Let us set the scene. It’s mid-morning, your coffee’s gone cold, and someone on the team pings in Slack asking:

“Hey, how many remote weeks can I get each year?”

You vaguely remember the answer hiding somewhere deep in the company playbook. A few Notion pages and bookmarks later, you’re still searching.

Understanding the struggle, we at CodeLink decided to build an internal assistant that could instantly answer company policy questions right in Slack, using a combination of Notion, a low-code orchestration tool (n8n), vector search, and a large language model.

No guessing page names. No digging through tabs. Just answers.

The Real Pain That Sparked It

At CodeLink, we document a LOT of useful things: company values, benefits, workflows, policies, you name it. But the more we document, the harder it gets to find stuff. 

Faced with the current number of 260+ pages, team members often:

• Slam CTRL+K in Notion trying random keywords

• Ask in Slack and hope someone knows

• Guess document titles like it’s a game of corporate Wordle

And when you finally do find the answer? It’s buried three sections deep in a page you didn’t expect.

The Idea of Letting Slack Just... Answer

We asked ourselves: “What if you could just type a Slack command like ‘What’s our remote meeting policy?’ and get an instant, accurate answer?”

That kicked off a simple but powerful internal initiative: using a lightweight Retrieval-Augmented Generation (RAG) workflow to connect Slack with our company knowledge, orchestrated entirely through n8n.

High-Level Workflow

1. Extract: Fetch Notion playbook content, including nested pages.

2. Chunk: Split content into manageable text blocks.

3. Store: Save these blocks in a vector database to enable semantic search.

4. Retrieve: When a Slack command is sent, convert the query into a vector and find the most relevant content.

5. Generate: Feed the results to an LLM to summarize the most relevant answers.

6. Respond: Send a human-friendly answer back to Slack with source references.

A high-level workflow to turn company policies to a Q&A Slackbot

Why We Chose Low-Code for Orchestration

Rather than building a full backend service to handle Slack events, query pipelines, and logic routing, we leveraged n8n to manage much of the integration and data flow logic.

Why That Worked Well for Us

• We could design and iterate workflows quickly without scaffolding a backend.

• Built-in integrations (Slack, Webhook, HTTP, conditional logic) saved a lot of time.

• It remains visual and accessible to both devs and technical stakeholders.

• Custom code is still possible using function nodes when needed.

This allowed our team to focus on the core value: getting information into people’s hands, fast.

Lightweight Slack-to-answer pipeline powered by n8n, from user command to response delivery

For example, instead of scaffolding a backend code logic just to call an embedding service, we simply configure an HTTP node in n8n. The model is specified, authentication is already handled via a predefined credential type of n8n, and the user’s input can be dragged directly into the request body. No boilerplate, no server code, just clean input/output that’s easy to visualize and test right inside the workflow.

An HTTP node configured to call the embedding API with dynamic input and built-in credentials, no backend code needed

Real Examples: The Bot in Action

Here are some real examples of how the assistant is being used inside Slack:

Example 1: Daily standup optionality

Example 2: Online meeting etiquette

What n8n Workflows Are (and Aren’t)

While n8n gives us incredible speed and flexibility, it’s not a catch-all solution.

Where it shines:

• Automating internal workflows

• Building quick prototypes or MVPs

• Bridging systems using APIs or webhooks

Where it falls short:

• High-scale, latency-sensitive production systems

• Scenarios needing deep observability or reliability

• Complex business logic with edge cases

💡 Pro tip: n8n is a jetpack for your workflow. But jetpacks aren’t built for long-haul flights. Use with context. Low-code tools are best seen as accelerators. They don’t replace solid engineering; they amplify it.

Why Not Just Use Notion Enterprise?

We get this question a lot, and it’s valid. Notion’s Enterprise Search is a robust and centralized solution for finding information across a large workspace.

But for our team, context is everything.

This assistant is scoped specifically to our internal playbook and lives directly inside Slack, the place where our team collaborates most. That means faster answers, less context-switching, and more relevance to how we actually work day to day. 

All in all, this small project isn’t meant to compete with or replace what Notion is building. Instead, it complements it. A lightweight solution focused on one high-impact use case: surfacing key internal answers fast.

Final Thoughts

This project wasn’t just about answering questions in Slack. It was about making company knowledge accessible, contextual, and usable, exactly when and where people need it. By combining the right tools, Notion for content, a vector database for search, an LLM for natural answers, and n8n for orchestration, we created a simple but powerful assistant that actually gets used.

So the next time someone asks, “What’s our policy on…”

You can just smile and say: “Ask Slack.”

And that’s the real win.