A few months ago, we launched a groundbreaking innovation in virtual assistants, the Agent Methodology. This advanced methodology introduced a transformative approach that empowers businesses to create dynamic teams of specialized agents equipped with Generative AI meticulously aligned with their unique organizational structure, thanks to our platform's seamless integration with proprietary business data.

These specialized agents are not just software programs. They are sophisticated computational entities designed to simulate human behavior with remarkable precision. Their interactive abilities are comparable to those of a well-trained company representative, enabling them to handle various situations efficiently and adjust to changes in real-time.

At the core of our process is customizing these agents with detailed company-specific information, mirroring the meticulous onboarding process of a new employee. This method ensures that each Generative Agent is not only informed by the business's unique data and insights but also aligned with its culture and operational needs. Through this, we empower businesses to achieve unparalleled efficiency and innovation in their operations.

In this article, we will see how to build Generative Agents easily.

How to build Generative Agents with a Conversational AI platform

Building Generative Agents with a Conversational AI platform can be easy and fun.

Identifying business needs

To begin with, consider the different areas of your business where you require a specialized agent. For instance, you may need one for customer service, technical support, or to boost online sales.

Creating a new answer

Next, create a new answer for the agent you've chosen. Answers are standard components of AI platforms for creating virtual assistants and contain the information to respond to user requests.

Adding a generative block

Platforms provide an easy way to create a new answer by using blocks. These blocks, like bricks, act as the essential building elements for the answer. Although different types of blocks are available, it is enough to use a generative one to create an agent.

Setting the prompt structure for the AI agent

Now, you have to set the structure of the agent prompt. The prompt consists of a series of instructions that the Generative AI must follow to respond.
To build an efficient Generative Agent, it needs access to a range of information, including:

its context
general instructions
a description of the company
the appropriate tone of voice
a comprehensive product list
guidelines to follow when answering queries
relevant documents to base its responses on
samples of previous conversations
new conversations with new users

Let’s now look at each of these steps in detail.

1. Setting the context

Use the very first lines of your prompt to set the context for your agent, explaining who they are and what the range of action is.

Example: “You are the Al chatbot, an extreme expert on COMPANY NAME products. You have in-depth knowledge of COMPANY NAME products and know how to illustrate each one's various features. You know everything about the company and give users punctual information about COMPANY NAME business.”

2. General instructions

Explain to your agent what the specific task is to accomplish.

Example: “Guide users in discovering COMPANY NAME and products. Answer their frequent questions and help them solve issues about orders.”

3. Describing the Company

Include a section describing your company, highlighting key elements like the company name, history, vision, mission, and core values. This information gives your agent a foundation for its interactions.

Example: “The COMPANY NAME is a CORE BUSINESS company headquartered in CITY. Founded in YEAR in CITY, it is the world's largest company operating in the BUSINESS industry. COMPANY NAME is committed to being CORE VALUES. In an increasingly digital world, the goal is to MISSION through VISION. In YEAR, the company reported annual sales of approximately NUMBER billion, NUMBER employees.”

4. Tone of Voice

Specify the tone of voice for your Generative Agent. Depending on your company's branding and the nature of the interaction, this could range from professional and understanding to friendly and engaging.

Example: “COMPANY NAME's Al chatbot adopts a calm, professional, and understanding tone of voice. Its communication is user-centered, always attentive to the user's needs, and eager to find the best product to meet them fully. His writing style is short and concise, just like chatting with a human being. He avoids multi-paragraph or extremely long messages, prefers a chat, and often uses emoji to enrich his answers.”

5. Product List (Optional)

If relevant, provide your Generative Agent with a list of products, including descriptions and prices, enabling it to make recommendations or offer detailed information.

6. Answering Guidelines

Outline a set of rules for your agent to follow when generating answers, ensuring that its responses remain professional, relevant, and within the scope of its knowledge.

Example:

“If someone writes you something unprofessional, reply, "This is not a topic for me. Let's talk about COMPANY NAME!" or something polite.”
“If someone asks you about COMPANY NAME competitors, answer with "I am a COMPANY NAME expert. Ask me anything you want about our products!". You cannot make comparisons, and you cannot make other companies look bad. You have to be diplomatic. List some of COMPANY NAME's competitors: Competitor 1, Competitor 2, Competitor 3.
Never give email addresses.
Do not invent any type of product, but only choose from those in the #Products List.
Information on product descriptions and features are contained within #Product Lists and {{documents}}

7. Reference Documents (Optional)

If you wish for your agent to use specific documents or URLs when generating answers, include these as references, enhancing the accuracy and relevance of its responses.

In this section, you just have to put the variable {{documents}} as follows: “#Documents on which to base your answers: {{documents}}”

8. Sample Conversations

Provide examples of user and virtual assistant interaction to guide the generated answers' content and structure. Cases help in shaping the agent's conversational flow.

Example:

User: Can I kiss you?
Al assistant: This is not a topic for me. Let's talk about COMPANY NAME!
User: Payment methods
Al assistant: COMPANY NAME offers multiple ways to purchase. All major credit and debit cards accepted, as well as Apple Pay and PayPal
User: how much does shipping cost?
Al assistant: Shipping is free for major appliance product purchases $100 & above. Otherwise, for product purchases below $100, it has a cost of $10.00
User: What are the dimensions of Product 1?
Al assistant: Product 1 has a depth of 50 cm, a width of 60 cm and a height of 80 cm. For more details, visit the <a href="https://companyname.com" target="_blank">product page</a>.”

9. Initiating a New Conversation

Finally, describe how to initiate a conversation with a new user, emphasizing the importance of context in generating appropriate responses.

A step forward in using AI for business operations

Deploying a team of Generative Agents is a major step forward in utilizing AI to improve business operations, customer service, and overall customer experience. However, the true potential of these agents can only be realized through continuous improvement, monitoring, advanced customization, and a culture of innovation.

Try indigo.ai, in a few minutes, without installing anything.

Try a demo

Keep in mind that the goal is not just to automate tasks but to create a dynamic, intelligent entity that significantly contributes to your business's success. Many companies have already embraced this innovative approach by choosing the indigo.ai platform, which enables them to explore new horizons of operational excellence and customer satisfaction.

FAQs

How do these Generative Agents integrate with a company's existing CRM, ERP, or other business systems?

indigo.ai's Generative Agents are built to easily integrate with existing business systems like CRM and ERP platforms. The process involves customizing agents with company-specific information to ensure they use unique business data and insights. This customization makes the agents adaptable to different systems, allowing businesses to increase their operational efficiency without significantly changing their existing infrastructure.
The platform uses APIs or similar integration methods to connect with a company's data sources. This feature ensures that agents can access and utilize necessary information in real-time during customer interactions.

How does indigo.ai ensure the privacy and security of the business data fed into these Generative Agents?

indigo.ai follows standard security practices, including data encryption, secure data storage, and compliance with privacy regulations such as GDPR for European customers. Given the potential sensitivity of the information involved, the privacy and security of business data fed into Generative Agents are of the utmost importance.

What are the technical requirements for deploying Generative Agents using indigo.ai?

The no-code nature of indigo.ai’s platform allows businesses to customize and deploy agents without deep technical expertise. This power implies minimal technical requirements for deploying these agents, emphasizing user-friendliness and accessibility over complex technical infrastructure. Businesses would need to provide access to relevant data and documents and follow this guide to create Generative Agents and enable them to learn and interact effectively with users.