LlamaAgents Builder: From Immediate to Deployed AI Agent in Minutes

On this article, you’ll learn to construct, deploy, and take a look at a no-code document-processing AI agent with LlamaAgents Builder in LlamaCloud.

Matters we’ll cowl embody:

• Methods to create a document-classification agent utilizing a pure language immediate.
• Methods to deploy the agent to a GitHub-backed utility with out writing code.
• Methods to take a look at the deployed agent on invoices and contracts within the LlamaCloud interface.

Let’s not waste any extra time.

LlamaAgents Builder: From Immediate to Deployed AI Agent in Minutes (click on to enlarge)
Picture by Editor

Introduction

Creating an AI agent for duties like analyzing and processing paperwork autonomously used to require hours of near-endless configuration, code orchestration, and deployment battles. Till now.

This text unveils the method of constructing, deploying, and utilizing an clever agent from scratch with out writing a single line of code, utilizing LlamaAgents Builder. Higher nonetheless, we’ll host it as an app in a software program repository that might be 100% owned by us.

We are going to full the entire course of in a matter of minutes, so time is of the essence: let’s get began.

Constructing with LlamaAgents Builder

LlamaAgents Builder is among the latest options within the LlamaCloud internet platform, whose flagship product was initially launched as LlamaParse. A barely complicated mixture of names, I do know! For now, simply needless to say we’ll entry the brokers builder by way of this link.

The very first thing you need to see is a house menu just like the one proven within the screenshot beneath. If this isn’t what you see, attempt clicking the “LlamaParse” icon within the top-left nook as a substitute, after which you need to see this — no less than on the time of writing.

LlamaParse residence menu

Discover that, on this instance, we’re working underneath a newly created free-plan account, which permits as much as 10,000 pages of processing.

See the “Brokers” block on the bottom-right facet? That’s the place LlamaAgents Builder lives. Despite the fact that it’s in beta on the time of writing, we are able to already construct helpful agent-based workflows, as we’ll see.

As soon as we click on on it, a brand new display will open with a chat interface just like Gemini, ChatGPT, and others. You’ll get a number of advised workflows for what you’d like your agent to do, however we’ll specify our personal by typing the next immediate into the enter field on the backside. Simply pure language, no code in any respect:

Create an agent that classifies paperwork into “Contracts” and “Invoices”. For contracts, extract the signing events; for invoices, the whole quantity and date.

Specifying what the agent ought to do with a pure language immediate

Merely ship the immediate, and the magic will begin. With a outstanding stage of transparency within the reasoning course of, you’ll see the steps accomplished and the progress made thus far:

AgentBuilder creating our agent workflow

After a couple of minutes, the creation course of might be full. Not solely are you able to see the complete workflow diagram, which has steadily grown all through the method, however you additionally obtain a succinct and clear description of how one can use your newly created agent. Merely wonderful.

Agent workflow constructed

The subsequent step is to deploy our agent in order that it may be used. Within the top-right nook, you might even see a “Push & Deploy” button. This initiates the method of publishing your agent workflow’s software program packages right into a GitHub repository, so ensure you have a registered account on GitHub first. You’ll be able to simply register with an current Google or Microsoft account, as an example. After you have the LlamaCloud platform linked to your GitHub account, this can be very straightforward to push and deploy your agent: simply give it a reputation, specify whether or not you need it in a non-public repository, and that’s it:

Pushing and deploying agent workflow into GitHub

The method will take a couple of minutes, and you will note a stream of command-line-like messages showing on the fly. As soon as it’s finalized and your agent standing seems as “Operating“, you will note just a few last messages just like this:

The “Uvicorn” messages point out that our agent has been deployed and is working as a microservice API throughout the LlamaCloud infrastructure. In case you are acquainted with FastAPI endpoints, it’s possible you’ll wish to attempt it programmatically by way of the API, however on this tutorial, we’ll preserve issues less complicated (we promised zero coding, didn’t we?) and take a look at every thing ourselves in LlamaCloud’s personal person interface.

To do that, click on the “Go to” button that seems on the high:

Deployed agent up and working

Now comes probably the most thrilling half. It’s best to have been taken to a playground web page referred to as “Evaluate,” the place you possibly can attempt your agent out. Begin by importing a file, for instance, a PDF doc containing an bill or a contract. Should you don’t have one, simply create a fictitious instance doc of your individual utilizing Microsoft Phrase, Google Docs, or the same software, comparable to this one:

LlamaCloud Agent Testing UI: processing an bill

As quickly because the doc is loaded, the agent begins working by itself, and in a matter of seconds, it can classify your doc and extract the required knowledge fields, relying on the doc sort. You’ll be able to see this end result on the right-hand-side panel within the picture above: the whole quantity and bill date have been appropriately extracted by the agent.

How about importing an instance doc containing a contract now?

LlamaCloud Agent Testing UI: processing a contract

As anticipated, the doc is now labeled as a contract, and on this event, the extracted data consists of the names of the signing events.

Nicely finished! As you retain working examples, ensure you approve or reject them based mostly on whether or not they have been processed appropriately: this helps the agent be taught from suggestions.

Agent testing circumstances and their standing

Wrapping Up

We have now seen how one can construct and deploy, step-by-step and with no strains of code, an AI agent able to classifying paperwork and processing them in numerous methods relying on the doc sort — all in a matter of minutes and inside LlamaCloud’s newly added characteristic, LlamaAgents Builder.