The JEX Agent Workspace gives you fine-grained control over how your agent-
s use the knowledge base.
Temperature
Consider temperature as the "creativity dial" of your Agent.
- At 0, the Agent decides for itself whether to stick close to the source material
or be a bit more free.
- At 0.1, the Agent is extremely strict: it copies facts exactly as they are written in the knowledge base, with
.
- At 1.0, the Agent is highly creative: great for brainstorming, rewriting, or ideation for "
."
In short: Lower values = safer, fact-based answers. Higher values = more flexible,
creative output.
Similarity Threshold
This setting determines how closely a piece of knowledge must match the user's question
before the Agent uses it.
- A lower threshold means the Agent accepts looser matches (faster
answers, but sometimes irrelevant).
- A higher threshold means the Agent only accepts very close matches
(more precise, but might answer “I don’t know” more often).
In short: Use a higher threshold if accuracy is more important than always giving
an answer.
Reranking
Most search systems grab results quickly, but don’t always understand intent. A
reranker takes the first set of results and re-checks them with a smarter AI model.
Instead of just matching keywords or vector similarity, it looks deeper at the
relationship between the question and the content.
That means:
- More accurate answers for complex queries.
- Fewer "near misses" where the result looks close but isn't really relevant.
- Especially useful when you have large or messy datasets.
- Reranking slows down the Agent.
In short: Reranking ensures that the best results rise to the top, not just the first-
.
Number of Embeddings
Embeddings are the small chunks of knowledge your Agent looks at before answering
.
- More embeddings = more context, better coverage, but higher cost.
- Fewer embeddings = cheaper and faster, but the Agent might miss some
info.
- The sweet spot for most use cases is around 20 embeddings.
In short: Start with 20 and adjust based on cost versus coverage.
Built-in capabilities
- Email integration
Every Agent can send emails directly. Useful for escalations, follow-ups, or lead notifications. Agents can also receive incoming emails and trigger actions based on their content, for example, parsing an order request and sending it to your CRM.
- Agent-to-Agent communication
Agents can “talk” to each other inside your ecosystem. One Agent can pass tasks or information to another, chaining together more advanced workflows. You can trigger Agent-to-Agent communications in your prompts by typing @ and picking an Agent e.g. @sales-Agent
- Function calling
Agents can use functions to trigger actions in other applications, send data, or perform tasks beyond conversation. You can add a function to an Agent through the prompt editor by typing / and selecting the function you want from the dropdown menu.
- Custom-built connectors
. You can develop your own integrations for very specific use cases. And if you have a strong idea but no development capacity, our team can help you build it.
Quick customization
You can quickly customize an Agent by selecting it, going to the Customize
interface and selecting the appearance tab.
Agent interfaces
Each Agent can be deployed through different interfaces, depending on the use
. The interface defines how users interact with your Agent and how it appears.
Focus
A full-screen interface designed for internal or creative use cases.
It looks and feels similar to ChatGPT, but it connects directly to your branded and
fully customized Agents.
Best for:
- Workflow automation
- Creative writing or ideation
- Internal knowledge bases that teams need to quickly access
Embedded
An interface optimized to live on websites.
It’s designed to look familiar (similar to the AI tools people already use), but is fully
optimized for web journeys.
The embedded interface has two elements:
- Trigger: a button or input field (or both) that asks the user what they're on the website for
. This prompt sets the intent for the Agent.
Fully customizable, but some CSS knowledge is needed.
- Chat interface: appears in a modal overlay. Styling is limited to color (to en
sure usability), but it has been thoroughly tested for clarity anduser-
.
Best for:
- Support Agents on websites and online stores
- Discovery or lead conversion flows
- Guiding visitors to the right product or content
Tip: Always choose Embedded over Chat for public-facing agents. It feels natural to users and avoid
, and avoids associations with outdated chatbots.
Chat
A more traditional chat interface. It works, but it lacks the web-first optimizations
of the Embedded interface. Many users still associate “chatbots” with poor
experiences, so we don’t recommend it for customer-facing use cases.
Best for:
- Internal use
- Simple proof-of-concepts
- When you explicitly want a chat-only experience
In short:
- Focus = deep work, full screen.
- Embedded = websites, lead generation, customer journeys.
- Chat = legacy option, use only if you must.