Become a design partner.
60 days of Fairmeter free, with the router tuned around your traffic instead of somebody else’s.
Built for the first ten teams.
60 days of Fairmeter free (bring your own provider keys; Fairmeter does not cover your provider bills) across every current capability primitive plus every new feature that ships during the trial.
A live capability editor and a session dashboard wired to your real agent traffic from day one.
A weekly telemetry rollup covering total spend, token and cache totals, p99 latency, and breakdowns by model, complexity, and thinking budget. Sent to you directly until the dashboard view for it ships.
A direct line to the founder in a shared Slack channel for routing tweaks, incident triage, and product road map input.
One onboarding call to set up the partnership and get your traffic flowing.
Early access to new features and capabilities behind a feature flag before the public beta.
Light by design.
No recurring calls, no standing meetings. Just the minimum signal we need to tune routing against your real traffic and produce a credible case study at the end.
Written feedback at the end of the 60-day trial covering top wins, top frustrations, and feature requests.
Consent to share anonymised usage data for a case study. Team and company name are optional.
Willingness to be quoted on the landing page after the case study lands.
Willingness to test pre-release builds behind a feature flag.
Every Monday, one row for your org.
Every Monday the gateway rolls up your prior seven days across cost, tokens, cache, latency, and routing, and writes one row for your org. Your numbers come from the gateway's own session tables, so they reconcile against your invoice rather than being computed separately. The dashboard view for it is not built yet, so for now we send you the rollup directly.
Example values below. The field names are what the gateway emits. The numbers are invented, not a customer and not a measured result.
{
"request_count": 18342,
"total_cost_usd": 4287.42,
"total_input_tokens": 41293807,
"total_output_tokens": 6118442,
"total_thinking_tokens": 2841920,
"cache_hit_rate": 0.61,
"p99_latency_ms": 4127,
"model_distribution": [
{ "model": "claude-haiku-4-5", "requests": 11204, "cost_usd": 612.88 },
{ "model": "gpt-5-mini", "requests": 4918, "cost_usd": 1043.20 }
],
"complexity_distribution": [
{ "label": "simple", "requests": 9820 },
{ "label": "moderate", "requests": 6411 }
],
"thinking_budget_burn": [
{ "label": "none", "requests": 12750, "thinking_tokens": 0 },
{ "label": "standard", "requests": 4102, "thinking_tokens": 2841920 }
]
}What follows is a modelled example of the shape a partnership takes, built from the same routing arithmetic the home page prices step by step, because somebody deciding whether to apply should be able to see the shape of the thing before they talk to us. Read it as a model of what the mechanism does at a given spend level rather than as a result somebody achieved. 60 days into your trial, the numbers on your own dashboard replace it.
Hypothetical fintech startup, Series B, twelve engineers
$84K → $27K−68%
Monthly agent spend reduced across a 60-day pilot.
Every figure in this box is modelled rather than observed. There is no such company, there was no such pilot, and no customer has run this. The 68 percent also sits a little above the best result we have actually measured, which is 66 percent against a top-tier reasoning baseline — read it as the optimistic end of the model rather than a typical outcome. The measured numbers are on the savings methodology page.
Ready to talk?
All ten slots are open, and the first one carries the most influence over what gets built. Drop us an email and we will get your onboarding call on the calendar within the week.