What you'll build

You'll build:

• A backend API with a database. A REST API endpoint that accepts a POST request and saves the data as a document in the built-in BuildShip Database.

• An AI workflow on a schedule. A nightly "Business Health Monitor" that fetches metrics, uses an LLM to spot anomalies, saves a report and emails a summary.

• An AI agent with tools. An agent with instructions, a model of your choice, and tools made from library nodes, your own flows and MCP servers.

Everything starts from your workspace home, where Start building offers New Flow, New AI Agent and Start with a guided example.

The BuildShip home screen: start a new flow, a new AI agent, or a guided example.

Part 1: Build a backend API with a database

The first build is a classic backend task: an endpoint that receives data from a contact form and saves it to a database.

Step 1: Create a new workflow

Click New Flow. BuildShip asks "What do you want to build today?" and offers three starting points: Workflow (build your API or automation flow on a blank canvas), AI Agent and Tool for AI Agent. Below them is a template gallery with Search templates....

Choose Workflow for an API or automation, AI Agent for an agent, or Tool for AI Agent to build a tool an agent can call.

Choose Workflow, give the flow a name in the New Flow dialog and click Create.

Name the workflow and click Create.

Step 2: Add a REST API Call trigger

Every BuildShip workflow starts with a trigger. Open the Connect tab and click Add trigger to see the trigger library. Under Rest API you'll find REST API Call, which creates an API / HTTP endpoint as the starting point for your workflow. The library also has integration triggers such as Supabase Trigger and Buildship Database Trigger, and scheduled triggers: Custom Schedule (Cron) and Interval Schedule (Cron).

The trigger library: REST API endpoints, integration and database triggers, and scheduled triggers.

Select REST API Call. In Setup, BuildShip pre-fills a Path; change it to something readable, such as /contact. Leave Method as POST for this example (BuildShip also supports GET, PUT and DELETE). Advanced Options sets the request content type (application/json by default). The payload size limit is 32MB per request.

Click Connect to connect the trigger to the workflow. This generates the endpoint URL for your workflow.

Set the Path and Method, then click Connect to create the endpoint.

Step 3: Define the inputs

Switch to Build to see the canvas: the REST API Call trigger, an Inputs block, an Add node button, and a Flow Output block that returns the Last Node Output by default. Test and Ship sit in the top bar, and Logs along the bottom.

A blank flow: trigger, inputs, nodes and output, with Test, Ship and Logs always in reach.

Click Add Input and create the fields your API expects, for example name, email and message. Every node in the flow can use them.

Step 4: Add a database node

Click Add node to open the node library. Search for a node, or use Generate to have AI create one. Most popular includes OpenAI nodes and BuildShip Database — Create Document; Flow nodes cover Branch, Loop, Parallel, Switch and Execute AI Agent; Integrations lists third-party services.

The node library: AI models, database operations, flow control and integrations in one place.

Pick BuildShip Database — Create Document. Every BuildShip project comes with a ready-to-use Firestore database, so there's nothing to provision. Configure the node:

• Collection Name: the collection to write to, for example contact_messages.

• Data: open the Editor and map your inputs into the document, such as name, email, message and a timestamp.

• Additional Settings: optionally set a Document ID (otherwise one is generated) and whether to merge with an existing document.

The node's output, Firestore Response, flows into Flow Output, so your API returns the result of the write.

The Create Document node writes request data straight into the built-in BuildShip Database.

Prefer another database? The same pattern works with nodes for Firestore, Postgres, Supabase, MySQL and MongoDB. You can view and edit BuildShip Database data in BuildShip Tables, a spreadsheet-style UI.

Step 5: Test the flow

Click Test to open the test panel. Enter values in Form view or paste a request body in JSON, then click Test Flow. BuildShip runs the workflow and marks each node with a success or failure state so you can inspect its output. Save inputs as Examples and revisit Recent Executions.

Test any flow with form or JSON inputs before you ship it.

Step 6: Ship it and call your API

When the test looks right, click Ship in the top-right corner. Deployment takes a few seconds, and BuildShip shows your API endpoint URL in a pop-up. Changes go live only when you click Ship again, and each ship creates a new version (version history is on paid plans).

