What is an AI agent?

An AI agent is a software program designed to perform specific tasks autonomously. Think of it as a digital helper that can handle repetitive or time-consuming tasks, such as managing schedules, answering customer inquiries, or organizing data.

How are AI agents different from employees?

Employees typically work in roles that consist of multiple tasks requiring a range of skills and decision-making abilities. For example, a receptionist might answer calls, schedule meetings, and greet visitors.

AI agents, on the other hand, are task specific. They excel at performing one or a few focused tasks but cannot handle a full role. Multiple AI agents can collaborate on workflows, but they remain task oriented.

What are horizontal and vertical AI agents?

Horizontal AI agents are general purpose and can perform tasks across various industries or departments, like managing emails or generating reports.

Vertical AI agents are industry-specific, designed to address challenges unique to a particular field, such as healthcare scheduling or retail inventory tracking.

Can AI agents work together?

Yes! AI agents can collaborate to complete complex workflows. For instance, a scheduling agent might sync with a customer support chatbot to arrange follow-up meetings after resolving an inquiry. However, even when working together, they remain task specific and do not take on broad roles like human employees.

How can AI agents benefit my small business?

AI agents can:

• Save time by automating repetitive tasks.
• Enhance customer service with 24/7 availability.
• Reduce operational costs by handling tasks efficiently.
• Enable employees to focus on higher-value work.

Are AI agents customizable?

Yes! AI agents can be tailored to fit your business's specific needs. For example, an agent can be designed to match your workflows or integrate seamlessly with your existing software.

How do I choose the right AI agent for my business?

Identify tasks that are time-consuming or prone to errors. Determine if you need a:

• Horizontal agent for general tasks (e.g., email management).


• Vertical agent tailored to your industry (e.g., inventory tracking for retail).

Consult with AI specialists to design and implement the best solutions for your business.

What are AI agents made of?

AI agents combine traditional computer programming and large language models (LLMs):

• Traditional program code: Handles quantitative, deterministic tasks, ensuring predictable outputs based on input data.


• Large language models (LLMs): Manage qualitative, non-deterministic elements, generating semantically correct answers and reasoning through language-based tasks.

This hybrid approach allows agents to balance precision and flexibility in their operations.

How do AI agents “know” things?

AI agents utilize knowledge graphs to store and retrieve information. A knowledge graph is a structured database where entities (e.g., people, places, objects) and their relationships are mapped out. This enables agents to:

• Understand context and relationships between data points.
• Retrieve precise information efficiently.
• Augment LLM responses with verifiable, structured knowledge.

How do AI agents interact with external systems?

Our agents integrate seamlessly with Software-as-a-Service (SaaS) APIs. These integrations allow agents to:

• Access and manipulate data from external platforms like CRMs, scheduling tools, or financial systems.
• Automate workflows by bridging multiple services together.
• Provide real-time updates and actions through API calls.

Where do AI agents run?

Our AI agents can operate in two environments:

• Locally: Agents run directly on your company's hardware, offering enhanced security and faster response times for sensitive data.


• On the cloud: Agents leverage scalable cloud infrastructure, providing accessibility, ease of updates, and reduced maintenance for your team.

Agents that take over an employee's PC: This capability, enabling agents to act directly on a desktop environment, is still in the research and development phase.

What are the pros and cons of different agent frameworks?

There are multiple frameworks available for developing AI agents. Here's a comparison:

Framework	Pros	Cons
LangGraph	Well-suited for complex workflows and integrating multiple agents; strong focus on task orchestration.	Steep learning curve for new developers; requires significant customization for niche tasks.
CrewAI	Optimized for collaborative agent tasks; lightweight and efficient.	Limited documentation and community support compared to older frameworks.
AutoGen	Powerful for autonomous task generation and LLM integration; strong pre-built capabilities.	Resource-intensive, leading to potential performance bottlenecks in low-powered environments.
Swarm	Excellent for distributed agent systems; great for scalability and fault tolerance.	Complex setup and configuration; not ideal for simple, standalone agents.

What are the advantages of AI agents running locally versus cloud?

Local Deployment:

• Pros: Enhanced data security, reduced latency, better control over agent operations.
• Cons: Higher setup costs, limited scalability, and potential hardware constraints.

Cloud Deployment:

• Pros: Scalability, reduced maintenance, easy accessibility, and lower upfront costs.
• Cons: Reliance on internet connectivity, potential latency for real-time tasks, and ongoing subscription costs.

How was this website built?

1. Prompting Method: This website was built using a method called "prompting," where we ask computer programs specific questions, and it helps create content based on our questions.


2. Design Elements: We used a tool called DALL-E to create the website's visuals, like its icon and background. We kept trying different designs until we found one we loved.


3. Code Generation: Once we had our design, we used another tool, ChatGPT, to help write the code for our website. This code was written in languages: HTML, PHP, and JavaScript.


4. Testing: After creating the code, we put it on a test website (a local webserver) to see how it looked and worked. We created follow-up prompts based on what we saw.


5. Content Creation: For the written content you see on the site, we first made an outline of what we wanted to say. We then used ChatGPT again to fill in the details based on our outline.


6. Contact Page: Our contact page seamlessly integrates with HubSpot, a platform known for its user-friendly, no-code landing pages. Any information you submit through the form is securely stored in the HubSpot CRM. Leveraging reliable off-the-shelf software allowed us to meet our requirements without the need for custom development.


7. Final Touches: Before making the website live for everyone to see, we made some last-minute adjustments to the site's look, such as its colors and spacing (using a technology called CSS).


8. Going Live: Once everything was perfect, we put the website on the internet, and that is the version you are seeing now!

How are your services priced?

Our typical rate stands at $50/hour, although project-specific rates may vary. If part of the project is subcontracted out, we can coordinate with subcontractors for a 10% markup.

How do you handle proprietary data in my business?

We recommend using mock data. While AI's input and output are sent over the internet, any code designed to function locally will, ensuring your data's security. You may want to run a private cloud AI model or a fully private local one. We will help you determine the level of security appropriate for your business.