Ask Your Database from Slack

Most data questions do not start in a dashboard. They start in a Slack thread.

Someone asks "how many users signed up last week?" in a channel. An analyst sees it between meetings, opens a SQL client, writes a query, runs it, copies the result, and pastes it back into the thread. If no one with database access is around, the question sits unanswered until someone catches up.

The answer takes minutes to produce. The wait takes hours.

The Question Happens in Chat — the Answer Should Too

Limerence ships native integrations for Slack and Google Chat that let anyone on a team ask business questions in plain language — and get answers grounded in real data, right in the thread where the conversation is happening.

In Slack, mention the bot and specify which agent to use:

@Limerence [sales-agent] what were our top 10 customers by revenue last quarter?

@Limerence [sales-agent] what were our top 10 customers by revenue last quarter?

The bot parses the agent name from the brackets, routes the question to that agent, and replies in the same thread. Direct messages work too — type your question and the workspace's default agent handles it.

In Google Chat, no bracket syntax is needed. Mention the bot, ask your question, and the configured default agent takes it from there. Replies stay in-thread so context is preserved across follow-ups.

What This Replaces

The old workflow is not bad because the tools are bad. It is bad because there are too many steps between the question and the answer.

The difference is not the accuracy of the answer. It is how long the question has to sit there before someone gets around to it.

How It Works in Practice

Both integrations follow the same flow, whether you are in Slack or Google Chat.

1. 1

   Mention the bot in a channel or send a direct message. In Slack, you can specify an agent with bracket syntax — [sales-agent] — or let the workspace's default handle it. Google Chat always uses the configured default agent.

2. 2

   The agent processes your question. It translates the plain-language question into SQL, runs it against your connected data source, and streams the response back. In Slack, you see the answer build in real time as the response streams in. In Google Chat, the full response arrives as a single message.

3. 3

   Follow up in the same thread. Each thread maintains its own conversation context, so you can refine the question, ask for a different breakdown, or drill into the data. The agent remembers what you asked earlier in the thread.

Every thread is linked to the agent that handled it, so there is a full history of what was asked and what was returned. That is useful when someone else joins the channel later and wants to understand where a number came from.

Two Platforms, Same Architecture

Both integrations use a pull-based architecture.Pull-based means Limerence reaches out to the messaging platform rather than requiring the platform to push events to a public URL. This matters for self-hosted deployments behind firewalls — no need to expose a public endpoint or configure webhook URLs. Slack connects via Socket Mode — persistent WebSocket connections that do not require an inbound URL. Google Chat connects via Cloud Pub/Sub — Limerence subscribes to a topic and pulls messages from it.

For self-hosted deployments, this means no public endpoints to open and no webhook traffic to route. There are practical boundaries worth knowing: each Slack workspace has one default agent — channels can override with bracket syntax, but direct messages always use the default. Google Chat uses the configured default for all messages with no override. And responses longer than a few thousand characters are truncated on both platforms, so queries that return large result sets may need narrower filters.

What It Enables

When data access lives inside the team's chat tool, a few things change.

Non-technical team members start asking questions they would not have bothered with before. The friction of finding someone with database access and waiting for a response was enough to kill the question entirely. Now the question takes thirty seconds.

Conversations move faster. When a thread about campaign performance includes a live data answer, the discussion continues with real numbers instead of stalling while someone looks them up.

And the answers are visible to everyone in the channel. Instead of one person getting a private response from an analyst, the whole team sees the data in context. That changes how decisions get discussed.

Getting Started

Setup takes a few minutes per platform. In Limerence's settings, the integrations page walks you through connecting your Slack workspace or Google Chat space — including the tokens, scopes, and permissions each platform needs. Once connected, you assign a default agent and the bot starts listening.

The only prerequisite is a Limerence agent with a connected data source. If you have already set up an agent in the web interface, that same agent is available in Slack and Google Chat. Same knowledge, same access rules, different surface.