Call it from any frontend, app builder or script:

curl -X POST "https://your-project.buildship.run/contact" \
-H "Content-Type: application/json" \
-d '{"name": "Ada", "email": "ada@example.com", "message": "Hello!"}'

Before going public, follow BuildShip's guide to securing your API with an Authorization header check.

Step 7: Monitor with logs

Open Logs to see every execution, filtered by time range, by Success or Failed, and by Prod or Test. BuildShip logs node inputs, outputs and trigger request data; production logs appear within about 45 seconds, or instantly with Refresh.

Execution logs filtered by time range, success or failure, and Prod or Test environment.

That's a real backend API: an HTTP endpoint, database writes, test tooling, versioned deploys and production logs.

Part 2: Build an AI workflow that runs on a schedule

Plenty of backend work happens on a timer: reports, syncs and monitoring. Here's a complete AI workflow, the Nightly Business Health Monitor.

A scheduled AI workflow: fetch data, calculate, analyse with an LLM, save the report and email a summary.

Node by node:

1. Custom Schedule (Cron): the trigger. Set the Schedule with cron syntax (for example 0 0 * * * runs every day at midnight), choose the Time Zone, and the Location is filled in from your workspace's Google Cloud region. Click Connect.

2. Fetch Daily Metrics: reads the day's numbers from your data source.

3. Calculate Metrics & 7-Day Averages: computes today's figures and compares them with a rolling weekly average.

4. AI Health Status & Anomaly Analysis: an LLM node that reads the metrics and returns a status and a plain-English explanation of anything unusual.

5. Save Health Report: writes the report to the database so you have a history to query later.

6. Email Health Summary: sends the summary to your inbox through a Gmail node.

7. Flow Outputs: returns the final result, which also lands in the logs.

To build your own version, start a new Workflow, choose Custom Schedule (Cron) (or Interval Schedule (Cron) for a simple "every N minutes" job) and add nodes with Add node. Test it with Test, then click Ship to make it active. It then runs on schedule with no server to manage, and every run shows up in Logs.

This is the gap between a chat assistant and a workflow builder. A chat assistant can explain your metrics if you paste them in. A shipped workflow fetches them, analyses them, stores the result and tells you, every night, unprompted.

Part 3: Build an AI agent with real tools

An AI agent is an LLM that decides which tools to use to reach a goal. Create one with New AI Agent. The example is a support agent that triages and updates tickets.

Step 1: Write the instructions

In Instructions, describe your agent's purpose, tone and tool usage. Be specific about when to use each tool. For example: "When a ticket is urgent, use the Slack Message tool to alert the on-call team."

Step 2: Choose the model

Under LLM Configuration, pick an AI Provider and Model. The example uses Anthropic with Sonnet 4.5. For Key, you can Use credits (BuildShip's keyless option, billed from your credits) or bring your own API key. Set Max Tokens to cap the response length.

You can also give the agent knowledge and context. Websites and Files add reference sources, and Context Variables define values the agent needs when it's called from a flow; they appear as inputs on the Execute AI Agent node. Try it in the chat panel (Ask the Agent...).

The agent editor: instructions, model, knowledge sources and context variables, with a live chat panel to test.

Step 3: Give the agent tools

Tools are what turn a chatbot into an agent. Click Add Tool and choose from:

• Node from Library: any node from the BuildShip library, such as sending an email or posting a Slack message.

• Custom Tool (Flow): one of your own BuildShip workflows. This is the powerful one: everything you built in Parts 1 and 2 (database writes, API calls, multi-step logic) can become a single tool the agent calls.

• Agent Handoff: pass the conversation to another agent that specialises in a different job.

• MCP server: connect tools exposed by an MCP (Model Context Protocol) server.

The In use section lists the agent's current tools. In this example they are Send Email and Slack Message (both library nodes) and Update Ticket Status (a custom flow).

Agent tools can be library nodes, your own flows, handoffs to other agents, or MCP servers.

Because a tool can be a full workflow, your agent works with your real data, wrapped in the validation you build into that flow. It works the other way too: BuildShip tools can be deployed as MCP-ready endpoints for clients such as Claude, Cursor and ElevenLabs.

Step 4: Use the agent anywhere

Add the Execute AI Agent node to run the agent inside any workflow, so you can call it from a REST API (for an in-app chat) or a scheduled flow (for a nightly run).

