> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.recursiv.io/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.recursiv.io/_mcp/server.

# Recursiv vs Build It Yourself

## How Does Recursiv Compare to Building It Yourself?

Building production agents from scratch means stitching together a framework, model providers, a vector store, a job queue, a sandbox runtime, auth, a deploy host and billing. Recursiv gives you all of that behind one SDK. You ship the agent and skip the integration project.

The differentiator is time and maintenance saved. Each piece you self-assemble is a service to wire, secure, scale and keep running. Recursiv runs the agents for you with orchestration, scheduling, sandboxes, deploys and observability built in, so you can prove what ran.

## Quick Comparison

| Capability         | Recursiv                                     | Build It Yourself                |
| ------------------ | -------------------------------------------- | -------------------------------- |
| Agent framework    | Built in (agents, swarms, dispatcher)        | Pick and wire a library          |
| Model support      | Any model, swap anytime                      | Integrate each provider yourself |
| Memory             | Brain and memory primitives                  | Stand up a vector DB             |
| Task orchestration | Dispatcher and jobs                          | Build a queue and scheduler      |
| Code execution     | Managed sandboxes                            | Run and isolate your own         |
| Auth               | Built in                                     | Add an auth provider             |
| Data and storage   | Postgres and object storage                  | Provision and connect each       |
| Deploy             | Deployments built in                         | Set up a host and pipeline       |
| Integrations       | Hundreds built in                            | Build each connector             |
| Govern             | Scoped identity, approvals, audit, self-eval | Design and build it all          |
| Billing and usage  | Built in                                     | Wire your own metering           |

## The Realistic Shopping List

| Layer               | DIY                                        | Recursiv                  |
| ------------------- | ------------------------------------------ | ------------------------- |
| Agent framework     | Choose, wire, maintain                     | `client.agents.create()`  |
| Model access        | Per-provider SDKs and keys                 | Model agnostic, one call  |
| Vector memory       | Provision and operate a vector DB          | Brain and memory built in |
| Queue and scheduler | Build and run                              | Dispatcher and jobs       |
| Sandbox runtime     | Isolate and scale                          | Managed sandboxes         |
| Auth                | Add a provider                             | Built in                  |
| Deploy host         | Configure and pipeline                     | Deployments built in      |
| Billing             | Meter and invoice                          | Usage tracking built in   |
| **Result**          | **Months of integration plus ongoing ops** | **Hours**                 |

## When Building It Yourself Is the Better Choice

* You have a hard requirement to own and operate every layer in your own cloud
* Your needs are narrow enough that a single library covers them and you will not grow into orchestration, memory or govern
* You have a platform team whose job is to maintain this infrastructure
* You need a bespoke runtime that no platform can match

## Get Started

If you would rather build the agent than the infrastructure, start with the [quickstart](/guides/start-here/quickstart) and have a governed agent running in minutes.