Getting Started with the GenAI Platform

The GenAI Platform, introduced astatine Deploy 2025, is DigitalOcean’s consequence to the quickly increasing request for AI based solutions to quality problems pinch Large Language Models. In practice, this work allows users to create various AI powered agents which tin beryllium utilized to powerfulness a myriad of different applications.

In this tutorial, we will look successful extent astatine the capabilities of the GenAI Platform, research its unsocial features that make it truthful useful, for everyone from hobbyists to business leaders to developers and show the limitless possibilities provided by specified a product. Afterwards, we will talk getting started pinch utilizing the level for ourselves.

Capabilities of the GenAI Platform Overview

The GenAI level is acold much than conscionable a serving level for Large Language Models, and while it tin beryllium utilized to create civilization chatbots, it is acold from constricted to specified tasks. AI agents are besides awesome for doing analyzable information analyses, fraud detection, cybersecurity, healthcare and more. All it requires is the observant exertion of information to the knowledge base.

Potential of utilizing GenAI

The imaginable for the GenAI level is genuinely endless. We envision systems wherever users attack each imaginable types of problems pinch an AI solution. Some examples of what Agentic AI from the GenAI level could beryllium utilized for include:

• Healthcare: specified arsenic supporting diagnostic systems pinch AI
• Finance: making informed decisions pinch nonstop information from your individual finances
• IT: automate overmuch of the difficult activity for IT professionals, specified arsenic password resetting
• Marketing: analyse contented for SEO, brainstorm engagement ideas, and moreover make content
• Cybersecurity: Agentic AI tin beryllium utilized to place threats

Features of the GenAI Platform

The main features of the GenAI level are based astir the expertise to return successful caller accusation into an existing LLM agent, and past usage that accusation to behaviour immoderate action. These actions tin beryllium applied to immoderate functionality we plug into the model, and tin beryllium artificially constricted by defender obstruction exertion to forestall misuse aliases hallucination.

In this section, we will research the GenAI Platform’s features successful greater detail, and explicate why each characteristic helps make GenAI truthful powerful for everyone - from business users to developers to hobbyists.

Function calling

Arguably the astir captious exertion of LLMs arsenic agentic ai, usability calling is the expertise to usage accusation from the LLM to make a consequence that tin beryllium utilized to trigger a codification usability elsewhere. This capacity is autochthonal to galore AI models, but often lacks the expertise to grip caller accusation owed to the deficiency of vulnerability to caller information successful the training set.

The GenAI platform’s remedy to this problem is the summation of Retrieval Augmented Generation (RAG) from group knowledge bases, which let the exemplary to accurately usage caller accusation to create meticulous and utile usability calls.

Retrieval Augmented Generation pinch Knowledge Bases

Retrieval Augmented Generation, aliases RAG, enhances the capabilities of a ample connection exemplary (LLM) by allowing it to entree and incorporated applicable accusation from outer knowledge bases, for illustration databases aliases documents, earlier generating a response, ensuring much meticulous and contextually applicable outputs compared to a modular LLM alone. In practice, it allows the exemplary to look up accusation astir the information provided. For example, RAG could let an LLM to look astatine receipt information to cognize really overmuch effective costs were.

The Knowledge Base is different cardinal constituent of RAG. RAG requires these knowledge bases, fundamentally a corpus of rich | matter data, to tie accusation from. Knowledge Bases connected the GenAI Platform are optimized to champion service our users aft upload.

Guard rails

LLM Guardrails are a group of rules and practices that guarantee AI systems are safe, ethical, and reliable. They thief forestall harmful, biased, aliases incorrect outputs, which are communal hallucinations we tin find successful AI Agents. The guardrails connected the GenAI level are group up to guarantee that our models ne'er veer excessively acold disconnected topic, are incapable to springiness malicious responses, and forestall insecurity from the model.

Agent routing

GenAI functions tin usability arsenic a strategy that uses artificial intelligence to automatically nonstop personification queries aliases requests to the astir due AI supplier wrong a network, a process called agentic routing. With supplier routing, it’s imaginable to link aggregate agentic AI systems together, and person the champion result from the optimized agent. This typically useful done the exertion of a superior supplier which oversees which routes to nonstop the consequence to the output. This is incredibly powerful because it allows for the nonstop relationship betwixt aggregate agents, removing the limitations from utilizing a azygous supplier for a solution.

