AI agent implementation pricing
Fixed build fees and a flat monthly operating fee. No hidden token billing, no per-seat pricing, and no metered surprises at the end of the month.
Three tiers
Single Agent Pilot
Deploy one high-impact workflow into production.
- 1 custom autonomous agent workflow
- Integration with up to 3 existing systems (CRM, email, database)
- Company Brain grounding & guardrails
- Full monitoring, tuning, and prompt maintenance
Agent System
Full multi-agent orchestration for a complete department.
- System of up to 4 specialized agents
- AI Manager, AI Judge, and Improvement Manager loop
- Deep CRM, ERP, and communication tool integration
- Dedicated uptime SLA, output scoring, and weekly optimization
Enterprise System
Custom multi-agent operations across your enterprise.
- Unlimited agent workflows & custom company brain build
- Hardened security, SOC 2 compliance, and audit trails
- On-premise or custom deployment options
- Dedicated account engineer & 24/7 monitoring
What the build fee covers
The build fee covers everything between a mapped workflow and a running system. Agent design, orchestration, the company brain build, integration with the systems you already run, the scoring rules the AI Judge applies, and the guardrails that decide what a person still approves.
It also covers the unglamorous part: shadow running against live volume, tuning against real exceptions, and the runbook your team gets at handover. A pilot that only works on clean inputs is not a delivered pilot.
What the monthly fee covers
Monthly maintenance covers model usage, vector database hosting, continuous scoring, and ongoing system tuning. It is one number. You are not reconciling token invoices from three vendors at the end of the quarter.
It also covers the improvement loop. Every output the AI Judge rejects becomes a new rule or a test, and those changes ship continuously. That is the difference between a system that drifts and a system that gets better each month.
Model prices move. If underlying costs shift materially we talk about it before anything changes, not after.
Where the scan fits
Every engagement starts with the two-week AI-First Scan, priced separately and scoped to one workflow. The scan stands alone, with no obligation to continue. If the scan says agents are the wrong answer for that workflow, you have spent two weeks instead of a quarter.
What you build on runs on open infrastructure that you own. If we are ever not the right partner, the company brain leaves with you. Nothing is locked inside a product only we can open.
Pick one workflow. We'll price it honestly.
Two-week AI-First Scan first, then a fixed build fee and one flat monthly operating fee.