Skip to content
Platform capabilities

Everything you need to build with AI

The Chat Agent is a fully managed platform that unifies every major AI provider, powerful tool integrations, autonomous agents, and real-time team collaboration — all under one roof, with nothing to install.

Conversations

Chat that keeps up with you

Switch model mid-thread, stream responses live, search your history, attach files, and stay in complete control of every system prompt.

Multi-model conversations

Connect OpenAI, Anthropic, Azure OpenAI, Google Gemini, Ollama, or any OpenAI-compatible endpoint — and switch provider or model mid-thread without losing context.

  • GPT-4o, Claude Opus/Sonnet/Haiku, Gemini 1.5 Pro/Flash
  • Swap model at any point in a conversation
  • Per-thread provider preferences

Live streaming responses

See tokens appear in real time as the model generates them. Streaming works across all supported providers and is preserved through tool-call roundtrips.

  • Token-by-token streaming for every provider
  • Streaming preserved through tool calls
  • Cancel generation at any time

Conversation history & search

Every thread is persisted and full-text searchable. Pick up where you left off or surface past insights in seconds from any device.

  • Full-text search across all threads
  • Infinite scroll with cursor pagination
  • Pin, archive, and label conversations

File attachments

Upload images, PDFs, and code files directly into the chat. The model reads and analyses your attachments inline, referencing them throughout the thread.

  • Images, PDFs, and plain-text files
  • Inline preview inside the thread
  • Attachment-aware context management

Rich markdown & code rendering

Responses are rendered with full markdown support — headers, tables, syntax-highlighted code blocks, and collapsible long outputs.

  • Syntax highlighting for 50+ languages
  • One-click copy for every code block
  • Collapsible long response sections

Prompt & system-message control

Set a custom system prompt per conversation, dial in temperature and max tokens, and build a reusable template library your whole team can share.

  • Per-conversation system prompts
  • Temperature, top-p, and max-token controls
  • Shareable prompt template library

Tools

Give the model superpowers

Ship with ready-made built-in tools or connect any REST API as a dynamic HTTP tool — the model decides when to call them, automatically.

Built-in tools

A curated toolkit available the moment you sign in: web search, code execution, calculator, date/time helpers, and more — zero configuration required.

  • Web search and URL browsing
  • Code interpreter with sandbox execution
  • Calculator, date/time, and utility functions

Dynamic HTTP tools

Turn any REST endpoint into a model-callable tool by supplying a URL, HTTP method, auth headers, and a JSON schema. No deployment needed.

  • Point-and-click tool builder in the UI
  • Full JSON Schema parameter definitions
  • Per-tool authentication and headers

Automatic tool-calling

The model determines which tools to invoke and when. Multi-step tool chains execute transparently — every call and its result are visible in the thread.

  • Zero-shot tool selection by the model
  • Multi-step and chained tool calls
  • Full call-and-result history in-thread

Tool result streaming

Tool outputs stream back in real time so long-running calls never block the interface. Intermediate results appear as they arrive, with progress indicators.

  • Partial results stream while tools run
  • Progress indicators for async operations
  • Graceful timeout and error surfacing
Multi-model chat · Tool-calling panel · Workflow editor — one unified interface

Agents & Workflows

Automate anything with AI agents

Build purpose-built AI agents and chain them into multi-step workflows that run autonomously on demand.

AI agents

An agent bundles a model, system prompt, and enabled tools into a reusable unit. Deploy once and call it from chats, workflows, or directly via the API.

  • Bind any model + system prompt + tools
  • Invoke from chat, workflow, or REST API
  • Version-controlled agent definitions

Multi-step workflows

Chain prompts, agents, and tool calls into a directed workflow. Each step can branch on conditions, loop until a goal is met, or fan out to parallel sub-tasks.

  • Visual drag-and-drop workflow editor
  • Conditional branching and while-loops
  • Parallel and sequential step execution

Reusable & shareable agents

Publish agents to your team library so colleagues can use them in their own chats and workflows — no duplicated configuration, always in sync.

  • Team-wide shared agent library
  • Clone and customise any shared agent
  • Per-agent usage analytics

Branching & sequencing

Route workflow execution based on model output, retry failed steps automatically, or run independent branches in parallel to cut total latency.

  • Output-conditional routing rules
  • Automatic retry with configurable back-off
  • Fan-out for fully independent parallel tasks

Collaboration

AI that works better together

Invite your team into shared workspaces, collaborate on chats and agents in real time, and manage access with fine-grained roles. Learn more about team collaboration.

Team workspaces

Organise your team into workspaces. Each has its own members, shared resources, model-spend budgets, and a centralised activity feed.

  • Multiple workspaces per account
  • Per-workspace model and cost limits
  • Centralised activity feed

Shared chats, agents & workflows

Share any chat thread, agent definition, or workflow with the whole team. Everyone works from the same resource — changes sync instantly for all viewers.

  • One-click sharing inside a workspace
  • Live sync across all concurrent viewers
  • Read-only link sharing for external stakeholders

Real-time presence & typing

See who is active in a shared thread, who is typing, and which agents are currently running — all without a page refresh, powered by WebSockets.

  • Live user presence avatars in the header
  • Typing and agent-running indicators
  • Sub-second latency over WebSocket

Roles & permissions

Assign Owner, Admin, Member, or Viewer roles per workspace. Control who can create agents, execute workflows, or manage API keys.

  • Four built-in workspace roles
  • Per-resource permission overrides
  • Invitation links with role pre-assignment

Operations & Security

Security & controls

Built-in analytics, encrypted key management, and role-based access keep you in full control. Everything is encrypted in transit and at rest — and your data is never used to train any model. Explore the security details.

Analytics & cost tracking

Monitor token usage, cost per provider, latency percentiles, and error rates across every model and user — updated in real time.

  • Per-user and per-model cost breakdown
  • Latency P50/P95 and error-rate charts
  • Budget alerts and hard spend caps

API-key management

Store provider API keys centrally in the platform. Keys are encrypted at rest and never exposed to browser clients or logs.

  • AES-256 encryption at rest
  • Scoped keys per workspace or individual user
  • Key rotation without service interruption

Role-based access control

Enforce least-privilege access across every resource. Owner, Admin, Member, and Viewer roles keep each person scoped to exactly what they need.

  • Granular resource-level permissions
  • Four built-in workspace roles
  • Per-resource permission overrides

Encryption & data privacy

All data is encrypted in transit (TLS) and at rest (AES-256). Your conversations, agents, and attachments are scoped to your account and never mixed with anyone else's data.

  • TLS encryption in transit
  • AES-256 encryption at rest
  • We never train on your data

Ready to see The Chat Agent in action?

Connect your providers, add tools, and have your first AI-powered workflow running in minutes. Free to start — no credit card required.