TL;DR: Repeated external service calls, not model tokens, are often the biggest AI agent infrastructure cost, Archestra reports. Agent workflows need usage limits, caching and retry controls because downstream search, scraping and storage bills can quickly outweigh the model layer.
Editorial analysis by NHI Mgmt Group, based on content published by Archestra: “When the LLM is not the expensive part”.
Key questions
Q: Why do AI agents create unpredictable SaaS and API spend?
A: Because agents do not behave like humans.
Q: What are the warning signs that agent economics are getting out of control?
A: Look for flat token spend paired with rising external service bills, higher retry counts, frequent cache misses and more calls per task over time.
Q: How should teams control AI agent costs when tool calls matter more than token usage?
A: Treat every external call as part of the agent’s cost envelope.
Practitioner guidance
- Map every downstream service call Inventory each external dependency an agent can invoke, including search, scraping, transcription and retrieval, then classify which calls are required, optional or duplicative.
- Set hard usage guardrails Apply per-task and per-session limits on retries, query fan-out and call volume so the agent cannot turn one request into uncontrolled service consumption.
- Introduce caching for repeated lookups Cache normalised retrieval results and common search outputs so the same content is not purchased repeatedly within a single workflow or across similar sessions.
Bottom line: The article’s central point is that AI agent cost often comes from repeated external service usage, not from the model itself.
Glossary terms: Agentic Cost Leakage, Tool Call Fan-Out, Usage Governance
Explore further
View Full Forum → | NHI Foundation Course → | Our Services → | Read the full analysis →
Agent infrastructure economics are now a governance problem, not just a procurement problem. When an agent can trigger many paid services in one workflow, the organisation is effectively delegating spend authority to software. That means access scope, retry behaviour and service selection all influence cost exposure. In identity terms, the question is not only what the agent can reach, but how often and under what constraints it can do so. Practitioner conclusion: cost controls need to sit alongside authorisation controls.
A few things that frame the scale:
- The average enterprise SaaS platform connects to 42 or more third-party applications through OAuth tokens, API keys, webhooks and automation platforms.
A question worth separating out:
Q: How can organisations compare model costs with downstream tool costs?
A: Compare them at the task level, not the invoice level. A model may look cheap while the surrounding search, scraping, vector and speech services dominate total cost. The useful metric is the fully loaded cost of completing one workflow, including retries and fallbacks.
👉 Read our full editorial: AI agent infrastructure cost is driven by tool calls, not tokens