Niteshift raises $7 million to make AI coding agents less sticky

Niteshift, founded by former Datadog engineers, raised a $7 million seed round led by Greylock and is selling infrastructure for AI coding agents. Its bet is not another autocomplete, but the ability to switch between GPT, Claude and open source models when the model provider becomes a competitor.

Former Datadog engineers are building a cloud layer for coding agents

Niteshift was founded by Sajid Mehmood and Conor Branagan, both former early engineers at Datadog. TechCrunch reports that the company raised a $7 million seed round led by Greylock’s Jerry Chen. Angels include Reid Hoffman, Datadog’s Olivier Pomel and Alexis Lê-Quôc, Braintrust’s Ankur Goyal and Reflection AI’s Misha Laskin.

The product is not framed as a direct replacement for Claude Code or Codex. Niteshift is building an AI coding cloud that routes work across different models, including GPT, Claude and open source options.

The business model sounds closer to infrastructure than token resale. Niteshift talks about per-minute usage rates and a layer for orchestration, testing, vetting and maintenance of agent-generated code.

Lock-in hurts when the model vendor starts selling the application too

The Datadog analogy works because the issue is not only price. Mehmood points to e-commerce companies that avoided AWS because Amazon was also their market rival. Niteshift argues that the same pattern is emerging around AI models.

For enterprise teams, code is a more sensitive asset than an ordinary prompt. If an agentic workflow attaches itself to one model vendor, a technical choice becomes a commercial dependency. Switching models is not just changing an endpoint. It affects evals, tests, permissions, logs and review flow.

That is the control Niteshift is selling. It is not claiming to own the best model. It is saying customers should not let the best model also own the whole traffic hub for code work.

A multi-model layer can be insurance, or just another middleman

The weak spot is obvious: routing across models is easy to promise and hard to do well. Each model has different behavior, tools, context limits and failure modes. An abstraction that smooths too much may hide exactly the differences a reviewer needs to see.

The market is also crowded. Cursor, Cognition, Claude Code, Codex, Bedrock and OpenRouter already press on different parts of the stack. A $7 million seed is a solid trust signal for the team, but in this category it can vanish quickly into product, infrastructure and enterprise sales.

Real portability on a live repo will decide the pitch

The next proof will not be a launch post. It will be migration. If Niteshift shows that a team can move an agentic workflow across models without breaking tests, audits and review queues, the pitch starts to make economic sense.

Pricing is worth watching too. Per-minute billing sounds cloud-like, but buyers will want to compare the cost of a completed pull request, not the time an agent spent running.

Lilith's verdict

Niteshift is selling an emergency exit from a house where the model vendor also rents the rooms and changes the locks. If that exit only leads to another hallway with a startup logo, enterprise teams will notice fast.