[01]
Open infrastructure
The core auction server is open source.
Inspect it. Run it yourself. Extend it. Build on top of it.
Own how your media is bought.
An open platform for agencies and brands to run their own buying logic directly against publisher supply.
The bidder is open. The models are open. The docs are open. Anyone can get started.
Bring your own data, models and agents — or start with ours.
Join the waitlist
We’re building the first version now. Sign up for early access and help shape it.
( 01 / 03 )
Why
Programmatic should be open. Today, much of the intelligence behind media buying sits inside platforms you cannot see, change or own.
We want to change that.
[02]
Open models
Start with open models for common media decisioning problems, then train or fine-tune them on your own data.
Your models remain yours.
[03]
Your agents
Use our always-on campaign agent or bring your own.
Manage campaigns from the tools you already use, including Claude and ChatGPT.
[04]
Direct supply
Buy through clear paths to publishers, without unnecessary layers or reselling.
[05]
Clear economics
See what the media costs, what the technology costs and where margin is being made.
[06]
Built for efficiency
Less unnecessary compute means lower infrastructure cost, lower latency and lower carbon.
( 02 / 03 )
Open
Built in the open.
- [A]Open bidder.
- [B]Open models.
- [C]Open APIs.
- [D]Open docs.
No private sales process required to get started.
Anyone can sign up, experiment and build.
Join the waitlist
( 03 / 03 )
FAQ
Questions, answered.
Something we haven’t covered? Join the waitlist and tell us what you need.
01What are you building?
An open execution platform for programmatic media.
Bring your own models and agents, run them directly against publisher supply, and own the decisions between campaign objective and impression.
02Is this a DSP or an SSP?
It overlaps with both, but we are not trying to recreate either in the traditional sense.
We are building an open layer for media decisioning and execution.
03What is open source?
The core auction and bidding infrastructure is open source.
We also release an open family of media decisioning models, along with the tools and examples needed to run them.
04Do I need to build my own models?
No.
You can start with open models provided by the community and fine-tune them using your own data.
More advanced teams can bring completely custom models.
05Can I train models on the platform?
Yes.
The platform includes infrastructure for training, evaluating, fine-tuning, testing and deploying models into live buying.
06Where do AI agents fit?
Agents manage the longer-running work around a campaign.
They can monitor performance, adjust settings, run experiments, retrain models, investigate problems and produce reports.
You can use our agent or connect your own.
07Can I use Claude or ChatGPT?
Yes.
The platform is API-first, so agents are native operators rather than an interface added afterwards.
You can manage a campaign from Claude, ChatGPT or another compatible agent.
For example:
“Check this campaign every morning. If CPA rises above £20, investigate why and tell me before making a major change.”
08What happens to my logs?
They are yours.
Auction logs, model decisions, campaign activity and agent actions are available to you without additional logging or data-access charges.
09What does direct supply mean?
Every impression should have a clear path back to the publisher.
Supply can still come through infrastructure partners, but we want to avoid unnecessary reselling and opaque supply chains.
10How can agencies make money?
Transparency does not mean agencies cannot make margin.
Platform fees, optimisation fees, model fees and performance-based economics can all be explicitly built into campaigns.
11When will it be available?
We are building the first version now.
We are looking for agencies, brands, publishers, developers and adtech teams interested in helping shape it.