Skip to main content

Can’t find what you’re looking for? Join the Discord community or email support@bothive.cloud.

​
General

Because the hard part of building AI agents isn’t generating code — it’s everything around the model: memory, integrations, channels, observability, deployment, and multi-user state.If you build it yourself, you’re also responsible for auth and token refresh, session storage, retry queues, run logs, end-user identity separation, deployment infrastructure, and scaling. Bothive handles all of that as one unified platform.The real question isn’t “can I build it myself?” — it’s “do I want to maintain that stack once real users arrive?”
LangChain is a Python library for building LLM chains. Bothive is a complete platform.With LangChain, you still need to provision hosting, build your own memory system, implement each integration, manage OAuth tokens, wire up your deployment channels, and add observability separately.With Bothive: define your agent in HiveLang, connect integrations in the dashboard, deploy with one click. Everything else is handled.Full comparison: Observability
Vercel AI SDK is a JavaScript library — building blocks for developers. Bothive is a platform.Vercel AI SDK gives you primitives for calling models and streaming responses. You assemble the rest: hosting, memory, integrations, channels, observability.Bothive gives you everything assembled: HiveLang for defining agents, multiple integrations, multiple deployment channels, built-in memory, and per-run observability — out of the box.Full comparison: Bothive vs Vercel AI SDK
Yes — this is one of the primary use cases Bothive is designed for.Call the Bothive API from your backend, pass your user’s ID as the session ID, and the agent handles tenant-isolated memory, tool execution, and observability per user. Each of your users gets a fully isolated agent experience.See: Building SaaS Agents
Both. Bothive has three authoring modes:
  1. HiveMind — describe your agent in plain English, HiveMind builds it
  2. Visual Builder — configure instructions, tools, memory, and schedules in the UI
  3. HiveLang — write agent behavior in a readable declarative language
Non-technical users can ship production agents using HiveMind and the Builder. Developers can write HiveLang and deploy via the API.

​
HiveLang

HiveLang is Bothive’s native declarative language for defining agent behavior. It’s designed to be readable by anyone — not just developers.Instead of writing 50 lines of Python to set up an agent with tools and instructions, you write 10 lines of HiveLang:
HiveLang v5 supports: bot, swarm, agent, capabilities { }, instructions { }, before input { }, after response { }, on event { }, schedule, and loop.
No. The visual Builder and HiveMind AI assistant both produce the same agent definition under the hood. HiveLang is the underlying representation, but you never have to write it if you don’t want to.HiveLang is most useful for teams who want their agent definitions in version control and want to deploy via CI/CD pipelines.
HiveLang v5 introduces:
  • Curly-brace blocks for instructions { } and capabilities { } — no more string escaping
  • bot keyword at the top level (previously agent)
  • swarm blocks for multi-agent orchestration, native in the language
  • loop and if expressions for control flow inside event handlers
  • delegate to AgentName for swarm coordination
  • remember and recall built-in keywords for memory

​
Integrations

Bothive has 25+ native integrations including: GitHub, Gmail, Slack, Telegram, Google Calendar, Google Drive, Notion, Airtable, Stripe, Paystack, Lemon Squeezy, Jira, Linear, Twitter/X, Discord, WhatsApp (beta), OpenAI, Anthropic, Google Gemini, and Groq.Full list: Integrations
Yes. Any REST or GraphQL API can be connected as a custom integration. Define the endpoint, auth method (API Key, Bearer, OAuth2), request parameters, and response schema. Custom capabilities appear in the same capabilities { } block as native ones.Guide: Custom Integrations
Automatically. When you connect an OAuth integration (GitHub, Gmail, Google Calendar, Slack, etc.), Bothive stores the access token and refresh token securely. Before each tool call, it checks if the token is expired and silently refreshes it. You never need to re-authorize unless you revoke access.

​
Pricing & Plans

The Free plan includes: 3 bots, 1,000 runs/month, 2 custom integrations, Web and API channels, basic observability, and community support.It’s designed to let you build, test, and validate your first agent before committing to a paid plan.
A run is one execution of your agent — one user message processed and responded to. Runs include all the tool calls, memory lookups, and model inference that happen within that single request-response cycle.Scheduled bot executions count as one run each. Background tasks count as one run each.
Yes. The HiveStore marketplace lets you publish your agents for others to subscribe to. You set the price (monthly or per-use), users subscribe directly, and Bothive handles payments and revenue sharing.Your agent runs on Bothive’s infrastructure — no ops overhead for you.

​
Deployment & Channels

Web chat (hosted link + embeddable widget), Telegram, Slack, and REST API. WhatsApp is in beta.One agent definition deploys to all channels. Updating your agent propagates to all connected channels automatically.
Yes. Add one <script> tag to your HTML and the Bothive web widget appears as a floating chat button. Configure the color, greeting, position, and user identity. Full docs: Web Widget.
Yes. Use the REST API to call your agents from any server-side language. Pass a stable X-Session-ID per user for scoped memory. Supports streaming via SSE. Full reference: API Reference.

​
Support

  • Community: Discord server — fastest for general questions
  • Email: support@bothive.cloud — Pro and Business plans
  • Docs: You’re here — search is your friend
  • HiveMind Debug: Inside any run trace — one-click AI-powered root cause analysis
Email support@bothive.cloud with:
  1. Your bot ID
  2. A run ID that demonstrates the issue
  3. What you expected to happen
  4. What actually happened
Including a run ID is the single most helpful thing you can do — it lets us see exactly what happened in your trace.

​
Next reads

Quickstart

Discord Community

API Overview