LlamaAgents Builder: From Prompt to Deployed AI Agent in Minutes

On this article, discover ways to construct, deploy, and take a look at a no-code doc processing AI agent utilizing LlamaCloud’s LlamaAgents Builder.

Subjects coated embody:

The right way to create a doc classification agent utilizing pure language prompts. The right way to deploy brokers to GitHub-based purposes with out writing any code. The right way to take a look at brokers deployed with LlamaCloud interface invoices and contracts.

Let’s not waste any extra time.

introduction

Creating an AI agent that autonomously performs duties like analyzing and processing paperwork required hours of near-endless configuration, code orchestration, and deployment struggles. Till now.

This text reveals the method of utilizing LlamaAgents Builder to construct, deploy, and use clever brokers from scratch with out writing a single line of code. Even higher, we host it as an app in our 100% owned software program repository.

Time is of the essence as all the course of takes only a few minutes. Let’s get began.

Constructing with LlamaAgents Builder

LlamaAgents Builder is without doubt one of the latest options of the LlamaCloud net platform, whose flagship product was initially launched as LlamaParse. It is a barely complicated mixture of names, however you get the concept. Please word that for now, you may be accessing the Agent Builder via this hyperlink.

The very first thing you may see is the house menu, as proven within the screenshot beneath. When you do not see this, strive clicking the “LlamaParse” icon within the high left nook as a substitute. Then, a minimum of as of this writing, that is what you need to see.

LlamaParse residence menu

Observe that on this instance, we’re working with a newly created free plan account that’s allowed to course of as much as 10,000 pages.

See the “Brokers” block within the backside proper? That is the place the LlamaAgents Builder resides. It is in beta on the time of writing, however you’ll be able to already construct helpful agent-based workflows as we’ll see later.

Clicking this can open a brand new display with a chat interface much like Gemini, ChatGPT, and so on. You may see some advised workflows for what you need the agent to do, however specify your individual by getting into the next immediate within the enter field on the backside: Simply pure language, no code in any respect.

Create an agent to categorise paperwork into “contracts” and “invoices”. For contracts, extract the signing events. For invoices, the full quantity and date.

Specify what the agent ought to do with pure language prompts

Simply submit your immediate and let the magic start. An unimaginable stage of transparency within the inference course of allows you to see the steps you have accomplished and the progress you have made to date.

Creating agent workflows with AgentBuilder

After a couple of minutes, the creation course of shall be accomplished. Not solely will you see an entire workflow diagram that regularly expands all through the method, however you may additionally obtain concise and clear directions on learn how to use your newly created agent. Simply great.

Constructing an agent workflow

The subsequent step is to deploy the agent and make it out there to be used. You might even see a Push & Deploy button within the high proper nook. This begins the method of publishing the agent workflow software program bundle to your GitHub repository, so first ensure you have a registered account on GitHub. For instance, you’ll be able to simply join utilizing your present Google or Microsoft account. Upon getting linked the LlamaCloud platform to your GitHub account, it is very simple to push and deploy brokers. Simply give it a reputation, point out whether or not you need it to be in a non-public repository, and also you’re achieved.

Push and deploy agent workflows to GitHub

This course of takes a couple of minutes and shows a command line-like message on the fly. As soon as it’s full and the agent standing reveals “Operating”, you will note some last messages much like the next:

The “Uvicorn” message signifies that the agent has been deployed and is working as a microservice API throughout the LlamaCloud infrastructure. When you’re acquainted with the FastAPI endpoint, you would possibly need to strive it out programmatically utilizing the API. However on this tutorial, we’ll make issues easier (we promised zero coding, proper?) and take a look at all of it out your self with LlamaCloud’s distinctive consumer interface.

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

Deployed agent is up and working

Now comes probably the most thrilling half. You possibly can check out the agent by going to the playground web page known as ‘Opinions’. First, add a file, akin to a PDF doc containing an bill or contract. If you do not have one, create your individual hypothetical pattern doc utilizing Microsoft Phrase, Google Docs, or an identical device akin to:

LlamaCloud Agent Check UI: Bill Processing

As quickly as a doc is loaded, the agent begins working by itself and inside seconds classifies the doc and extracts the required knowledge fields relying on the doc kind. This outcome could be seen in the best panel of the picture above. The entire quantity and bill date are appropriately extracted by the agent.

Why not add a pattern doc, together with a contract, as we speak?

LlamaCloud Agent Check UI: Processing Contracts

As anticipated, this doc is assessed as a contract, the place the extracted info consists of the names of the signing events.

Effectively achieved! As you proceed to run the pattern, be sure you settle for or reject it primarily based on whether or not the pattern was processed appropriately. This enables the agent to study from suggestions.

Agent take a look at instances and their standing

abstract

We have proven you learn how to construct and deploy an AI agent inside minutes that may classify paperwork and course of them in several methods relying on the doc kind, with out writing a single line of code, inside LlamaCloud’s new function, LlamaAgents Builder.