Limits per ticket and per agent
Technology and software shouldn't come with surprise bills. While agentic spend can vary based on complexity of existing configuration and difficulty of the work you assign to the agents, you can control your spend ceiling in advance. In knowvation, spend limits are set for each ticket and for each agent, and the platform enforces the limits. Work is paused when spend limits are reached. Administrators review and choose to resume, or not, tickets that hit this limit.
Cost you can see
For every ticket, knowvation shows tokens and dollars spent by the Business Analyst, Architect, Developer and QA Engineer agents, so you know exactly what it cost to deliver the work.
An example ticket
An example: a story to require a manager's approval for discounts over 20%.
- Business Analyst: $0.71 (212k tokens)
- Architect: $0.98 (288k tokens)
- Developer: $1.86 (547k tokens)
- QA Engineer: $0.93 (271k tokens)
- Agentic delivery: $4.48, against a ticket limit of $15.00
- Human delivery, estimated: 1 to 2 hours, $100 to $200
These numbers are illustrative.
Pick the model for each agent
Each agent's persona supports model selection from the model providers you choose to configure. Use a larger model where judgment matters more and a smaller one where it doesn't. Using the right model for the right job is important to balance quality and spend.