Getting Started with Chimera Chat

Chimera Chat is a conversational AI built into the Content Chimera web interface. Instead of navigating menus and building charts by hand, you type questions in plain English and get answers about your website data — counts, tables, charts, and even summaries of what your pages actually say.

What You Need Before You Start

Chimera Chat works with data that has already been crawled and processed. Depending on the type of question you want to ask, different processing steps need to be complete:

Question type	What needs to be ready	Processing step
Structured queries (counts, distributions, field values)	The flattened table must exist	Optimize step complete
Semantic search (finding content by meaning)	The vector database must be loaded	Load Semantic DB step complete
Graph queries (broken links, link depth, orphan pages)	The graph database must be loaded	Load Graph DB step complete

If you ran a standard crawl pipeline, the Optimize step runs automatically. The semantic and graph databases may need to be loaded separately — you can load them from the My History page in the web UI, or by asking your AI assistant via MCP.

Tip

Not sure what is ready for your site? Ask Chimera Chat: “What data is available?” and it will tell you which query types are supported.

Two Modes of Asking Questions

Chimera Chat supports two fundamentally different ways of querying your data. You select which mode to use before asking your question.

Content Analysis (Structured Queries)

Content Analysis asks questions about the structured data in your flattened table: field values, counts, averages, and distributions. Think of it as a smart query builder that understands English.

Good for:

• Counting pages (“How many pages are in the /blog folder?”)

• Distributions (“Show me a bar chart of content types”)

• Averages and statistics (“What is the average word count?”)

• Filtering (“Which pages have a title longer than 70 characters?”)

• Comparing (“How does word count differ between /blog and /products?”)

Examples:

• “How many pages did we crawl?”

• “What are the top 10 folders by page count?”

• “Show me a treemap of content by folder”

• “What percentage of pages have a meta description?”

Semantic Search

Semantic Search uses AI to search the actual content of your pages — the words, topics, and meaning — rather than structured metadata. It has two sub-modes that Chimera Chat selects automatically:

RAG (Retrieval-Augmented Generation) — finds specific facts from a small number of relevant pages. Use this when you have a precise question.

• “Find pages that mention pricing”

• “What does the site say about return policies?”

• “Which pages discuss accessibility?”

Summarization — synthesizes broad themes across many pages. Use this when you want an overview.

• “What topics does this site cover?”

• “Summarize the site’s approach to sustainability”

• “What tone of voice does the blog content use?”

Note

Semantic search results are drawn from a sample of pages, not the entire site. They are excellent for exploration and hypothesis generation, but do not treat them as exhaustive. If you need precise counts, use a structured query instead.

When to Use Chimera Chat vs. the Web UI

Chimera Chat and the Web UI are complementary. Here is a rule of thumb:

Use Chimera Chat when you want to…

• Get a quick answer without clicking through menus

• Explore data you have not looked at before

• Ask follow-up questions that refine your understanding

• Generate a chart from a description rather than configuring options

Use the Web UI when you want to…

• Build persistent charts and save them to reports

• Set up rules and run rulesets

• Do manual page-by-page review with assignments

• Configure crawl settings, scraping patterns, or LLM fieldsets

• Manage projects, users, and permissions

Example Conversation

Below is a typical Chimera Chat session showing how questions build on each other.

You: How many pages did we crawl?

Chimera Chat returns a single number — for example, 4,218 pages.

You: Show me a treemap of content by folder

Chimera Chat creates an interactive treemap chart. You can see at a glance that /blog dominates the site while /products is comparatively small.

You: What percentage are blog posts?

Chimera Chat calculates the ratio — perhaps 62% of all pages sit under /blog.

You: (switch to Semantic Search mode) What topics does the blog cover?

Chimera Chat summarizes the main themes it found across a sample of blog posts.

You: Find blog posts that mention pricing

Chimera Chat uses RAG to locate specific pages whose content discusses pricing, returning URLs and relevant excerpts. (RAG vs. Summarization is selected automatically within Semantic Search mode.)

Each question takes seconds. The same exploration in the Web UI would involve navigating to charts, configuring fields, and opening individual pages — Chimera Chat lets you stay in a single conversation flow.

Tips for Better Results

• Be specific with structured queries. “How many pages have a word count above 500?” works better than “Tell me about word count.”

• Name the chart type if you have a preference. “Show me a bar chart of status codes” is more predictable than “Show me status codes.”

• Switch to Semantic Search for content questions. If your question is about what pages say rather than how many pages there are, select Semantic Search mode before asking.

• Ask follow-up questions. Chimera Chat maintains context within a conversation, so you can refine results step by step.