The LLM Is the Cheapest Part: Why Enterprises Now Sell You the “Agent Harness”
Last Updated on May 29, 2026 by Editorial Team
Author(s): Vivek Acharya
Originally published on Towards AI.
“Harness” went from slang to paradigm in three weeks
A client asked me a fair question last quarter: if a single model call costs a fraction of a cent, why is the agent platform quoting $2 per conversatio, and charging it even when the bot fails and hands the customer off to a human?
The article argues that the “agent harness” is where real complexity, cost, and risk live—around the model—because the model alone can’t reliably handle tool use, timeouts, ordering, or production constraints. It traces how “harness engineering” became the dominant framing in early 2026 as benchmark gaps narrowed but environment and orchestration gaps widened. It explains the harness as an operating-system layer (permissions, routing, context management, crash handling) that vendors bundle as the actual product, which is why they charge on top of commodity model tokens. The piece also dissects pricing narratives (e.g., per-conversation fees that charge for activity rather than outcomes), then recommends buyers push vendors to itemize harness components, identify mandatory prerequisite “taxes,” price LLM passthrough separately, and prefer outcome-based or resolution-based pricing while deciding what to build vs. buy per harness pillar.
Read the full blog for free on Medium.
Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.
Published via Towards AI
Towards AI Academy
We Build Enterprise-Grade AI. We'll Teach You to Master It Too.
15 engineers. 100,000+ students. Towards AI Academy teaches what actually survives production.
Start free — no commitment:
→ 6-Day Agentic AI Engineering Email Guide — one practical lesson per day
→ Agents Architecture Cheatsheet — 3 years of architecture decisions in 6 pages
Our courses:
→ AI Engineering Certification — 90+ lessons from project selection to deployed product. The most comprehensive practical LLM course out there.
→ Agent Engineering Course — Hands on with production agent architectures, memory, routing, and eval frameworks — built from real enterprise engagements.
→ AI for Work — Understand, evaluate, and apply AI for complex work tasks.
Note: Article content contains the views of the contributing authors and not Towards AI.
Related posts
Recent Posts
