> For the complete documentation index, see [llms.txt](https://docs.praiz.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.praiz.io/get-started/key-concepts.md).

# Key concepts

The Praiz vocabulary on one page: conversation, transcript, agent, conversational database, Insights, Smart Actions.

## Conversation

A conversation is a call captured by Praiz: a sales call, a customer meeting, a check-in, an onboarding session. Praiz records on three channels: video meetings (a bot joins the meeting), phone calls (Praiz pulls the call from your phone system) and in-person meetings (mobile app).

## Transcript

Praiz transcribes every conversation automatically, a few seconds after the call ends. Agents work on that transcript. See "Transcription and languages".

## AI agent

An agent reads a conversation and always fills in the same fields, defined in advance. Import a ready-made agent from the library, or build your own. There are three types of agent:

| Type       | What they do                                        | Examples                                                    |
| ---------- | --------------------------------------------------- | ----------------------------------------------------------- |
| Generation | Produce content from the call                       | Summaries, follow-up emails, customer stories, testimonials |
| Scoring    | Evaluate the call against a method or your playbook | Call scoring, MEDDIC, SPICED, BANT evaluation               |
| Tracking   | Detect and structure signals                        | Objections, competitors, churn risks, product feedback      |

See "What is an agent?".

## Conversational database

Every agent output feeds a structured database. Every call is filled in the same way, so the information can be filtered, sorted and analyzed: by period, by team, by rep, by customer, or by any field an agent extracted.

## Querying your conversations: Ask Praiz, Insights and the MCP server

Praiz offers three ways to query your conversations. What sets them apart is their scope.

| Feature    | Scope                     | Where you use it                            | What for                                                       |
| ---------- | ------------------------- | ------------------------------------------- | -------------------------------------------------------------- |
| Ask Praiz  | One call                  | From the call page                          | Digging into one specific point of a conversation              |
| Insights   | All of your conversations | From the Insights view                      | Surfacing aggregated trends and signals, inside Praiz          |
| MCP server | All of your conversations | From your AI tool (Claude, ChatGPT, Gemini) | Querying your database in plain language, from an outside tool |

## Automation

Once the data is extracted, Praiz acts on its own.

* **CRM sync**: Praiz fills in your CRM fields from the agent's fields, not just the call note. See "CRM sync".
* **Smart Actions**: Praiz sends an agent's output for a call according to your rules, by email or to a Slack channel. See "Smart Actions".
* **Recurring Insights**: Praiz sends scheduled reports that aggregate several calls, at the frequency you choose, on Slack or by email. See "Recurring Insights".

## Related pages

* The "Glossary" gives the definition of each term and its label in the application.
* To record your first call, follow the "Quickstart".