Meet Macroscope: An AI Tool to Understand Codebases and Fix Bugs

2025-09-18

Founders Who Sold Live Video Startup Periscope to Twitter Return With New AI-Focused Venture Former Periscope founders Kayvon Beykpour and team have launched Macroscope, a new startup dedicated to AI solutions tailored for developers and product leaders. The system aims to streamline codebase updates and identify errors efficiently. The company was co-founded by Beykpour alongside childhood friend Joe Bernstein, who previously worked on Periscope and Terriblyclever — a startup acquired by Blackboard in 2009. Rob Bishop, another co-founder known for selling his computer vision startup Magic Pony Technology to Twitter in 2016, also joins the team. Officially incorporated in July 2023, Macroscope positions itself as an “AI-powered understanding engine” designed to save engineers valuable time — a tool the founders wish they had during their early entrepreneurial journeys.

“Engineers today spend too much time managing workflows using tools like JIRA, Linear, and spreadsheets, often at the expense of actual development,” Beykpour explained. “Macroscope was built to solve that problem directly.”

Beykpour shared in an interview with TechCrunch: “I’ve experienced this pain in every company I’ve worked with — whether a startup or a large public company like Twitter. We’ve all faced this challenge the hard way.” He added, “Understanding what every engineer is doing, especially in a large organization like Twitter with thousands of developers, was a major part of my role — and definitely the part I disliked the most.”

To address these challenges, Macroscope clients start by installing its GitHub app, granting the system access to their code repositories. Additional integrations like Slack, Linear, and JIRA apps can be added. The software then analyzes code changes and provides insights.

The system performs a process called “code walking,” utilizing abstract syntax trees (ASTs) — structured representations of code — to gather contextual understanding of clients’ codebases. This knowledge is then combined with large language models (LLMs) to generate actionable insights.

Once operational, engineers can use Macroscope to identify and fix errors in pull requests (PRs), summarize PRs, receive updates on codebase changes, and ask code-related questions. Product leaders can gain real-time summaries of product updates, productivity metrics, and natural language answers regarding product development or engineering activities, enabling better prioritization of engineering tasks.

“You can ask natural language questions regardless of your technical skill level,” Beykpour emphasized. “It’s incredibly useful if you want to understand a codebase without interrupting senior engineers. For example, a CEO asking, ‘What did we accomplish this week?’ can either consult Macroscope or disturb colleagues — one option is far more efficient than the other.”

While no single product offers the exact combination of Macroscope’s capabilities, the startup competes in the code review space — where developers inspect and test code before implementation — against tools like CodeRabbit, Cursor Bugbot, Graphite Diamond, and Greptile. According to internal benchmark tests, Macroscope detected 5% more real-world errors than the next best tool across over 100 test cases. It also reduced review time by 75%. Macroscope’s pricing starts at $30 per active developer per month, with a minimum of five seats. Enterprise pricing and custom integrations are available for larger organizations. The software requires GitHub Cloud to operate. Even before its official launch, the product is already in use by several startups and major companies, including XMTP, Things, United Masters, Bilt, Class.com, Seed.com, ParkHub, and A24 Labs. Headquartered in San Francisco, Macroscope employs a 20-person team. The company recently closed a $30 million Series A funding round led by Lightspeed’s Michael Mignano in July. Additional investors include Adverb, Thrive Capital, and Google Ventures. To date, Macroscope has raised a total of $40 million.