Build faster with Compass, your AI builder

You don't have to wire every node by hand. Click Build with AI to open Compass - Your AI Builder (currently V2 (Beta)). Tell Compass what you want in plain language and it helps with:

• Building an agent from scratch or improving an existing one

• Creating tools for your agent (nodes, flows, MCP servers and more)

• Debugging failed executions and fixing errors

• Refining your agent based on feedback

Compass sits beside your agent or flow and can build, extend and debug it from a chat prompt.

Unlike a chatbot, Compass works on your actual canvas: it builds a flow you can test and ship, not a snippet you still have to host. Every node's code stays editable.

Why this beats a chat assistant alone

ChatGPT, Grok and other chat assistants are excellent thinking partners, but on their own they stop at the conversation. BuildShip gives you:

• A deployed API. Your app calls a real HTTPS endpoint, not a chat window.

• A database that persists. Data is written to the built-in BuildShip Database or your own Postgres, Supabase, Firestore, MySQL or MongoDB, and is still there tomorrow.

• Jobs that run without you. Cron and interval triggers run flows on a schedule, unprompted.

• Agents with real tools. Agents act through your flows, library integrations, other agents and MCP servers, not just text.

• Operational visibility. Test runs, production logs and version history.

• Your infrastructure, your keys. Workflows run on Google Cloud Run in the region you choose, and secrets are stored in Google Cloud Secret Manager in your own single-tenant project.

Think with a chat assistant; build and run with BuildShip.

How BuildShip compares with n8n, Zapier, Make, Xano, Supabase and Flowise

Each tool is strong at what it was built for. This table summarises how each describes itself on its official site, and where BuildShip fits.

If you only need to connect SaaS apps, an automation tool may be enough. If you need an AI agent backed by a real API, a database and scheduled jobs, that's what BuildShip was built for.

Frequently asked questions

Is BuildShip a real backend?

Yes. BuildShip workflows ship as HTTPS API endpoints (GET, POST, PUT and DELETE), run on Google Cloud Run, read and write databases (the built-in Firestore-backed BuildShip Database, or Postgres, Supabase, MySQL, MongoDB and Firestore), run on cron schedules and respond to database events. Every execution is logged.

What is the best AI agent builder?

It depends on the job. For connecting many SaaS apps, Zapier or Make are strong choices; for self-hosted automation, n8n is popular. If your agent needs its own backend (a database, API endpoints, scheduled jobs and tools built from your own workflows), BuildShip is designed for exactly that.

What is the best AI workflow builder?

The best AI workflow builder lets you combine LLM steps with real logic, data and deployment. BuildShip offers LLM nodes, branching, loops and parallel steps, plus REST API, scheduled and event triggers and one-click deployment.

What is a no-code backend builder?

A no-code backend builder lets you create server-side logic, such as APIs, database operations, scheduled jobs and integrations, visually instead of writing and hosting server code. BuildShip is low-code: you build visually, and every node's JavaScript or TypeScript is there when you need it.

Can I self-host BuildShip or own my data?

You choose your project's region, and your workflows are deployed there, so your data stays in that region. API keys and secrets are stored in Google Cloud Secret Manager in your own single-tenant project. For full control, you can export workflows and tools as JavaScript and deploy them on your own infrastructure, such as Vercel, GCP, AWS or Azure. Self-hosting and bring-your-own-cloud are offered on the Enterprise plan.

Can BuildShip AI agents use MCP?

Yes. You can add an MCP server as a tool on any BuildShip agent. You can also build tools in BuildShip and deploy them as MCP-ready endpoints for clients such as Claude Desktop, Cursor and ElevenLabs.

Do I need my own OpenAI or Anthropic API key?

No. BuildShip's keyless nodes let you run AI models using your BuildShip credits, shown as Use credits in the agent's Key setting. You can switch to your own API key at any time.

How much does BuildShip cost?

Paid plans start at $19 per month for the Starter plan, which includes 20,000 credits per month, and you can start with a 7-day free trial. Pro is $59 per month, Business is $449 per month, and Enterprise is custom. See buildship.com/pricing for current details.

Start building

You've seen the three building blocks: an API that writes to a database, an AI workflow that runs nightly, and an agent with real tools, all on one canvas. Pick one and build it today.

Start your free trial at buildship.com, open New Flow or New AI Agent, or ask Compass to build the first draft for you. Browse ready-made templates at templates.buildship.com and the full guides at docs.buildship.com.