Skip to main content

Threads

📌 This article is for users of elvex 2.0. Find out which version you're using.

Threads in elvex

Threads in elvex are ongoing dialogues between you and the AI platform, either through the Home Agent or a specific agent. They allow you to maintain context and build upon previous interactions.

Starting a New Thread

Click the + in the sidebar to start a new thread. From there you can choose your starting context — your personal context, a Space, or a specific agent — before sending your first message.

Consider starting a new thread when:

  1. You're beginning a completely new topic or task.

  2. You want to ensure there's no lingering context from previous discussions.

  3. You're switching between different projects or areas of focus.

  4. You want to start fresh.

When to Return to an Old Thread

Returning to an old thread can be beneficial when:

  1. You need to access information or context from earlier in the thread.

  2. You're iterating on ideas or solutions discussed previously.

How Context Works with LLMs in elvex

Large Language Models (LLMs) used by elvex maintain context within a thread to provide more relevant and coherent responses.

  1. Short-term Memory: The LLM retains information from recent messages in the thread, allowing it to understand and respond to follow-up questions or references to previously discussed topics.

  2. Context Window: There's a limit to how much previous thread history the LLM can consider, often referred to as the "context window." This varies depending on the specific model being used.

  3. Relevance Decay: As the thread progresses, older information becomes less relevant, and newer information takes precedence in shaping the AI's responses.

  4. Resetting Context: Starting a new thread effectively resets the context, allowing you to begin fresh without any lingering information from previous discussions.

  5. Datasources and Instructions: elvex can enhance the LLM's context by incorporating information from datasources and following specific instructions set for the agent.

Finding Old Threads

  1. Use the search function to find specific threads by keywords, dates, or topics.

  2. Browse through your thread history in the sidebar, organized chronologically.

Pinning Threads to a Space

When a thread contains useful context, decisions, or outputs worth preserving, you can pin it to a Space. Pinning a thread turns it into a resource — meaning it becomes part of the Space's shared foundation and flows into every future conversation that takes place within that Space, automatically.

This is how individual work becomes institutional knowledge. A thread you pinned about a key project decision, a customer call debrief, or a research summary is now available to everyone in the Space without them needing to find or attach it manually.

To pin a thread, open the three-dot menu on the thread and select Pin to Space.

Best Practices for Managing Threads and Context

  1. Be Explicit: If you're referencing something from earlier in the thread, be clear about it to help the AI understand the context.

  2. Use Clear Transitions: When switching topics within a thread, clearly indicate the change to help the AI adjust its context.

  3. Start Fresh When Appropriate: Don't hesitate to start a new thread when shifting to a significantly different topic or when you want to ensure a clean slate.

  4. Leverage Datasources: Utilize elvex's datasource feature to provide additional context and information to the AI when needed.

  5. Pin threads with durable value: If a thread contains decisions, outputs, or context your team will need again, pin it to the relevant Space so it becomes a shared resource rather than a personal artifact.

By understanding how threads and context work in elvex, you can more effectively communicate with the AI and make the most of your interactions on the platform.

Did this answer your question?