In 2023, when attending business school at Harvard, friends Brian Elliott and Sid Pardeshi were introduced to a small Boston bakery that wanted an app for customers to place orders. Its owners had planned to spend $300,000 and six months on the project, and the pair offered to take it on as part of a consulting project for school.

They didn’t need six months. ChatGPT had just been released, and the two realized they could combine the strengths of multiple AI models to speed up development. They built the app in a single weekend.

That was the spark that led them to start AI coding company Blitzy in November 2023. Today, Blitzy’s AI is designed to take on entire engineering projects, like building apps or systems. It can ingest a company’s codebase to construct a map of how different components work together before it plans, writes, tests and updates code. That systemic focus gives the AI tool a clear sense of the context surrounding each piece of code, which the founders say enables Blitzy to do large chunks of work autonomously.

It’s a different approach from AI coding agents like Claude Code and Codex, which can search files, edit code and execute commands. But Elliott and Pardeshi argue that these tools, while helpful, still depend heavily on human direction and lack a holistic understanding of how systems fit together.

“Other tools are getting a slightly faster bicycle to an individual. We’re autopilot for a plane,” says Blitzy cofounder and CEO Brian Elliott, 34.

Blitzy doesn’t build its own AI: It coordinates models such as Gemini and GPT-5.5. Over time, it builds up a structured map of how a company’s code and apps all fit together so it can reason about the system holistically and make more informed, end-to-end coding decisions.

“In 2023, the whole world was telling us that the model was going to just do it all,” Elliott says. “Everybody told us we were incorrect. You're either wrong or you're onto something really interesting that people are overlooking.”

Investors are sold. On Tuesday Blitzy announced it had raised $200 million in a funding round led by Northzone at a $1.4 billion valuation, with participation from PSG, Battery Ventures and Liberty Mutual. The round comes after raising $4.4 million in initial capital from Link Ventures, Bessemer, Flybridge, NFX, Picus and Asymmetric in 2024.

Sanjot Malhi, a partner at Northzone who led the most recent funding round, says developers have quickly adopted Blitzy into their workflow. Many customers now start projects on Fridays so Blitzy can run through the weekend, leaving engineers to review results on Monday, he says. In one instance, a single project lasted over a month, yielding 500,000 lines of code that shipped to production without requiring a single manual edit, according to Elliott.

Another customer had planned a modernization project to update its system that would have required hiring 300 engineers over two years, at an estimated cost of $50 million to $60 million annually, Malhi says. With Blitzy, the company completed the project in six weeks, without adding a single engineer.

“This is amongst the strongest feedback I've heard on any product from a customer in my career,” says Malhi. “It literally takes all of this context, goes away for seven days and comes back with a 90%-ready product.”

In large organizations, the same code is reused across thousands of applications, says CTO Pardeshi, 33. But AI systems lack visibility into those relationships. That means a small change can quietly break systems elsewhere, sometimes only surfacing once the code goes into production. Recent concerns reinforce that risk: Anthropic’s Claude Opus 4.6, for instance, has reportedly declined in coding quality, with cybersecurity experts warning that developers relying on such models may fail to catch flaws or serious defects before deployment.

Of course, Anthropic (valued at $380 billion), OpenAI (valued at $852 billion) and Cursor (which agreed to be acquired by SpaceX for $60 billion or pay $10 billion) are extremely well capitalized competitors that have grown exceptionally fast. Anthropic, for example, has experienced unprecedented growth, with an annualized revenue run rate surpassing $30 billion in April, up from roughly $9 billion at the end of 2025. Blitzy is still nascent by comparison. Malhi isn’t worried, though.

“It competes with everything, but also it competes with nothing because nobody else can do this," he says.

Elliott says Blitzy’s starting packages range from $1 million to $10 million a year, charging 20 cents per line of code (the company declined to share revenue figures). Its customers include large enterprises in industries like finance and insurance, such as Builders FirstSource, one of the country’s largest suppliers of construction materials with $3.4 billion in 2025 net sales, and software consulting firm Galatea Associates, which works with top Wall Street investment banks.

Enrique Ibarra, CIO of Grupo Nacional Provincial (GNP), one of Mexico’s largest insurance providers, says Blitzy was a challenge to use at first. Engineers had to learn how to effectively structure prompts and define requirements for the system, with the help of Blitzy’s founders. But while the developers were initially skeptical, he said they were quickly won over after seeing the system handle 95% of the work. “Some of them were not expecting that level of quality,” Ibarra says.

Sid Pardeshi’s fascination with software started early during an internship at Nvidia, where CEO Jensen Huang regularly circulated research papers on the rapid advance of AI—an early signal, Pardeshi says, that machines would one day build and run much of the world. He left in 2022 after accumulating 25 patents in early forms of generative AI to pursue a dual master’s in business and engineering at Harvard. There, he met Elliott, a former U.S. Army chief of staff who previously founded two startups, including an at-home engagement ring try-on service and a chatbot-based therapy platform.

If AI can build software largely on its own, Elliott argues, companies can tackle far more ambitious problems faster than ever before.

“When you're able to design and build a system that can deliver autonomous outcomes, you are moving GDP,” says Elliott. “That's a world worth living in, and that's a mission worth dedicating your life to.”

MORE FROM FORBES

ForbesMeet The $250 Million Startup Challenging Salesforce With AI AgentsForbesThis $1.5 Billion AI Startup Steps In When Software BreaksBy Sofia ChierchioForbesInside Suno’s $2.5 Billion Bet That AI-Made Music Is Here To StayBy Rashi Shrivastava