NEWMEN

About Newmen

Your partner, not your supplier.

Frontier labs sell you tokens and move on. They don’t fix the bad responses, they hide what served the call, and they make you wear every regression. Newmen exists because the contract should run the other direction. Every call is cheaper than going direct, you see the model that served it, and you can thumbs-down anything you don’t like for a full refund. Newmen is the platform, Atlas is the optimization layer, and the proof lives on your own traffic.

Values

Three commitments

Your partner, not your supplier

Frontier labs sell you tokens and move on — they don't fix bad responses, they hide what served the call, and they make you wear every regression. Newmen runs the contract the other way. Every call is cheaper than going direct, you see the model that served it, and you can thumbs-down anything you don't like for a full refund.

We train the optimization engine, not a language model

We train Atlas to learn which route, model, and precision hold quality on your traffic at the lowest cost. Your prompts and code are never absorbed into a model, sold, or shared — your IP stays yours. Turn off usage-based optimization any time, or get full data isolation on Enterprise.

Verified on your own traffic

Every claim points at a receipt you can see. Every call shows the model that served it and exactly what you saved, reconcilable against the provider bill. Evaluators score served outputs; anything below your threshold isn't metered. The proof lives on your own traffic, not a benchmark.

The double guarantee

Two guarantees, in writing.

Price: never more than going direct — and you start saving on call number one, at least 5% off and climbing as Atlas ramps. Quality: thumbs-down any call you don't like and we refund it in full — no ticket, no explanation, no argument.

Why now

Intelligence is becoming electricity.

LLMs are commoditizing and capability ceilings are converging across labs. The next axis isn't which model you call — it's whether every call gets cheaper while the quality is proven to hold, on your specific production traffic.

We started Newmen because we kept watching the same pattern repeat at customer after customer: an AI bill that climbs faster than usage, with no clean way to cut it without quietly trading away quality. Buyers were asked to take "it's still good" on faith — no per-call proof, no refund when a call missed, no way to check the math.

Atlas was designed around that gap. It serves each call the cheapest way that holds quality, records every call, and shows the model that answered alongside the direct price and what you paid. Its optimization engine tunes routing to your workload over time, so the discount climbs as it sees more of your traffic — while quality stays the guaranteed thing, not the asserted thing.

Team

Who is here

Placeholder roster for v1. The team is small and on purpose.

Investors

Backing

Careers

Open roles

We are a small team. Every role is a founding role in practice — you will write code, methodology, and customer memos in the same week. Send a short note plus your most concrete work to jobs@newmen.ai.

Founding Optimization Engineer

San Francisco · On-site

  • ·Design and run optimization experiments — which model, route, and precision serve each operation the cheapest way that holds quality.
  • ·Own the evaluator suites that gate every routing change — write rubrics, validate against human labels, publish methodology.
  • ·Partner with platform engineering on the pipelines that turn recorded traffic into a better-tuned optimization engine — never a model trained on customer data.
  • ·Publish at least one rigorous technical report per quarter.
  • ·Strong preference for candidates who have shipped cost-and-quality measurement end-to-end.
Apply for Engineer

Inference Platform Engineer

San Francisco · On-site or remote (NA / EU)

  • ·Own the call-recording, evaluator, and savings-reconciliation surfaces in the Newmen console.
  • ·Build the verification and billing infrastructure — eval scheduling, refund automation, per-call inference manifests, and direct-vs-Newmen price reconciliation.
  • ·Production TypeScript + Next.js + Postgres; comfort with Kubernetes and GPU scheduling a plus.
  • ·You care about latency budgets, error budgets, and writing systems your replacement can understand.
  • ·Bonus: you have rebuilt this kind of pipeline at a previous AI company and have opinions about what you would not do twice.
Apply for Engineer

Enterprise Solutions Engineer

San Francisco or New York · Hybrid

  • ·Embed with enterprise customers from first call through production launch.
  • ·Translate a customer's workload into operations, evaluators, and savings targets — write the first dataset alongside them.
  • ·Run quarterly business reviews focused on verified savings and quality held, not vanity adoption.
  • ·Background in ML platform consulting, ML engineering, or a prior solutions role at an AI company or platform team.
  • ·Strong written communication. The job is half code, half memos to VPs of AI.
Apply for Engineer

Contact

Talk to the team.

We work with AI teams running production workloads who want a smaller bill — proven on their own traffic, not benchmark scores.

30 minutes

We walk through how Atlas would serve a workload that looks like yours — the cheapest path that holds quality, with the savings shown per call. No deck, no canned demo.

Bring a real workload

Come with a production operation — something where the bill is climbing or quality is hard to trust. We'll work through how Atlas optimizes it and how you'd verify the result.

Same day response

A solutions engineer reaches out within one business day. If you need faster, email us directly at sales@newmen.ai.

No sales pressure

We only make money when you save money. We'll tell you if your workload is a bad fit — it saves both sides time.

Talk to a solutions engineer

Atlas is sold to teams who commit to meaningful production volume. That commitment unlocks the reliability loop.