Relay.app

Relay.app enters the AI workflow automation space with a clear thesis: automation should not mean ceding control to machines. While many tools in this category focus on fully autonomous execution, Relay.app deliberately builds human oversight into the core of its platform. The result is a system that feels less like a black-box bot and more like a collaborative teammate one that can draft, summarize, and extract data, but always knows when to pause and ask for approval or input. This makes it particularly well suited for teams that need the speed of AI but cannot afford to lose the guardrails of human judgment. At its heart, Relay.app is a platform for creating what it calls playbooks multi-step workflows that can span over one hundred apps, including Gmail, Slack, Notion, and many others. Users can chain together actions, add conditional logic, and inject AI steps powered by models like GPT, Claude, and Gemini. The AI can perform tasks such as data extraction, text summarization, and even transcribing audio. But the standout feature is the human-in-the-loop step, where a workflow can pause and wait for a person to approve, edit, or provide additional information before proceeding. This is not merely a nice-to-have; it is a fundamental design choice that sets Relay.app apart from simpler automation tools. For instance, a customer support team can build a playbook that automatically routes a ticket, generates a draft response using AI, and then sends it to a human agent for review before it goes out. The AI handles the grunt work, but the human retains the final say. This pattern is especially valuable in contexts where errors carry high stakes, such as legal communications, financial reporting, or customer-facing messaging. Another strength is the platform's flexibility in handling complex logic. Relay.app supports paths, webhooks, and custom AI prompts, allowing power users to create nuanced automations that adapt to different scenarios. A meeting briefing generator, for example, could pull notes from a Notion database, check the calendar for upcoming events, and then compose a summary with key action items all without a single line of code. Similarly, a marketing team could build a LinkedIn engagement tracker that monitors interactions on posts and compiles a weekly report in a spreadsheet. These use cases demonstrate the tool's ability to go beyond simple trigger-action pairs and into the realm of orchestrated, multi-step processes. However, the platform's sophistication comes with a pricing model that demands careful consideration. Relay.app charges based on two separate metrics: steps and AI credits. Steps are counted for every successfully completed action, including sending a Slack message, updating a HubSpot contact, or running a utility step. AI credits are consumed when an AI model is invoked. The free plan offers 200 steps and 500 AI credits per month, which is generous for testing but will be consumed quickly in a production environment. The Professional plan starts at $19 per month for 750 steps and 5,000 credits, and the Team plan at $69 per month for 2,000 steps and 5,000 credits. Additional steps and credits can be purchased, but the dual-metered billing can make cost estimation tricky. Heavy users may find themselves surprised by the combined consumption, especially if they run frequent AI-powered tasks. That said, Relay.app does allow users to bring their own AI model credentials, bypassing the credit system entirely. This is a significant advantage for organizations that already have paid accounts with OpenAI or Anthropic, as they can route AI costs through their existing subscriptions. For such users, the step-based pricing becomes the primary cost driver, which may be more predictable. The platform's audience is clearly teams that need collaboration features. The Team plan includes unlimited teammates, and the entire workspace can share playbooks, credentials, and human-in-the-loop tasks. This multiplayer experience is a differentiator compared to many no-code automation tools that are essentially single-user. Founders, marketing teams, customer success teams, and HR departments are all likely beneficiaries. A founder might build a custom Slack bot that answers employee questions using company data, while an HR team could automate onboarding tasks with approval gates for manager sign-off. The common thread is that these workflows require both AI efficiency and human judgment. On the cautionary side, Relay.app may be overkill for users who only need simple automations like sending a Slack notification when a form is submitted. Simpler, cheaper tools exist for those scenarios. Additionally, the platform's reliance on polling for triggers (checking for new data) does not count as steps, which is a welcome detail, but the overall complexity of the interface and the learning curve for building multi-step playbooks should not be underestimated. Non-technical users will appreciate the no-code builder, but mastering paths, webhooks, and custom prompts takes time. In summary, Relay.app is a powerful, team-oriented automation platform that excels when human oversight is non-negotiable. Its human-in-the-loop capability, combined with broad app integrations and flexible AI, makes it a strong candidate for organizations that want to automate without abdicating responsibility. The pricing requires careful monitoring, but the ability to bring your own AI credentials mitigates some of that concern. For teams that need to balance speed with control, Relay.app offers a compelling middle ground.