AI coding agent startup Niteshift has raised a $7 million seed spherical led by Greylock’s Jerry Chen. That’s a modest sum by AI requirements, however the startup, based by two former early Datadog engineers, has attracted some big-name angels like Reid Hoffman, Datadog’s Olivier Pomel and Alexis Lê-Quôc, Ankur Goyal of Braintrust, and Misha Laskin of Reflection AI.
Based by Sajid Mehmood and Conor Branagan, who helped develop Datadog from its early days to a multi-billion valuation, the corporate has entered the crowded AI coding area with a compelling concept: Why would any firm belief its most delicate property — code that runs its merchandise — on to mannequin makers like OpenAI and Anthropic, on condition that these corporations are consistently “killing” startups and companies by launching competing apps?
Mehmood, who’s CEO, likens it to Datadog’s early development, when the monitoring firm gained e-commerce prospects who refused to construct on Amazon Internet Providers. It was an inexpensive concern, on condition that Amazon was concurrently placing a lot of those self same retail shops out of enterprise in what grew to become often known as the “retail apocalypse.”
The AI equal, as Mehmood sees it, is already underway. Anthropic, OpenAI, and others are transferring quick into vertical software program markets — what some are calling the SaaSocalypse.
“At Datadog we noticed this clearly,” Mehmood mentioned. “An enormous a part of our multicloud enterprise got here from e-commerce companies who didn’t wish to run on Amazon, proper? … We’re completely going to see the identical dynamic as Anthropic goes to compete in authorized and healthcare and finance and no matter else.”
The guess is that corporations will more and more search infrastructure that separates the coding mannequin from all the opposite orchestration wanted to make sure AI-generated code is correctly vetted and maintained (and that they’ll desire a vendor with no competing agenda).
To be clear, Niteshift isn’t changing Claude Code or Codex, the 2 hottest coding brokers. It argues that it reduces dependence on them.
Niteshift’s AI coding cloud will route between these fashions — together with open-source choices and others — primarily based on the wants of every undertaking.
“Having the ability to change between GPT and cloud fashions is vital,” Mehmood mentioned, “All people’s apprehensive about getting stepped on by these giants.”
That concept is what acquired Greylock’s Chen to chew.
“Because the frontier labs transfer up the stack, there’s a possibility to supply prospects an alternate path: unbundling their brokers from the infrastructure they run on,” Chen advised TechCrunch. “Niteshift is constructing the platform that allows this for coding brokers, letting prospects make investments deeply of their developer tooling with out locking themselves right into a single mannequin or agent vendor.”
Greater than that, Niteshift isn’t promoting tokens. It sells infrastructure, charging like a cloud supplier, with per-minute utilization charges.
“All people else is promoting labor alternative intelligence,” Mehmood mentioned. “We’re promoting software program to brokers, versus people — however we’re nonetheless out right here promoting software program.”
Even so, Niteshift is getting into a crowded market of AI coding instruments. Mannequin independence isn’t a novel concept, and Niteshift’s opponents have a large head begin. That features Cursor, though it could soon be gobbled up by SpaceX; Cognition, which simply raised $1 billion at a $26 billion valuation; Amazon Bedrock; and AI gateway platform OpenRouter, which just raised $113 million at $1.3 billion valuation. The record goes on.
Mehmood’s reply to all of that’s the founding staff’s depth. Mehmood and Branagan didn’t simply examine these issues — they lived them, scaling Datadog by means of the precise rising pains that giant engineering organizations now face with AI-generated code. Groups, he mentioned, must run, check and confirm software program autonomously of their actual manufacturing environments, and so they want infrastructure constructed by individuals who’ve executed it at scale.
Once you buy by means of hyperlinks in our articles, we may earn a small commission. This doesn’t have an effect on our editorial independence.
