You pay per conversation. Not per token.
A conversation handled with the Niadra memory counts once, no matter how many reads or searches the agent makes. What you spend on your AI model is not part of the bill, and the compact context helps you lower it.
Production
per thousand conversations, depending on volume
Fully managed. You connect the SDK, and Niadra runs all the infrastructure.
- Every channel, agent and vendor
- Context and history search, with no per-conversation limit
- Full Console, with the read log
- AES-256-GCM and TLS 1.3, with an exclusive key per company in an HSM
- Support during business hours
Regulated
for banking, insurance and healthcare
A dedicated environment for your company, also run by Niadra.
- Servers, database and keys exclusive to your company
- SSO through SAML 2.0 or OIDC, and logs in your SIEM
- Contractual availability SLA
- Security review with your team
- Extended support, with a direct line
What each plan includes
Questions about pricing
What counts as a conversation?
A conversation in which the agent used the Niadra memory. Ten reads and five searches inside the same conversation count as one. A conversation that never used Niadra is not charged.
Do you charge per token, per search or for storage?
No. Only per conversation. If the bill followed the amount of text, Niadra would earn more by sending larger contexts, which is the opposite of what you want.
Is there a volume discount?
Yes. The US$ 5 to 15 range is exactly that: the more conversations per month, the closer you get to US$ 5.
How do I start?
Through the contact form. You talk to the people who write the code, and the proposal comes out of what you are building. Then you install the SDK: the infrastructure is on us.