Quickstart
How an evaluation starts and what the first month looks like.
There is no package to install. Saifuro attaches to production traffic and decides whether a spend is allowed, so evaluation starts with a conversation. What follows is what the first month actually looks like.
Write to us
contact@saifuro.com. Useful to include: which agents spend money today, roughly how much, through which providers, and what worries you about it.
Technical call
We walk through the system on real data, not slides: how a mandate is written, what a denial looks like, what lands in the log. Bring whoever will own the policies.
Draft policies
Agents, spend categories, limits, mandate owners. This usually takes one working session, and the first draft is deliberately loose.
Observe mode
Saifuro attaches to production traffic and blocks nothing. Integration tooling is issued at this point, along with credentials for each agent that will call us, so a mandate attaches to an identity from the first request.
Report and calibration
After about thirty days you get the analysis: what would have been denied, where the rules would have fired wrongly, where money went outside the categories anyone expected, and measured authorization latency on your own traffic. Policies are adjusted against that. Evaluation is free; nothing is invoiced until the service agreement takes effect.
Enforcement
Blocking switches on per policy class, starting with hard limits.
More detail on why this shape in getting access and observe mode.