Recommendation: use a hybrid model, but do not price raw “requests” the way LLM vendors do. For an orchestrator like workingagents.ai, the cleanest structure is:

Recommendation: use a hybrid model, but do not price raw “requests” the way LLM vendors do. For an orchestrator like workingagents.ai, the cleanest structure is:

1. Platform fee
2. Usage fee based on weighted executions/credits
3. Optional implementation/consulting at your hourly rate

$300/hour for integration and custom interface work is reasonable if you are doing senior-level API integration, workflow design, security, and custom MCP/tool work. Keep that separate from product pricing so buyers understand they are paying for delivery, not just runtime.

For the product itself, I’d avoid “per request” pricing because one request can vary wildly in cost and value. Market comps are closer to workflow runs / tasks / credits:

• n8n prices by workflow executions, not users or steps: https://n8n.io/pricing
• Zapier prices by tasks per month: https://zapier.com/pricing
• Make prices by credits/operations: https://www.make.com/en/pricing
• Retool mixes seat/platform pricing with workflow runs and agent hours: https://retool.com/pricing

My recommended pricing model:

• Starter: $99/month Includes 5,000 credits, basic MCP tools, email support, low concurrency
• Growth: $299/month Includes 25,000 credits, team features, audit history, higher concurrency
• Business: $999/month Includes 100,000 credits, SSO, SLA, governance, private deployments/options
• Overage: Charge per 1,000 credits, for example $10 to $25 / 1,000 depending on margin

Define credits by cost intensity, not by HTTP request count:

• Simple tool call / webhook / CRUD action: 1 credit
• Multi-step orchestration run: 5 to 20 credits
• LLM call passthrough: cost-plus markup or separate metered line item
• Long-running jobs, scraping, embeddings, or agent loops: higher weighted credit cost

For LLM usage, I would either:

• pass through model cost plus a markup, or
• bundle a monthly allowance and meter overage separately

That is better than pretending all orchestrator requests are equivalent.

Consulting recommendation:

• Keep $300/hour as your list rate
• Also offer fixed-scope packages:
  • Integration setup: $3k to $7.5k
  • Internal tool / MCP rollout: $10k to $25k
  • Custom orchestrator deployment: $25k+

Why this works:

• Subscription creates predictable revenue
• Credits scale with customer value
• Consulting monetizes implementation without distorting product pricing
• Enterprise buyers usually accept platform + usage + services much more easily than “per API request” pricing