Usage & cost control
AI fair use & usage limits
Yes—if AI usage were unlimited, model tokens and runtime cost would rise with every workflow. We design hard scope and monthly allowances so expenses stay predictable.
Why limits exist
Large language models charge by tokens (roughly the amount of text read and written). Extra loops, huge documents, voice minutes and retries all increase cost. Unlimited “run forever” pricing would either bankrupt the provider or force surprise invoices—exactly what DIY credit meters do. We avoid that with fair-use caps.
Website demos
- Default: local demo brain only — no billable third-party model calls.
- Optional live AI is off unless explicitly enabled for testing.
- Do not paste real secrets into demos.
First Workflow Pilot
- Fixed fee covers design, connection, testing and go-live for one agreed workflow.
- Build-time model usage for that scope is included.
- Out-of-scope features, extra workflows or uncontrolled high-volume load during pilot are quoted before we run them.
AgentOps Monthly (standard allowance)
Unless your statement of work says otherwise, the starting monthly plan includes:
- One live workflow (one primary agent job)
- Up to 5,000 agent runs / month (a run = one triggered handling of a case/message/task)
- Reasonable document size per run (guide: under ~50 pages of text equivalent unless quoted)
- Monitoring, approval support and monthly review
Voice minutes, multi-agent packs, or very large file pipelines are separate allowances when scoped.
What happens near the limit
- Alert — we notify you around ~80% of the monthly run allowance.
- Protect — non-urgent or bulk jobs can be queued or rate-limited so costs don’t spike overnight.
- Choose — upgrade the monthly tier, add a quoted top-up, or tighten the workflow (shorter prompts, fewer retries, human batching).
Overage (if you opt in)
We do not silently rack up open-ended token bills. Overages run only with your written OK, typically as:
- a same-month top-up block of runs, or
- a move to the next AgentOps tier.
Your controls that save money
- Human approval gates before expensive external sends
- Clear job boundaries (what the agent must not do)
- Prefer structured tools over stuffing entire mailboxes into prompts
Related
Pricing · Acceptable use · Terms · Discuss limits for your volume
These pages are provided for transparency. They are not a substitute for legal advice. If you need a signed contract for a pilot or AgentOps retainer, we issue a written statement of work.
Last updated: 14 August 2026