Wall Street Journal, Michael Pizzi, the bank’s Head of Technology and Operations, shared that this model has already reviewed over 9 million lines of COBOL and Perl code this year. That’s the equivalent of 280,000 developer hours saved.
What thousands of developers and IT service providers have struggled to do profitably for decades —modernising banking infrastructure —may now be solvable in months with the help of AI. And this is not a niche use case. Legacy systems are the number one bottleneck to innovation in large enterprises, especially in the financial services sector. Without clearing this technical debt, true architectural modernisation remains impossible.
My post on the topic (in French) sparked a wide-ranging debate on LinkedIn, especially among developers with real-world experience working on legacy codebases. Here’s what stood out:
Scepticism is high: Many developers are unconvinced that LLMs can reliably extract specifications from legacy code, especially in critical sectors like banking.
But they do see potential: Some engineers pointed out that even partial success can be valuable. Using LLMs to extract architecture diagrams, list dependencies, or summarise functionality could dramatically accelerate modernisation projects.
Human-in-the-loop will be key: Several agreed that full automation is unrealistic (for now), but combining AI output with expert oversight could make a big difference.
The threat to IT services is real: Many noted that what’s at risk isn’t just the task of code review, it’s the entire consulting revenue tied to legacy system maintenance. As one comment put it, “The business model of many IT service companies will need to be reinvented fast.”
Now, let's dive into last week's developer-first transactions.
💰 Market Summary - Week of June 9th, 2025
7 companies raised $212 million across 5 product categories in 3 countries.
Europe-based companies attracted 0% of the total funding vs 11.3% for Asia-based companies (including Israel) and 88.7% for North America-based companies.
Software Delivery Lifecycle is the category that attracted the highest funding.
4 companies provide or contribute to an open-source product.
1 company was acquired this week.
🧩 Funding by Product Category
🌎 Funding by Region
🏢 Funding By Company
Linear, from San Francisco 🇺🇸 raised $82 million in Series C funding led by Accel. Linear is building an issue tracking and project management tool that empowers product teams by combining UI elegance with world-class performance. (more)
PostHog, from San Francisco 🇺🇸 raised $70 million in Series D+ funding led by Stripe. PostHog provides a fully integrated, open-source product platform including analytics, feature flags, heatmaps, session recording, and A/B testing. (more)
Antimetal, from New York 🇺🇸 raised $20 million in Series A funding led by Sound Ventures. Antimetal is building tools to automate the post-deployment software lifecycle so teams can spend more time building and less on maintenance. (more)
CloudQuery, from Tel Aviv 🇮🇱 raised $16 million in Series A funding led by Partech. CloudQuery extracts data from databases and SaaS apps into data lakes and visualisation tools, with a plugin-based open-source architecture. (more)
Infisical, from San Francisco 🇺🇸 raised $16 million in Series A funding led by Elad Gil. Infisical is an open-source, end-to-end encrypted tool that helps developers manage secrets and environment variables securely. (more)
Hirundo, from Tel Aviv 🇮🇱 raised $8 million in Seed funding led by Maverick Ventures Israel. Hirundo is the first machine unlearning platform designed to remove unwanted data and behaviours from AI models. (more)
Lightpanda, from Paris 🇫🇷 raised an undisclosed pre-seed funding led by ISAI. Lightpanda is a browser designed for headless usage, enabling fast web scraping and automation with minimal memory usage. (more)
🤝 Mergers & Acquisitions
Scale, from San Francisco 🇺🇸 received a $15 billion investment from Meta, with Scale’s founder, Alexandr Wang, joining Meta to work on its AI efforts (making the transaction an acquisition in practice). Scale powers the most advanced LLMs and generative models in the world through world-class RLHF, data generation, model evaluation, safety, and alignment. (more)
⭐️ Trending GitHub Repositories
sindresorhus / awesome (✩ 3,739 stars this week) - 😎 Awesome lists about all kinds of interesting topics.
tensorzero / tensorzero (✩ 2,321 stars this week) - TensorZero is an open-source stack for industrial-grade LLM applications. It unifies an LLM gateway, observability, optimisation, evaluation, and experimentation.
eyaltoledano / claude-task-master (✩ 2,239 stars this week) - An AI-powered task-management system you can drop into Cursor, Lovable, Windsurf, Roo, and others.
👍 Trending Developer Tools on ProductHunt
Bubble for native mobile apps (beta) (👍 1,102) - Bring your mobile vision to life without code — on a platform that offers everything you need to design, build, test, and publish to the app store. Users are already launching successful apps for sports training, travel, medical records, and word games.
Solar (👍 556) - Solar lets small teams collaborate with AI agents to build and deploy production apps in minutes. Like Figma, but no mockups — just real software with live data and backend functionality, built by our full-stack AI agent, powered by your imagination.
PageAI (👍 453) - The best AI website generator that can plan, design and code your landing page in minutes. Uses open source, well-documented components, builds wireframe, writes copy and selects colours, fonts and allows customisation with 1 click. Get the source or deploy!