Skip to content
Talk to us ↗

AI for every team. One set of rules.

Astraed deploys and operates your private AI gateway. You control models, access, and spend.

Talk to us ↗

Explore Astraed AI Gateway

Teams management screen with create and search controls, five teams, counts of members, models, and access keys, and each team's spend against budget Teams management screen with create and search controls, five teams, counts of members, models, and access keys, and each team's spend against budget

Give each team the right access.

Approve models and set a budget for each team. Review its members, access keys, and spend in one place.

Scroll the screenshot to explore.

Demo configuration

View full size ↗ View full size ↗

Every connected tool follows your rules.

Requests from your workspace, agents, and coding tools pass through the gateway before reaching an approved model.

  1. 01 · Request

    Legal team

    “Summarize the confidentiality clause in this NDA and flag anything we should change.”

  2. 02 · Rules

    Checked in the gateway

    Key
    Legal team key
    Model
    claude-sonnet-5-5, approved
    Personal data
    Masking rule applied
    Budget
    $25.31 of $500 used
  3. 03 · Response

    Answer, with its cost

    Route
    Approved outside provider
    Cost
    $0.00078
    Recorded to
    Legal team budget

Illustrative request using the demo configuration.

Blocked

Same key, asking for gpt-5.5. Not approved for Legal, so the request is refused and logged.

Your choice of models. Your approval.

Choose approved outside providers or models on your own servers. Availability follows your deployment and privacy requirements.

Match each workload to the right level of privacy.

Route each team to approved outside providers, dedicated European servers, hardware at your site, or a clear combination of these choices.

  1. 01

    Outside providers that agree not to store your data

    Use selected providers only when their contracts prohibit storing prompts and files or using them to train models.

  2. 02

    Private servers in Europe

    Run models on dedicated servers in the EU, Switzerland, Norway, or Iceland.

  3. 03

    At your site

    Run local models and the gateway on hardware you own or lease.

One endpoint for the workspace, agents, and code.

Connect tools that support an OpenAI-compatible endpoint using your gateway address and a team access key.

Python, with the OpenAI library
import os
from openai import OpenAI

client = OpenAI(
    base_url="https://ai.your-company.com/v1",
    api_key=os.environ["LEGAL_TEAM_KEY"],
)

reply = client.chat.completions.create(
    model="claude-sonnet-5-5",
    messages=[{"role": "user", "content": "Summarize this NDA."}],
)
print(reply.choices[0].message.content)

One subscription. Clear usage costs.

Operation, support, and maintenance are included. Outside model usage is passed through at provider cost, with no markup.

Setup
One-time fee, quoted after assessment
Subscription
Annual, priced by infrastructure, privacy requirements, and number of employees
Outside models
Monthly, at provider cost, no markup
On-site models
No inference charge for models running on your premises
Hardware
Optional lease under a separate contract

Questions teams ask about the gateway.

Which models can we use?+
Catalogue models that meet your privacy requirements, plus models on your own servers. Offline systems use local models only.
Where do prompts and logs live?+
Gateway logs live on your dedicated server or at your site, under your retention settings. Requests routed to an outside model are also processed by that provider. Astraed's support access follows your authorization.
Do outside providers keep our prompts?+
Approved routes run under terms that prohibit storing prompts or training on them. Workspace history and logging follow your own settings.
What happens when a team reaches its budget?+
Requests on that budget stop until an administrator raises the limit or the period resets.

Map your teams, models, and budgets.

A short call to map your teams, privacy needs, and budgets. Pilot scope is agreed with you.

Talk to us ↗