Vega’s performance is driven by context. The more precise the context, the better the output.
AI agents are how Vega gathers that context from the right sources at the right time. When you know how they work and how to trigger them, you get:
more accurate answers in the Vega chat
better pre-meeting preparation sheet and post-meeting analysis
stronger email drafts
Vega’s AI agents
What they are and how they work
At a high level, Vega is not a single model answering everything in isolation.
Instead, Vega acts as an orchestrator. When you ask a question, or when Vega processes a meeting or an email, it evaluates what the request actually requires and then decides which specialized AI agents to activate to handle each part of the work.
What is an AI agent in Vega?
An AI agent is a specialized component inside Vega that has:
domain-specific expertise
access to dedicated data sources
a well-defined role in the overall reasoning process
Rather than treating every request the same way, Vega routes pieces of your request to the AI agent (or agents) best equipped to handle them.
Each agent knows:
where to look for the right data
what information is relevant
how to interpret that information for the user's workflows
This is what allows Vega’s answers and outputs to feel informed, precise, and grounded in your actual environment, instead of generic AI responses.
In practical terms, you can think of Vega as coordinating a panel of subject-matter experts rather than relying on a single generalist.
How Vega uses RAG under the hood
This is also where retrieval-augmented generation (RAG) comes into play.
RAG is the process by which Vega:
retrieves the most relevant information from the AI agents (firm data, economic data, market data, and more)
injects that retrieved context directly into the AI’s reasoning process
then generates the final response using both your request and the retrieved information
Because each agent performs its own targeted retrieval before the final answer is composed, Vega’s responses are always built on current, specific, and context-aware data, not on generic training knowledge alone.
In other words: AI agents gather the right context first. Then Vega thinks and responds.
The main AI agents in Vega
The AI agents inside Vega were created specifically for advisor workflows.
Each agent is responsible for a distinct domain of knowledge. When Vega processes a request, it activates only the agents that are relevant. This is how Vega maintains both accuracy and speed.
Below are the main AI agents in Vega today.
CRM data
The CRM data agent retrieves information stored in your CRM, including contacts, meetings, notes, emails, phone calls, tasks, and SMS conversations. It gives Vega a complete understanding of your client relationships and communication history so it can generate personalized meeting preparation, follow-ups, CRM updates, and chat responses using real relationship data.
Chat query example*: "Give me the latest interactions with John Doe."
Research
The Research agent retrieves trusted external information that lives outside your firm’s data. Depending on your request, Vega activates one or more specialized research agents, including:
Tax (IRS): contribution limits, tax rules, retirement accounts, deadlines, and IRS guidance.
Economy (FRED): inflation, interest rates, employment, GDP, and other macroeconomic indicators.
Markets (Bloomberg): market performance, asset classes, securities, and financial market context.
Web: curated financial publications, regulatory sources, and other authoritative web content.
Chat query example*: “What are the 2026 IRA contribution limits for individuals under 50?”
Planning data
The Planning data agent retrieves information from your financial planning software, including financial plans, scenarios, assumptions, and planning outputs. This allows Vega to answer planning-related questions using your clients’ actual financial plans instead of generic assumptions.
Chat query example*: “Summarize John’s retirement plan and identify the biggest planning risks.”
Investment data
The Investment Data agent retrieves portfolio and investment information from your investment management platforms. It helps Vega understand portfolio allocations, holdings, performance, account values, and other investment-related information so it can provide more informed meeting preparation and client support.
Chat query example*: “Summarize my client’s current portfolio allocation and recent performance.”
Suggested actions
The Suggested actions agent analyzes meetings, emails, and conversations to identify CRM updates that should be made automatically. Rather than simply suggesting updates, this agent prepares structured CRM changes that can be applied with little or no manual work, reducing administrative overhead.
Chat query example*: “What CRM updates do you recommend from my meeting with Sarah?”
Calculator
The Calculator agent performs exact mathematical calculations whenever numerical precision is required. Unlike a language model, which may estimate or reason through arithmetic, this agent executes deterministic calculations to ensure accurate results.
Chat query example*: “If a portfolio grows 7% annually for 15 years, what will it be worth starting from $500,000?”
Knowledge center
The Knowledge center agent retrieves information from files and documents that have been uploaded into Vega, including firm documentation, procedures, playbooks, marketing material, compliance documents, and other internal knowledge. This allows Vega to answer questions using your firm’s own documentation and processes.
Chat query example*: “What is our firm’s process for opening a new advisory account?”
Support
The Support agent acts as Vega’s built-in product expert. Any time you ask how something works in Vega, how to configure a feature, or how to troubleshoot an issue, this agent is activated to provide precise guidance.
Chat query example*: “What are the most important things I should know when getting started with Vega?”
*Disclaimer: These examples are written as chat queries you can use inside the Vega chat. However, these AI agents are not limited to chat. They are also triggered automatically by Vega’s other features, including pre-meeting preparation sheet, post-meeting analysis, automatic email drafts, and more.
**Important note on data quality: The Research Web agent does not “browse the entire internet.” Vega intentionally restricts this agent to a curated set of relevant, trustworthy, and authoritative sources. For example, Vega will never retrieve information from community forums like Reddit or other unreliable sources the way ChatGPT often does. This design choice is what allows Vega to deliver professional-grade, compliance-conscious answers rather than generic web content.
Together, these agents allow Vega to reason across your firm’s internal knowledge, your client relationships, real-world financial data, external regulatory and market information, all within a single output.
Audit Vega's AI agents
One of Vega’s strengths is transparency. You can see exactly which AI agents were used for a given response and understand how Vega built that output.
In the Vega chat, Vega Outlook plugin, and Vega Gmail assistant
The AI agent system runs under the hood when you use the chat inside the Vega chat, the Outlook plugin, and the Gmail assistant.
If an AI agent is used for a chat response, Vega will include the source links at the bottom of that response.
Pre-meeting preparation sheet
For pre-meeting preparation, Vega primarily relies on the Contacts agent. This agent pulls context from your CRM and your communication history, including: past emails, CRM notes, tasks, and contact records.
Depending on how your meeting templates are configured, additional agents may also be activated to enrich specific sections.
For each major section of the preparation sheet (such as Recent activity, Vision and concerns, etc.), Vega explicitly shows the sources it used to generate the content.
This allows you to trace every insight back to its origin and understand exactly how the preparation was constructed.