The platform behind your agent workforce
Versioned agents, bring-your-own-model providers, managed knowledge bases, channel deployment, and governance enforced on every operation — everything you need to run AI agents like production software.
Agents & Apps with Version Control
Bot apps are the user-facing front door; agents are the specialists behind them. Both carry full lifecycle management, and every agent change is a version you can publish — and roll back.
- Agent versioning with publish, changelogs, and rollback
- App→agent routing links: single, round-robin, intelligent, custom
- Lifecycle status (draft → active → paused → archived) and visibility controls
- Templates and bucket folders to organize the fleet
Bring Your Own Model
Register the LLM providers your organization has sanctioned — your keys, your contracts. Botmanor resolves vaulted credentials at execution time and records which model ran every single execution.
- 11 provider types: OpenAI, Anthropic, Google, Cohere, Mistral, Groq, Together, HuggingFace, Azure OpenAI, AWS Bedrock, and custom OpenAI-compatible endpoints
- Vaulted API keys resolved at execution time, never echoed back
- Connection testing before any agent depends on a provider
- Per-model parameters and per-agent model bindings
Knowledge Bases & Retrieval
Managed RAG corpora: documents are chunked with the strategy you choose, embedded into per-knowledge-base vector collections, and validated with semantic search before anything goes live.
- Five chunking strategies: fixed, recursive, semantic, markdown, code
- Per-KB vector collections with an explicit index lifecycle
- Semantic search to prove retrieval quality on real queries
- Reusable retrieval profiles, shareable as store packages
Channels & Team Inbox
Connectors bind your workspace to Slack, Teams, Discord, WhatsApp, and webhooks; routes map platform channels to bot apps. Every conversation lands in one shared team inbox.
- Inbound webhooks acknowledge instantly; agent runs happen in the background
- Outbound messages stored in the conversation record and delivered in parallel
- Channel routes: different channels, different bot apps, one workspace
- Shared team inbox with both sides of every exchange
Governance, RBAC & Cost Audit
Every API operation is gated by an RBAC rule enforced on every request, and every agent run writes an execution record with token and cost metrics. Trust is structural, not aspirational.
- Enforced RBAC for humans and machine credentials alike
- Per-execution audit: status, input/output, tokens, cost, model
- Policy definitions across seven types with tool-risk levels
- Prompt templates with variable rendering to standardize instructions
Your providers, your channels
Bring the model providers you already trust and meet users where they already are