Datadog veterans launched AI coding startup Niteshift on a bet against restricting Big AI

AI coding agent startup Niteshift has raised a $7 million seed round led by Greylock’s Jerry Chen. That’s a modest sum by AI standards, but the startup, founded by two of Datadog’s early engineers, has attracted some big names like Reid Hoffman, Datadog’s Olivier Baumel, Alexis Lee Kwok, Braintrust’s Ankur Goyal, and Reflection AI’s Misha Laskin.

Founded by Sajid Mahmood and Connor Brannagan, who helped grow Datadog from its early days to a multibillion-dollar valuation, the company entered the crowded AI coding space with a compelling idea: Why would any company trust its most sensitive asset — the code that runs its products — directly to modellers like OpenAI and Anthropic, given that these companies are constantly “killing” startups and companies by releasing competing apps?

Mahmoud, the CEO, likens this to Datadog’s early growth, when the monitoring company won e-commerce customers who refused to rely on Amazon Web Services. This was a reasonable concern, because at the same time Amazon was putting many of its retail stores out of business in what became known as the “retail apocalypse.”

The equivalent of artificial intelligence, as Mahmoud sees it, is already underway. Anthropic, OpenAI and others are moving quickly into vertical software markets, what some call SaaSocalypse.

“At Datadog we saw this clearly,” Mahmoud said. “A lot of our multi-cloud business came from e-commerce companies that didn’t want to run on Amazon, right? … We’ll certainly see the same dynamic going for Anthropic to compete in legal, healthcare, finance and whatever else.”

The bet is that companies will increasingly seek infrastructure that separates the coding model from all other formats needed to ensure that AI-generated code is properly inspected and maintained (and that they will want a vendor without a competing agenda).

To be clear, Niteshift does not replace Claude Code or Codex, which are the most popular programming agents. It claims to reduce dependence on them.

Niteshift’s AI Coding Cloud will move between these models – along with open source and other options – based on the needs of each project.

“Being able to switch between GPT and cloud models is important,” Mahmoud said. “Everyone is worried about being attacked by these giant companies.”

This idea is what made Greylock’s Chin bite.

“As frontier labs move higher up, there is an opportunity to provide customers with an alternative path: decoupling their agents from the infrastructure they operate on,” Chen told TechCrunch. “Niteshift is building the platform that enables this for programming agents, allowing customers to invest deeply in their developer tools without restricting themselves to one model or one agent vendor.”

What’s more, Niteshift does not sell tokens. It sells infrastructure, charging like a cloud provider, with per-minute usage rates.

“Everyone is selling intelligence about labor replacement,” Mahmoud said. “We sell software to agents, rather than humans – but we’re still here selling software.”

However, Niteshift is entering a crowded market with AI programming tools. Model independence is not a new idea, and Niteshift’s competitors are off to a strong start. This includes Cursor, although it could be acquired by SpaceX soon; Cognition, which just raised $1 billion at a $26 billion valuation; Amazon Bedrock; and AI gateway platform OpenRouter, which just raised $113 million at a $1.3 billion valuation. The list goes on.

Mahmoud’s answer to all of this is the depth of the founding team. Mahmoud and Branagan didn’t just study these problems, they lived them, scaling Datadog through the increasing difficulties large engineering organizations now face with AI-generated code. Teams need to run, test and verify software independently in their real production environments, he said, and they need infrastructure built by people who have done this at scale.