VectorChat Docs

Feature Overview

A complete summary of everything you can do with VectorChat.

This page provides a quick reference to every feature in VectorChat.

Core Features

FeatureDescriptionLearn More
AgentsCreate AI agents with custom instructions, persona, constraints, and model selection. Agents can use tools, search knowledge bases, and remember users across conversations.Agents
ConversationsChat with your agents in real time. Agents respond using streaming and can call tools, ask clarifying questions, and search your knowledge bases.Agents
Knowledge BasesUpload documents, crawl websites, or add text content. Shared across multiple agents. Agents search automatically using document search.Knowledge Bases
SkillsReusable instruction sets you can attach to multiple agents. Generate skills from example questions or write them manually.Agents
SchedulingRun agents automatically on a schedule (hourly, daily, weekly, monthly). Includes approval workflows and prompt templates with date variables.Scheduling
OrganizationsCreate team workspaces with role-based access (owner, admin, member, billing). Invite members, manage seats, and share agents across your team.Settings
Embeddable WidgetDeploy agents as chat widgets on your website. Customize theme, initial messages, position, and branding.Agents
API AccessUse OAuth client credentials to interact with agents programmatically.Settings

Available Integrations

IntegrationTypeCapabilities
GmailOAuthRead, search, send, archive, label, and delete emails
SlackOAuthSend, update, and delete messages. List channels, members, and users. Read message history.
Google CalendarOAuthCreate, list, update events. Quick-add events. Invite attendees. Auto-generate Google Meet links.
Google DriveOAuthSearch, read, create, and update files. List folder contents.
Email (IMAP/SMTP)CredentialsRead, search, send, reply, move, and delete emails via any IMAP/SMTP provider.
Calendar (CalDAV)CredentialsList calendars and events. Create, update, and delete events via any CalDAV-compatible calendar.
Web SearchBuilt-inSearch the public web for current information.
Web FetchBuilt-inFetch and read content from any public URL.

Developer Tools

ToolDescription
Execute CodeRun sandboxed JavaScript/TypeScript functions with schema-validated inputs and outputs. Configure timeout, memory, and CPU limits.
Data Collection QueryQuery structured data collections with exact-match filtering.
Conversation StateStore and retrieve JSON state snapshots per conversation.
Ask User QuestionPresent multiple-choice questions to users during a conversation.
Knowledge Base SearchSearch shared knowledge bases and return relevant document chunks.
User MemoryRemember and recall information about users across conversations.

Agent Intelligence

SettingDescription
Agent ModeMulti-turn reasoning with tool calls, self-correction, and reflection.
Think ToolInternal reasoning step the agent uses before responding. Can be made visible to users.
Autonomy LevelControls when the agent asks for confirmation before taking actions.
Persistent MemoryAgent remembers facts, preferences, and decisions about users across conversations.
Context RetentionConfigurable token budget for conversation history. Older messages are auto-summarized.

Agent Reliability Settings

SettingDescription
Max TurnsLimit how many conversation turns the agent can take per session.
Max Tool CallsLimit how many tool executions are allowed per session.
Token Budget per TurnMaximum tokens the AI model can generate per request.
Request TimeoutMaximum time allowed for each AI model request.
Reflections per TurnHow many times the agent can self-correct within one turn.
Reflections per SessionTotal self-correction budget across the entire conversation.