AI Agents for in-company use

AI Agent for product manuals

With Eloquent you can easily create an AI Agent that helps staff work with product manuals and error codes. Many products generate error codes when issues occur. Looking these up manually takes time and can lead to mistakes. With an AI agent, your personnel can get instant explanations and step-by-step guidance to solve the problem faster and more reliably.

How it works

The agent is trained on the product manual, which is often available as a PDF or in structured documentation. You can upload these documents directly, and refresh them regularly to keep the knowledge base accurate. If the manufacturer publishes an online knowledge base, you can also connect it through a sitemap or integration to keep everything up to date automatically.

Our recommendations

  • Keep it fresh → Manuals and error codes change over time. Update the knowledge base whenever a new version is available.

  • Embed it where it matters → Place the agent in the apps, portals, or intranets where staff already work.

  • Use the right interface → The focus interface works best for staff training or problem-solving. The embedded interface can integrate into existing apps or web portals.

What can it do?

  • Answer questions about product manuals and error codes.

  • Guide staff step by step through troubleshooting processes.

  • Provide instant explanations, saving time and reducing reliance on paper manuals or scattered files.

What pain points does it solve?

  • Reduces time wasted searching for error codes and solutions.

  • Minimizes errors from misinterpretation of manuals.

  • Improves consistency in how staff troubleshoot issues.

  • Ensures knowledge is always up to date, even when manuals change.

  • Frees up senior staff by making knowledge accessible to everyone.

Who it’s for

  • Manufacturers → to support internal service teams.

  • Technical support providers → to equip their staff with instant troubleshooting help.

  • Field engineers and maintenance teams → who need guidance on the job.

  • Any organization that relies heavily on complex product documentation.

How to set it up

We’ll soon be releasing templates so you can just select this agent from the creation menu. Until then we recommend the following settings:

  • Agent temperature: 0.1

  • Reranking: On (when dataset is above 300 pages)

  • Embeddings: 20

  • Similarity: Default

  • Prompt template: Fill in the product details and troubleshooting criteria specific to your case.

# INSTRUCTIONS (PRODUCT HELP)

## ROLE
You’re an expert in technical product support, manuals, and troubleshooting guidance for {company-name}.

## GOAL
You help users understand and use {product-name} by providing clear, step-by-step explanations, interpreting error codes, and walking them through technical instructions with precision.

## TASK
- Answer questions as briefly as possible while ensuring complete clarity.
- Always give unambiguous instructions. If that’s not possible, ask exactly one clarifying question before answering.
- Be friendly but professional—users should feel supported and confident.
- Structure answers so users can follow steps easily (use ordered lists or bullet points).
- Provide direct solutions first. If multiple scenarios are possible, list them clearly.
- If the question is outside your scope, refer the user to {contact-details} and state it’s beyond your intended knowledge.
- Never answer questions outside manuals, troubleshooting, or product use cases.
- Never ask more than one question at a time, and only if it truly makes sense.

## LANGUAGE
Your default language is {default-language-settings}. Always respond in the same language the user uses.

## FORMATTING
- Use **Markdown** for clarity (steps, scenarios, error code tables).
- Keep the style conversational but precise.
- Use **bold** for important terms (e.g., buttons, settings, codes).
- Always format e-mail addresses, urls, and clickable resources with Markdown

Example in action

A manufacturer of industrial machines trained an agent on their technical manuals. Field engineers now use the agent through a tablet app: when an error code appears, the agent explains the issue and walks them through the solution step by step.
The result: faster troubleshooting, fewer mistakes, and reduced downtime for clients.

Vorige

WhatsApp