Question 1

What is Botmanor?

Accepted Answer

Botmanor is an enterprise-grade platform for building, governing, and deploying AI agents. Unlike basic chatbots that follow scripted flows, or developer frameworks that leave you to build the control plane yourself, Botmanor gives you the whole management layer: bot apps and the specialist agents behind them, versioning with publish and rollback, multi-provider LLM orchestration with vaulted keys, managed knowledge bases, channel connectors, MCP and A2A registries, and per-run cost and audit — all natively integrated into the Burdenoff Workspaces platform. Botmanor is pre-launch; the control plane is real today and the richer agent runtime is shipping in the open.

Question 2

What is the official tagline?

Accepted Answer

"Intelligent agents, deployed instantly." Botmanor is built around three pillars: speed to value (a workspace that is governed from the first minute, so idea to working agent is under 15 minutes), open architecture (bring your own model providers, the open MCP and A2A protocols, no proprietary lock-in), and enterprise trust (RBAC enforced on every operation, per-run cost and audit, and versioning with rollback — built into the platform, not bolted on).

Question 3

Who is Botmanor for?

Accepted Answer

Botmanor serves teams overwhelmed by repetitive inquiries and operational overhead: customer support leaders handling thousands of monthly tickets, HR teams answering the same benefits and payroll questions, sales teams qualifying inbound leads, IT help desks routing L1 troubleshooting, and engineering or platform teams who want an API-first agent platform instead of building one in-house.

Question 4

How is a Botmanor agent different from a chatbot?

Accepted Answer

A scripted chatbot follows a fixed decision tree and breaks the moment a question falls outside it. A Botmanor agent is a governed, versioned worker instead: it carries its own system prompt, model binding, and skills, it is published with a changelog and can be rolled back, and every run it makes is recorded with the model called, the tokens spent, and the cost. Around it sit the layers a serious agent needs — knowledge bases you build and validate, an MCP tool registry, an A2A agent registry, and policy definitions. Those layers are managed workspace entities today; consuming them inside the live answer (RAG grounding, in-chat tool calls, cross-agent delegation) lands with the agent runtime upgrade, and we say so rather than implying it already runs.

Question 5

How quickly can I deploy my first agent?

Accepted Answer

Botmanor is designed so you can go from signup to a working agent in under 15 minutes. Your workspace is bootstrapped with RBAC, audit, and billing already in place; you create a bot app, attach a specialist agent, register an LLM provider with your own key and run a connection test, and hold a chat with the agent — all in one sitting. Creating an app from a template is a single step that applies the template’s configuration as your app’s defaults.

Question 6

What platform does Botmanor run on?

Accepted Answer

Botmanor is natively integrated into Burdenoff Workspaces, a multi-tenant enterprise platform. That means workspaces, role-based access control, audit logging, notifications, billing, and authentication are inherited from a shared platform layer rather than bolted on — your agents live inside the same governed workspace as the rest of your team’s work.

Question 7

What is the App → Agent architecture?

Accepted Answer

Botmanor uses a two-tier hierarchy. An App is the user-facing bot your customers or employees talk to (for example "Customer Support Bot"). Behind it sit specialist Agents — internal AI workers with their own system prompts, knowledge, and skills (for example a Billing Agent, a Technical Support Agent, and an Escalation Agent). Each agent is linked to the app with its own routing configuration — priority, intent patterns, trigger keywords, a fallback agent, and a timeout — so one front door is designed to serve many specialisms. Automatic dispatch of a live message across those links lands with the agent runtime upgrade; today the links are configuration you author and version.

Question 8

How does an app decide which agent answers?

Accepted Answer

You author the routing, and Botmanor stores it as first-class configuration on the link between an app and each agent: a routing strategy on the app (single, round-robin, intelligent, or custom), then per-agent priority, intent patterns, trigger keywords (for example "invoice" or "refund" pointing at a Billing Agent), a fallback agent, a queue size, and a timeout. That configuration is versioned and reviewable today. Runtime dispatch — the app reading those patterns and picking the agent for a live message on its own — lands with the agent runtime upgrade; until then a conversation runs against the agent you address it to.

Question 9

What templates are available?

Accepted Answer

Templates in Botmanor are workspace entities, not a fixed catalogue we ship for you. An app template or agent template captures a reusable configuration — type, description, routing strategy, welcome and fallback messages, system prompt, model parameters — and creating from one applies that configuration as the defaults for a new app or agent, with usage tracked. Templates are private, workspace-visible, or public, and can be packaged and distributed through the store with linked or forked installs. Botmanor does not currently publish a curated per-industry template library; if that is what you need on day one, tell us on the waitlist and we will prioritise it.

Question 10

Can I create agents by just describing them?

Accepted Answer

Not yet from a single natural-language description. Today you author an agent directly — system prompt, model binding, parameters, skills, and knowledge — or start from a template or store package that carries a working configuration, then iterate with prompt templates that hold your reusable, parameterised instruction blocks in one versioned place. Natural-language generation of a whole app-and-agent structure is on the roadmap, not shipped.

