One AI policy doesn’t fit your whole company. Now it doesn’t have to.

Kilo, an AI coding platform, now lets enterprise admins create sub-orgs to apply separate AI policies, budgets, and model rules per business unit. The update includes one-click sub-org creation, a unified user view, per-sub-org model restrictions, and direct funding via Kilo Pass. Kilo reports that Auto Efficient routing achieves 71% of frontier-level results at 72% lower cost, and open-weight models now handle 81% of real usage, up from 3% a year ago.

Most Kilo customers aren’t one team. They’re a parent company with a dozen business units, each with its own budget, compliance rules, and risk tolerance. One policy for all of them always costs you something. Make it strict enough for your most regulated unit, and everyone else works under rules they don’t need. Make it loose enough for your fastest team, and that’s where the next security incident starts. The fix is to keep governance and spend under one roof, and let each team pick its own models and tools. Kilo now does this one level down, at the sub-org. Here’s what’s new: any admin can spin up a sub-org in one click, see every user across every sub-org from one screen, set model rules per sub-org, and fund each one directly, including with Kilo Pass https://kilo.ai/pricing/kilo-pass . Where the savings come from Kilo already cuts AI coding costs three ways. Now every sub-org gets all three. Smart routing. Auto Efficient reads each coding task and sends it to the cheapest model that can do the job: 71% of frontier-level results at 72% lower cost. Model freedom. Open-weight models now handle 81% of real usage on Kilo, up from 3% a year ago. Mixing open and closed models is how teams get the most work done for the least spend. Kilo Pass. Every dollar you put toward a subscription converts 1:1 into credits, plus up to 50% in bonus credits. Fund a sub-org with Pass instead of a flat top-up, and the same budget buys you more. None of this locks you into one inference path. A sub-org can run through Kilo’s shared gateway, plug in its own keys straight to Anthropic, OpenAI, Google, or Azure, or stay fully local with Ollama or Anaconda Desktop. You choose the routes per sub-org. Spin up a sub-org in one click New subsidiary. New client. New team standing up its own engineering org. Setting up a sub-org used to mean a call with us. Now it’s a button: create it, name an owner, done. See every user, everywhere, from one screen Ten sub-orgs used to mean ten places to check who has access to what. Now you see every user across every sub-org in one view. Add someone, move them, remove access, all without switching screens. Set model rules once, apply them everywhere or not Pick the approved models, providers, regions, and inference routes for every sub-org from one screen. Set a company-wide default, then override it where one team needs something different. Example: one subsidiary has to keep data in Europe https://kilo.ai/eu . Lock that sub-org to European inference options only. Everyone else keeps the full model catalog on Kilo’s gateway https://kilo.ai/gateway . A tighter model policy doesn’t cost you routing. Auto Efficient https://kilo.ai/auto-model still finds the cheapest option inside whatever pool a sub-org is allowed to use enterprise admins can set the model list for Auto Efficient . Fund each sub-org your way Send money from the top down to any sub-org: a flexible balance, a Kilo Pass allocation, or both, sized to match how your budget actually breaks down, not one number for the whole company. However a sub-org is funded, its model rules still apply. Funding never overrides what you’ve restricted it to. Try it today Keep governance and budget at the top. Leave model choice and how the work gets done with whoever’s doing the work, no matter which sub-org they’re in.