Skip to main content

Chat

The chat interface is where your conversations with Teela happen. It's designed to feel like messaging a knowledgeable colleague who knows your data inside and out. You ask questions, Teela answers, and together you can explore your data as deeply as you need to.

The Chat Interface Overview​

When you open Teela, the chat interface is front and center. Here's what you'll see:

  • The chat box at the bottom, where you type your questions.
  • The conversation area in the middle, where your questions and Teela's answers appear as a flowing conversation.
  • The sidebar on the left, where your saved threads (past conversations) are listed for easy access.
  • The connection selector at the top, which shows which data source you're currently querying. See Connections for more on switching between data sources.

Starting a New Conversation​

Click New Chat (or the "+" button) to start a fresh conversation. Each conversation becomes its own Thread that's automatically saved, so you can always come back to it later.

Starting a new conversation is useful when you want to switch topics. For example, if you've been asking about sales data and now want to explore customer support metrics, a new chat keeps things clean and organized.

Question Suggestions​

Not sure what to ask? Teela offers question suggestions to help you get started. These are pre-validated questions that Teela knows it can answer well based on your current data source.

You'll see suggestions when you:

  • Start a new conversation
  • Switch to a different connection

Just click any suggestion to ask it instantly, no typing required. Suggestions are a great way to discover what kinds of questions your data can answer.

Follow-up Questions​

One of Teela's most powerful features is its ability to remember the context of your conversation. After you get an answer, you can ask follow-up questions that build on it naturally.

For example:

You: "What were total sales last month?"

Teela: "Total sales last month were $284,500."

You: "Now filter to California"

Teela: "Total sales last month in California were $47,200."

You: "Break that down by product category"

Teela: (shows a table of California sales by category)

Teela carries the context forward, so each follow-up refines or expands the previous result. You don't need to repeat yourself. This makes it easy to start with a broad question and drill down into the details.

For more about writing effective queries, see Querying Your Data.

When Teela Asks You a Clarification Question​

Sometimes your question could mean more than one thing, or Teela isn't quite sure which data you're referring to. When that happens, Teela will ask for clarification rather than guessing. For a deeper look at how clarifications work and how to minimize them, see Working With Clarifications.

This can happen when:

  • Entity ambiguity: Your data has multiple things with similar names. For example, if you ask about "revenue" and your database has both gross_revenue and net_revenue, Teela will ask which one you mean.
  • Value mismatches: You reference a value that doesn't quite match what's in the data. For example, if you ask about "New York" but the database stores it as "NY," Teela will show you the options and ask you to pick the right one.

When Teela asks a clarification question, you'll see clickable options you can choose from. Just click the correct one to continue. This keeps things quick and easy, so you don't need to retype your question.

Conversational Messages​

Teela is more than a query engine. It can handle friendly conversation too. Feel free to:

  • Say hello: Teela will greet you back.
  • Ask "What can you do?": Teela will explain its capabilities and what kinds of questions you can ask.
  • Ask about your data: "What tables do I have?" or "What columns are in the orders table?" are perfectly valid questions.

These conversational interactions can be a great way to orient yourself, especially when you're working with a new data source.

Asking from Slack​

Teela also works in Slack. You can mention the bot in a channel, open a thread for a conversation, or send the bot a direct message. The conversation syncs back to your Teela account, so any Slack thread is also viewable and continuable on the web.

See Ask from Slack for the full guide.

Thread Management​

Every conversation is saved as a Thread in the sidebar. Threads make it easy to pick up where you left off or revisit a past analysis.

  • Auto-saved: You don't need to do anything. Every conversation is saved automatically as you go.
  • Rename: Give a thread a descriptive name (like "Q1 Revenue Deep Dive") by right-clicking or using the thread menu. This makes it much easier to find later.
  • Delete: Remove threads you no longer need to keep your sidebar tidy.
  • Navigate history: Scroll through the sidebar to find any past conversation. Threads are listed by most recent activity, with pagination as you scroll for long histories.

Archived Queries​

When you save a result as a DataClip, Teela automatically archives the thread it came from. This keeps your active sidebar focused on in-progress work without losing the original conversation.

To find an archived thread, click Archived Queries at the bottom of the sidebar. From there you can search across archived conversations, restore any thread back to the sidebar, or delete it permanently. There's also a "Clear all threads" option when you want a fresh start.

Threads are your conversation history. If you want to save a specific query result for repeated use, save it as a DataClip instead.

Asking "Why" Questions (Investigations)​

Teela can do more than answer "what" questions. Ask it "why" and it launches an investigation, running several queries in sequence, looking at the data from different angles, and synthesizing what it finds into a single narrative answer.

Try questions like:

  • "Why did revenue drop in October?"
  • "What's driving the increase in support tickets?"
  • "Why is the West region outperforming the East?"

Investigations come back with a written narrative, the supporting data, and suggested follow-up questions. You can export the full results as an Excel workbook or a branded PDF report straight from the chat. See Getting More From Your Questions for examples and tips.

"This Looks Wrong"​

If a result feels off, tell Teela. Say something like "that doesn't look right," "this seems too high," or "I expected more than this, can you check?" and Teela will run verification queries and explain what it finds. It's a useful second pair of eyes when a number surprises you.

Giving Feedback with Thumbs​

Every answer has a thumbs up and thumbs down button. Use them freely:

  • Thumbs up reinforces that Teela got it right. High-confidence positive signals feed into an automated learning system that can bake good answers into future results.
  • Thumbs down flags an answer as wrong. It goes into your admin's review queue along with an AI diagnosis of what likely went sideways, so they can correct it at the training level.

Either way, your feedback helps Teela get better for your whole team over time.

Explore and Drill Down​

Teela makes it easy to go deeper into your results without writing new queries.

Explore Aggregated Results​

When Teela shows you summarized data (like totals or counts grouped by category), you'll see an "Explore" button on each row. Click it to see the individual records that make up that summary.

For example, if you see "California: 142 orders," clicking Explore will show you all 142 individual orders from California.

Show Details on Single Numbers​

When the result is a single number (like a total or average), look for the "Show Details" button. Click it to see the underlying records that were used to calculate that number.

This is a quick and powerful way to go from "what happened" to "why it happened" without asking a separate question.

When you explore a record, Teela may show a Related Columns panel. This highlights connections between your data tables.

For example, if you're looking at an order record, the Related Columns panel might show you the associated customer details or the individual line items in that order. Click any related record to explore that connected data further.

This feature is especially useful when your data spans multiple tables and you want to follow the trail from one piece of information to another.

"Help Me Get an Answer" Button​

Sometimes a query doesn't work out the way you expected. Maybe Teela returned an error, or the results don't look right. When that happens, look for the "Help Me Get an Answer" button.

Clicking it gives Teela a chance to try a different approach. It might rephrase the query, ask you a clarifying question, or suggest an alternative way to get the data you need. Think of it as a friendly nudge that says, "Let's try this from a different angle."

Tips for Great Conversations​

  • Start simple, then refine. Ask a broad question first, then use follow-ups to zoom in.
  • Use natural language. You don't need to be formal or use technical terms. Just ask the way you'd ask a person.
  • Click suggestions when you're not sure what to ask. They're tailored to your data and guaranteed to work.
  • Save what matters. If you get a result you'll want again, save it as a DataClip so you don't have to re-ask the question.

What's Next?​