Core Lucien Concepts

Lucien is not a chatbot with plugins.

It is an operational intelligence layer that sits across your company’s tools, builds structured memory from activity, reasons over that memory, and proposes or executes work safely through auditable actions.

To understand how Lucien works, you need to understand these core systems.

1. Memory — Durable company intelligence

Lucien continuously builds a structured, long-term memory of your organisation.

This includes:

Conversations (Slack, Copilot threads)
Documents (Notion, website, internal docs)
Operational signals (metrics, events, integrations)
Decisions and their outcomes

Memory is not just chat history.
It is semantically indexed, time-aware, and linked to context such as organisation, conversation, and integration source.

When you ask a question like:

“Why did we defer that hire?”
“How many integrations per org?”
“What changed since last week?”

Lucien retrieves relevant memory fragments, reasons over them, and produces grounded responses and actions.

2. Context — What Lucien sees in the moment

Each conversation is augmented with contextual retrieval.

Before responding, Lucien:

Pulls relevant memory from long-term storage
Incorporates organization insights
References connected integrations
Reasons with context and makes decisions

Context is dynamically pulled as needed.

Lucien does not rely purely on what you type.
It reasons over:

Past conversations
Company structure
Active integrations
Pending plans and drafts
Key company insights (like financials, decisions, roadmap and goals)

3. Actions — Safe execution layer

Lucien rarely writes directly into your tools. Actions it deems risky (e.g sending emails) require human review and approval to execute.

All external work flows through Actions, which are:

Drafted first
Explicitly reviewable
Auditable
Executed only with approval (unless configured otherwise)

Examples:

Updating a Notion page
Sending a Slack message
Drafting an email
Creating a follow-up sequence

Actions separate reasoning from execution.

Lucien may think autonomously — but it acts deliberately.

4. Integrations — Operational surface area

Lucien connects to the tools where your company already operates:

Slack
Notion
Postgres
Website/product data
Mail
Calendar
Billing
and more

Integrations allow Lucien to:

Observe signals
Retrieve live data
Draft updates
Propose corrective actions

Integrations expand Lucien’s visibility.
Memory gives it continuity.
Actions give it leverage.

Together, they turn Lucien from “AI assistant” into “operational layer.”

How it all fits together

At a high level:

Integrations produce signals
Signals become structured memory
Memory is retrieved as context
Lucien reasons
Lucien drafts or executes actions

That loop runs continuously.

You can think of Lucien as:

Memory + Context + Reasoning + Actions
operating across your existing tools.

If you understand these systems, the rest of Lucien’s behaviour becomes predictable.

Use the sidebar to dive deeper into each concept.