Documentation

AI Equipment Assistant

A conversational AI assistant that helps you find and add equipment to your projects using natural language queries.

Overview

The AI Equipment Assistant understands your equipment needs and searches your product catalog to find matching products. Instead of manually searching and filtering, simply describe what you need in plain language.

Natural Language

Describe equipment like you would to a colleague: "I need a ceiling mic for a conference room"

Smart Search

Finds products by meaning, not just keywords - understands synonyms and related terms

Quick Add

Review suggestions and add products to your project with one click

Before You Start

The AI Equipment Assistant requires products to have embeddings generated. Without embeddings, the assistant won't find any products.

For best results, enrich your products prior to running embeddings. Anytime a product is enriched, its embedding will be removed and it will need to be re-embedded. You can choose to embed without enrichment, but it is recommended to include the enriched description in the embeddings for better downstream results.

One-time setup: Go to Products Click Generate Embeddings

How It Works

Powered by our 4-Stage Agentic AI Pipeline — the assistant doesn't just search, it understands your request, reasons through options, and refines until it finds the perfect match.

1. Parse
AI understands natural language and AV terminology
2. Search
Semantic search finds relevant products in your catalog
3. Select
Agentic AI reasons through options and refines results
4. Review
Confirm selections and add to your project
Tip: Save Credits with Specific Model Numbers

Using specific model numbers (e.g., "HD-TX-4KZ-101") results in direct matches and lower costs. Generic descriptions may require refinements.

Better Results with Enriched Products

Products with enriched descriptions produce significantly better matches. Use "Enrich with AI" (Products page) to add synonyms, use cases, and technical details.

Using the Assistant

Accessing the Assistant

  1. Open a project
  2. Click the Actions dropdown (top of project details)
  3. Select AI Equipment Assistant
  4. Type your equipment request and press Enter

Example Queries

Good queries:

  • "I need a wireless mic system for presenters"
  • "Find me a 4K PTZ camera with USB output"
  • "Add 2 ceiling speakers for the lobby"
  • "HDMI matrix that supports 4K60"

Too vague:

  • "Camera" - add specs like 4K, PTZ, USB
  • "Speakers" - add quantity, type, location
  • "Audio" - specify mic, DSP, amp, etc.

Reviewing & Adding Products

After processing, you'll see suggestions for each item:

  • High confidence (90%+) - AI is confident in the match
  • Needs review (below 80%) - Review carefully, consider alternatives
  • Change dropdown - Select a different product from alternatives
  • Quantity field - Adjust how many to add

Click Add to Room to add selected products, or continue the conversation for more equipment.

Tips for Best Results

  • Be specific - Include requirements like "wireless", "4K", "Dante", "PoE"
  • Mention the application - "for conferencing", "for digital signage"
  • Include constraints - "under $500", "ceiling mount", "USB connectivity"
  • Specify quantities - "2 mics", "4 speakers for the lobby"
  • Iterate - If results aren't right, refine your query with more details

Improving Search Quality

For best results, prepare your product catalog:

  1. Generate descriptions - Use "Fill with AI" for products missing descriptions
  2. Enrich products - Use "Enrich with AI" to add comprehensive content
  3. Generate embeddings - Required! Run "Generate Embeddings" on Products page

Credit Usage

The AI Equipment Assistant uses credits for each AI operation:

  • 1 credit for parsing your request
  • ~1 credit per item for AI selection (direct match)
  • +1 credit per refinement if the AI needs to search for alternatives

Typical costs:

  • Simple query with exact matches ("qm65c, neatbar") = ~3-4 credits
  • Multi-item query ("display with mount, tx and rx") = ~5-7 credits
  • Complex query with refinements = ~8-15 credits

Tip: Using specific model numbers (e.g., "HD-TX-4KZ-101") results in direct matches and lower costs. Generic descriptions may require refinements.

Troubleshooting

  • No products found - Products must be enriched AND have embeddings. Run "Enrich with AI" first, then "Generate Embeddings"
  • Wrong products suggested - Be more specific; use model numbers when possible. Ensure products are enriched for better context
  • "Insufficient credits" - Purchase more at Billing > Credits (coin icon in top bar)
  • Low confidence matches - The AI may refine its search automatically. Use the Change dropdown if needed
  • Missing TX or RX from a kit - Ask for "transmitter and receiver" separately, or add kit products to your catalog