Getting started pinch GenAI

Getting started pinch the GenAI level is easy! First, we request to do 2 things: login to our DigitalOcean accounts, and place information for an agent. The knowledge guidelines tin return accusation successful a assortment of forms, including archive formats, CSVs, JSON and more, truthful erstwhile we person identified our corpus, we are fresh to really get going.

Next, navigate to the GenAI tab connected the DigitalOcean website. We tin do this by clicking the negociate fastener connected the navigation barroom connected the near of the screen, and selecting the 2nd option.

Creating the Knowledge Base

From here, click connected the Knowledge Bases tab. We are going to create a caller knowledge guidelines by uploading our information onto a DigitalOcean Space. These are based connected OpenSearch database exertion to facilitate RAG. Fortunately if we do not already person one, we tin create 1 correct successful the Knowledge Base creation page.

First, sanction your Knowledge Base, past scroll down to the database section. Select the “Create New”. Select an due region from which to big the agent, we chose Toronto. Next, scroll down to prime the embeddings model. Embedding models person matter (like personification queries and knowledge guidelines documents) into numerical vectors, truthful we should prime the exemplary champion suited to our needs. Large Knowledge Base corpi will require the much costly model, but a elemental demo conscionable needs thing smaller for illustration MiniLM. Finally, prime the task you want the KB to beryllium in, and click Create. This will past return immoderate clip to embed your information utilizing the embedding exemplary to group up the knowledge base, but, erstwhile complete, we are fresh to move connected to the adjacent step.

Creating the GenAI Agent

Now it’s clip to create the existent supplier itself. Navigate backmost to the GenAI level homepage utilizing the barroom connected the near link. Then click “Create Agent” connected the apical correct of the screen. This will return you to the Agent creation page.

First, participate a sanction for your agent. Next, adhd successful applicable instructions for the supplier to follow. For example, pursuing our receipt illustration from earlier, we could instruct the supplier to wholly respond pinch numbers aliases to behave arsenic a adjuvant financial advisor.

We tin past prime our model. Here is wherever things will alteration greatly successful really our Agent will behave. The type of exemplary affects everything from really caller the information it has observed to really it will constitute retired its responses. We urge doing investigation connected each of the disposable models, spot here, but person a mates recommendations nary the less:

• DeepSeek-R1-Distill-Llama-70B - a distilled exemplary derived from LLaMA 3.2 70b supervised fine-tuned connected 80000 examples from the original DeepSeek R1. This exemplary has awesome capabilities wherever longer responses are needed, and the exemplary tin besides execute analyzable reasoning for coding and mathematics problems.
• Anthropic 3.5 Sonnet - 1 of the astir precocious closed root models connected the web, Sonnet is an exceptional LLM for some usability calling and axenic information. This exemplary requires an Anthropic subscription.

Look retired for models to beryllium made disposable successful the adjacent future!

Once exemplary action has been completed, we tin adhd successful our knowledge guidelines we created earlier. This will let the GenAI Platform to execute RAG connected the knowledge guidelines utilizing the recently created LLM Agent.

Finally, we tin delegate the supplier to a circumstantial project, and create it! This creation whitethorn return a fewer minutes. Once everything is ready, we will beryllium served pinch a specifications page wherever we tin interact pinch the caller agent.

Interacting pinch the agent

There are 2 main ways we tin interact pinch our agent: via API aliases the chat playground. The afloat functionality of the api tin beryllium recovered here. For this example, let’s return a look astatine the chat playground to show really the exemplary works.

This illustration is from the DigitalOcean Tutorial Expert disposable connected the website for each to trial out. As we tin see, the supplier is capable to behave for illustration a regular chatbot if we configure it this way. But it is acold from constricted to specified actions, arsenic we outlined previously.

Closing Thoughts

All successful all, the GenAI level is simply a robust instrumentality for creating and interacting pinch Large Language Model based agents. Thanks to the exertion of RAG pinch Knowledge Bases hosted connected the aforesaid system, we showed that the Agents are incredibly easy to usage and group up for anyone, from business leaders to ml engineers to hobbyists. We dream you each person a chance to research the GenAI level much going forward.