Claude Opus 5.5 is now available in GenViz. It joins our growing menagerie of AI chatbots for conversation analytics: ask a question about your business data, inspect the answer, and keep asking until you understand what changed.
You can choose Opus 5.5 from the Anthropic section of the model picker. It works with the same connections, conversations, and analysis tools you already use in GenViz. Your next question can be as simple as “Which customers ordered less this month?” or as involved as “Explain the change, separate price from volume, and show me the rows behind it.”
Conversation analytics, with another model to choose from
Conversation analytics in GenViz means working through a data question in plain language. You ask, the assistant uses your connected data to investigate, and the next question builds on the work already in the conversation. It is also called conversational analytics. Here, we mean a conversation with your business data, rather than an analysis of call recordings or chatbot transcripts.
Opus 5.5 adds another choice alongside our existing Anthropic and OpenAI models. Organizations with enabled OpenRouter models can choose those too. Different questions call for different balances of response time, reasoning, and cost. You choose the model; the GenViz workspace keeps the question and its supporting work together.
Anthropic’s Opus 5.5 documentation describes a model built for complex knowledge work. In GenViz, it starts at medium reasoning effort, with options from low through max when you want to adjust how much reasoning a question receives. Start with the default and judge the answer against your data.
Follow the question past the first answer
Imagine a sales manager investigating a slow month. A conversation analytics workflow might begin with:
Compare this month’s orders with last month’s, using the same number of days in each period. Break the change down by customer.
Then comes the useful part: the follow-up.
Which five customers account for most of the decline? Show their order counts and average order values in both periods.
And then a check on the conclusion:
Exclude cancelled orders, show the dates you used, and save the comparison as a table I can inspect.
The value of conversation analytics is in that progression. You can refine the definition, narrow the question, and examine the evidence without restarting the analysis in another tool. Saved extracts keep the returned tables in the conversation, where you can open or download them.
When a question needs more than a database query, GenViz can also run Python analysis on those extracts. A calculation can combine tables, summarize a distribution, or produce another table to investigate. You can inspect the code behind the result in the same conversation.
How GenViz keeps conversation analytics bounded
Adding a capable chatbot is only part of making it useful on business data. The tools around it determine what it can reach and what you can verify. Opus 5.5 uses the same GenViz controls as the other models:
- Read-only database tools. GenViz rejects write-shaped queries before execution. Use a database account with read-only permissions as an additional boundary around the data you connect.
- Connections you configure. Queries run from the desktop app or a connector you host inside your network. The model works through GenViz’s tools; it does not receive your database password.
- Results you can inspect. Review the returned tables and the query or calculation behind an answer. Check the filters, joins, and date ranges before using a result to make a decision.
- Isolated Python analysis. Analysis code runs in a disposable sandbox without network access or database credentials, using only the extracts provided for that run.
Safe conversation analytics also means being clear about data movement. Your prompts and the context needed to answer them, including schema information and query results used by the assistant, are sent to the model provider. Saved extracts are stored in GenViz cloud storage so they can be opened across devices. Your database does not need to be copied wholesale into a new warehouse to get started.
Choose connections and model providers that fit your organization’s data requirements. On the Team plan, you bring your own provider API key. Our privacy policy explains the services involved, and the connections guide covers the databases GenViz can work with.
Try Opus 5.5 on a question you know
Open a GenViz conversation, choose Anthropic → Opus 5.5, and ask a question whose answer you can check. Inspect the result, then ask the follow-up that would normally send you back to a spreadsheet: another grouping, a tighter date range, or the records behind a total.
That is where conversation analytics earns its place in the working day: less time reconstructing how you got an answer, and more time understanding what the data says. Opus 5.5 is another member of the menagerie ready to help.
Explore conversation analytics in GenViz or download the app to try it with your own data.