Bring your own machine
Use the models on your workstation or GPU server. Choose when, and with whom, to share access.
Your AI compute, working together
Your machines.
People you trust.
More places to run AI.
Reach models across machines your team already owns. Saylek routes inference to available capacity, while each owner’s own work comes first.
An emailed sign-in link. No password.
For small teams, research labs, and technical collaborators. One machine or one invitation is enough to start.
How a request is routed · illustration
saylek
Your applicationSends a request
WorkstationBusy with its own work
Lab machineNot busy, so it serves
Available capacity
- Use machines you already own
- Share with people you choose
- Keep control of your capacity
How it works
Your team’s models, working across machines.
Saylek connects the compute your team already owns, so the people you choose can use it.
01 / CONNECT
Connect machines
Saylek connects workstations and GPU servers and makes the models they serve accessible.
02 / REACH
Access models
Use models on your own machines and those your collaborators have shared with you.
03 / RUN
Run your applications
Send requests from chats, agents, and automated workflows. Saylek routes them to machines that can serve them.
Where your models get used
Connect the applications that need AI.
Coding agents, chat and assistant clients, automated workflows that run unattended, and code your team writes. Each one reaches the models your team’s machines are serving by changing a base URL and a key. Nothing else in the application changes.
Saylek speaks the OpenAI API, at the base URL https://api.saylek.com/v1, and Anthropic’s Messages API, at the base URL https://api.saylek.com. Both procedures are readable without an account: Use your models from anywhere and the Messages-API guide.
docs / connect an OpenAI client
from openai import OpenAI
client = OpenAI(
base_url="https://api.saylek.com/v1",
api_key="YOUR_API_KEY",
)
resp = client.chat.completions.create(
model="MODEL_ID",
messages=[{"role": "user", "content": "hello"}],
)Sharing stays with the owner
Share compute without
giving up your machine.
When sharing is on, your machine can serve AI requests from people you trust. You can keep using it, and your own work is prioritized.
- Choose who can use your models.
- Set the hours when shared requests may run.
- Stop sharing the machine at any time.
- Set priorities for your applications.
Your machine connects out to Saylek. Sharing it does not require opening an inbound port on your network.
With or without a machine
Two ways in.
Start with your machine, or with models your collaborators share.
Use models shared with you
Run requests on models your collaborators make available to you. No GPU or server of your own required.
Create account
Both begin with the same account. After that you connect a machine, or accept an invitation from someone who shares with you.
A few things to know
Before you get started
Who is Saylek for?
Founding teams, small research labs, and technical collaborators who want to use AI models across their own machines and machines shared by people they trust. You can start on your own or use models your collaborators share with you.
Do I need my own GPU or server?
No. You can use models your collaborators have shared with you.
Do I have to share my machine?
No. You can use it yourself and choose whether to share later.
If I share with someone, do they have to share with me?
No. Each owner independently chooses who can access their models.
Make a small start
One machine.
Or one invitation.
See what your team’s compute can do together.
Create account
An emailed sign-in link. No password.