Marketing ops · Production

AirOps uses Claude and the Claude Agent SDK to power AI search visibility, content generation, and evaluation at scale

The problem

Most AI models struggle to simultaneously satisfy brand voice, information gain, and search performance requirements, often falling short on quality or performing inconsistently as prompts grow more complex. AirOps' early workflow builder also required human involvement at every step—identifying underperforming pages, deciding actions, triggering workflows, and verifying output—making it impossible to scale to thousands of pieces of content across multiple brands.

First attempt

Previous orchestration frameworks were brittle—setup decisions sometimes required full refactors, and testing different configurations took enormous effort, with time to quality at 60 hours.

Workflow diagram · grounded in source
1
Human triggers content workflow
trigger
“required a human to identify which page was underperforming, decide what to do, trigger the workflow, and verify the output”
2
Claude interprets playbook
ai_action
“Claude interprets the playbook, decides which AirOps tools to call, and executes across sections rather than following a rigid sequence”
3
Sub-agents execute subtasks
ai_action
“Sub-agents from the SDK handle specific subtasks like research, drafting, and evaluation with isolated context windows”
4
Hooks enforce validation checkpoints
validation
“Hooks let them add determinism and validation at specific points in a workflow while keeping autonomous behavior elsewhere”
5
Pre-publication quality scoring
validation
“AirOps scores every piece of content against dozens of checks before it publishes: paragraph structure, source citations, whether the core answer appears in the first 150 words”
6
Human confirms; agent publishes
output
“an agent identifies an underperforming page, drafts updated content grounded in brand voice, and publishes it, with a human only needing to confirm strategy and provide edits when needed”
Reported outcome

AirOps 5x'd its revenue and doubled internal productivity, with engineering moving from prototype to production in weeks after adopting the Claude Agent SDK.
Customer results include Chime achieving a 3x citation increase with an 89% time reduction in content creation, Carta 4x-ing quarterly top-of-funnel output, and LegalZoom cutting Reddit response workflows from 48 hours to under 30 minutes. Internally, the team reached a consensus 2x productivity increase with individual savings of 20 to 40 hours per week.

Reported metrics
AirOps revenue growth5x
AirOps internal productivitydoubled internal productivity
time to quality (SDK development)60 hours to 5 hours
Team productivity increase2x
Show all 13 reported metrics
AirOps revenue growth5x
AirOps internal productivitydoubled internal productivity
time to quality (SDK development)60 hours to 5 hours
team productivity increase2x
individual weekly time savings20 to 40 hours per week
Chime citation increase3x
Chime content creation time reduction89%
Carta quarterly top-of-funnel output increase4x (5 to 20 articles)
Carta citation rate on AirOps-built pages75%
LegalZoom Reddit response workflow time48 hours to under 30 minutes
team member work completion ratetwo weeks of work in one week
weekly hours in Claude Enterprise15 to 20 hours per week
weekly hours in Claude Code40 hours weekly
Reported stack
ClaudeClaude Agent SDKClaude EnterpriseClaude CodeOpusSonnetMCP
Source
https://www.anthropic.com/customers/airops
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

AirOps 5x'd its revenue and doubled internal productivity, with engineering moving from prototype to production in weeks after adopting the Claude Agent SDK.

What tools did this team use?

Claude, Claude Agent SDK, Claude Enterprise, Claude Code, Opus, Sonnet, MCP.

What results were reported?

AirOps revenue growth: 5x; AirOps internal productivity: doubled internal productivity; time to quality (SDK development): 60 hours to 5 hours; Team productivity increase: 2x (source-reported, not independently verified).

What failed first in this deployment?

Previous orchestration frameworks were brittle—setup decisions sometimes required full refactors, and testing different configurations took enormous effort, with time to quality at 60 hours.

How is this marketing ops AI workflow structured?

Human triggers content workflow → Claude interprets playbook → Sub-agents execute subtasks → Hooks enforce validation checkpoints → Pre-publication quality scoring → Human confirms; agent publishes.