← Sticker Wall AGENT · DETAIL

Relay.app Workflow Agent Setup Service: Helping Teams Save $50,000/Month with 40 AI Agents

Workflow: Describe business processes in natural language daily to generate multi-step AI agents that execute tasks such as data s

AGENT

Key Fields

FIELD STAMPS

IndustryContent / Creator Economy

RegionGlobal

ScaleSME

ChannelOnline

🔧 Workflow

Describe business processes in natural language daily to generate multi-step AI agents that execute tasks such as data scraping, content generation, and email triggering; humans review outputs at critical nodes, and agent results automatically flow back into the CRM or spreadsheets. The specific rhythm involves checking the agent queue for pending review tasks in the morning, approving or rejecting them to let agents continue running; adjusting agent instructions in the afternoon based on client feedback, breaking down new business requirements into executable node flows; and reviewing daily agent execution logs at night, flagging error nodes and iterating on prompt words.

🛠 Setup Requirements

Requires familiarity with node orchestration on Relay.app or similar AI workflow platforms, along with mastery of MCP tool integration and API authorization; the technical barrier is medium, requiring only the ability to describe processes in natural language, with a learning curve of about 2 weeks. Setting up a complete multi-step agent requires first sorting out the client's existing business processes, inserting review steps at every node requiring human judgment, and then configuring data sources and output destinations. Following the shutdown of Relay.app, existing agents need to be migrated to n8n or Make, ensuring the agent logic remains unchanged during migration while only replacing the underlying platform's node implementation.

🧰 Toolchain

🔧 Relay.app
🔧 n8n
🔧 Make
🔧 OpenAI API
🔧 Google Sheets
🔧 Slack

💰 Revenue

① SME teams (Primary revenue): Charging service fees via monthly client subscriptions, approx. $500/month per client × serving 3-5 clients simultaneously = $1,500-$2,500/month, total reported figures around $2,000-$5,000/month, annualized at $18,000-$30,000. Exact proportion of total revenue is not specified (estimated figures; merchant-stated figures independently unverified). ② Deployment and migration project service fees: Clients pay a one-time setup and migration fee; the exact amount, number of projects completed, and revenue share are undisclosed and unverified. ③ Mature stage agent maintenance subscription: Daily maintenance per client compressed to under 30 minutes, renewed at a fixed monthly fee (exact fee not specified, renewed client count unverified, exact share in total revenue unquantifiable). ④ Opportunity item - Undertaking migration and hosting for legacy users of shut-down platforms: A former Google product lead used 40 AI agents at $500/month to complete the work of a $50,000 team (media-estimated figure, independently unverified); the exact revenue share from this remains unquantified.

💸 Cost

Tool subscription fees are approx. $50-$100/month, and API call costs range from approx. $30-$80/month depending on task volume; after migrating to n8n or Make, basic subscriptions remain at a comparable level. If clients request private deployment, server costs will rise to approx. $150/month, though this portion is typically passed on to the client.

⏱ Time Investment

2-3 hours daily for debugging and reviewing agent outputs, approx. 15 hours per week; client delivery and communication are interspersed, with peak weeks reaching up to 20 hours. Once entering steady operations, daily maintenance time per client can be compressed to under 30 minutes, with the remaining time used for acquiring new clients and iterating on templates.

🚀 Getting Started

Start by using the free version of Relay.app or n8n to build a marketing automation workflow for yourself. Once running successfully, record the process as a case study and post it on LinkedIn or indie developer communities to land your first $500/month setup order. The beginner's first step is to choose a familiar business scenario, such as email marketing or data scraping, describe the entire process as a node graph using natural language, and verify that the AI agents execute stably before showcasing it externally.

🔑 Keys to Success

✅ Human review nodes ensure output quality and prevent AI hallucinations from polluting client data
✅ The cost-reduction narrative of replacing a 5-person team with 40 agents appeals to clients, directly corresponding to a 580,000 RMB monthly salary cost
✅ Subscription model creates compound revenue, with the same agent template replicable across multiple teams
✅ Natural language agent generation lowers the barrier to entry for clients, with implementation cycles as short as 2 weeks
✅ Integrating clients' CRM and email systems creates high migration costs, thereby improving renewal rates

⚠️ 风险

⚠️ Relay.app has shut down, requiring migration to alternative platforms like n8n or Make, necessitating a tech stack update and posing a risk of interruption to existing client workflows
⚠️ AI agent execution errors could directly reach client inboxes or CRMs, potentially causing brand incidents and liability claims
⚠️ When clients tighten internal IT policies, changes to API authorizations and data permissions can cause agents to stall
⚠️ Subscription revenue relies on a handful of high-value clients; the churn of any single client will cause significant monthly revenue fluctuations

📌 Real Cases

📌 A former Google product lead used 40 AI agents for marketing at a monthly cost of $500, replacing a $50,000 team; related reports featured on 36Kr and NetEase Subscriptions
📌 An in-depth Relay.app Agents review article showcased a 24/7 AI team case study, covering end-to-end automation of data scraping, content generation, and email outreach
📌 A Relay.app vs. Zapier comparative analysis showed that Relay.app's AI orchestration capabilities are better suited for multi-step agent scenarios, prompting multiple teams to switch and generate setup demand
📌 The Relay.app product page featured on SimilarLabs emphasizes human-in-the-loop review nodes, with multiple enterprise clients sharing implementation experiences of automated workflows with human reviews in its community

Sources