All agents see the same conversation, attached files and results of earlier tool calls.
What if every agent knew what the others had already done?
Multiple agents, one conversation
Section titled “Multiple agents, one conversation”Specialized agents in LIVOI work within a shared chat context. Each agent has its own task, its own instructions, its own tools and, if needed, its own model. When the user switches agents, the new agent knows the conversation so far, including the results of tool calls.
Core principle: one conversation, clear responsibilities
Section titled “Core principle: one conversation, clear responsibilities”Picture a meeting with several experts: data analysis, sales, engineering and communications at the same table. You explain your request once, and each person picks up where the others left off as soon as you address them.
LIVOI works the same way: several specialized agents, one shared context. Nobody asks twice, nothing needs to be copied.
- Knowledge
- Custom tool
- Model
- Skill
- 1Knowledge Knowledge base agent
Finds product data and prices in the stored files.
rag_search_tool - 2Custom tool Sales agent
Gets the discount tier and delivery time live from the ERP and calculates the quote.
get_customer_conditions - 3Model Product advisor agent
Summarizes the technical benefits using a model suited to technical content.
- 4Skill Communications agent
Writes the quote email in the sales team's style.
Skill: quote email
The sales agent may query the ERP, the communications agent may not. Tools are assigned per agent.
LIVOI is model-agnostic. An agent for quick questions can use a fast model, an analysis agent a more capable one.
The user chooses the active agent directly in the chat. Attached files stay selected when switching.
15-second demo: feel the difference
Section titled “15-second demo: feel the difference”What changes day to day
- One do-everything bot with an overloaded prompt
- Information has to be entered multiple times
- Context is lost when switching tools
- Focused agents with a clear task and the right tools
- Results passed on seamlessly in the same chat
- Model and permissions chosen per agent
Benefits at a glance
Section titled “Benefits at a glance”- Clear separation of responsibilities and access rights
- The same context reused without losing data
- Scales by adding more specialized agents
- The right model for each task instead of one model for everything