Question 11

How do I organise a large number of apps and agents?

Accepted Answer

Botmanor provides bucket-based organisation for grouping related apps and agents, versioning of agent configurations so changes are traceable, and team-based development so multiple people can collaborate on the same bots inside a workspace with role-based permissions.

Question 12

Can an agent hand a conversation to a human?

Accepted Answer

Today a person takes over through the shared team inbox rather than through an automatic trigger. Every conversation — chat test, Slack thread, Teams channel, or inbound webhook — is stored centrally with both sides of the exchange and the execution record behind each reply, and your team supervises and intervenes from there under the same enforced RBAC as everything else. You can also model an escalation agent and give it intent patterns and trigger keywords now, so the topology is ready. Automated handoff — the app detecting a low-confidence or "speak to a human" turn and assigning a person mid-conversation — is designed and on the roadmap, not shipped.

Question 13

How do agents use my knowledge base?

Accepted Answer

Today you build and prove the corpus; grounding the live answer in it is the next runtime step. A knowledge base takes your documents, splits them with the chunking strategy you choose (fixed-size, recursive, semantic, markdown-aware, or code-aware), embeds them into that knowledge base’s own vector collection, and moves through an explicit index lifecycle — empty, indexing, ready, failed, rebuilding — so you always know whether it is queryable. You then validate it with semantic search: run the questions your users will actually ask and see exactly which chunks come back, with relevance scores. Injecting those retrieved chunks into the agent’s answer, with citations back to the source, lands with the agent runtime upgrade.

Question 14

What knowledge sources can I connect?

Accepted Answer

Documents you upload through the platform file service are the shipped ingestion path — each one is chunked with the strategy you pick, embedded into the knowledge base’s own vector collection, and tracked through its own processing status so a failed document is visible rather than silently missing. Automated site crawling and pull-based connectors for databases, cloud storage, and SaaS document sources are on the roadmap; until they land, anything you can export to a file can be ingested today.

Question 15

Can I tune how retrieval works?

Accepted Answer

Yes. Each knowledge base exposes retrieval configuration: vector, hybrid, or BM25-only mode; chunk size and overlap; top-K retrieved chunks; and a score threshold. Retrieval profiles capture a tuned recipe as a reusable object you can attach across agents or distribute through the store, so "the retrieval settings that work for legal documents" stops being tribal knowledge. Vector search is the implemented path today — hybrid and BM25 modes are stored configuration ahead of the runtime that consumes them.

Question 16

How does Botmanor keep answers accurate?

Accepted Answer

Three mechanisms are live today. Execution records: every agent run stores its status, input and output, the model called, token counts, and computed cost, so you can inspect exactly what happened rather than reconstruct it. Versioning: agents and apps publish with changelogs and roll back, so a prompt change that regresses quality is one rollback away. Retrieval validation: semantic search proves a knowledge base returns the right passages before it goes anywhere near a user. Grounded answers with citations, and confidence-based routing to a human, arrive with the agent runtime upgrade — we would rather name that boundary than blur it.

Question 17

Which LLM providers does Botmanor support?

Accepted Answer

Botmanor is multi-provider by design. The workspace provider registry covers 11 provider types — OpenAI, Anthropic, Google, Cohere, Mistral, Groq, Together, HuggingFace, Azure OpenAI, AWS Bedrock, and custom OpenAI-compatible endpoints — each registered with your own key, vaulted through the platform integrations service and resolved at execution rather than stored in an agent definition, and each verifiable with a connection test before an agent depends on it. Self-hosted inference (an Ollama, LM Studio, or vLLM server on your own infrastructure) connects through the custom OpenAI-compatible type. There is no single-vendor lock-in. OpenAI-compatible endpoints run in the live chat path today; native adapters for the remaining provider families are on the roadmap.

Question 18

Can different agents use different models?

Accepted Answer

Yes. Model selection is per-agent, so you can put a top-tier reasoning model behind your complex technical-support agent while a smaller, cheaper model handles simple FAQ routing. Temperature, max tokens, and top-p are tunable per agent as well.

Question 19

How does Botmanor control LLM costs?

Accepted Answer

Per-agent model selection lets you match model cost to task complexity, and every execution records its token usage and computed cost — so you can see exactly where spend is going, per agent and per run. You can register multiple providers and models with vaulted keys and switch an agent’s model binding as economics change, with each switch captured as a published agent version you can roll back.

Question 20

What agent technology is Botmanor built on?

Accepted Answer

The shipped control plane is Botmanor’s own service on the Burdenoff Workspaces platform, and the live chat path today calls your configured OpenAI-compatible provider directly. The richer runtime is being built on proven agent frameworks — Google’s Agent Development Kit (ADK) and Agno — unified behind a single API, with the open A2A protocol carrying cross-framework communication; Botmanor’s reference stack already proves that interop end-to-end across both frameworks. Either way you interact with one consistent platform and the framework plumbing stays our problem, not yours.

Question 21

Where can my agents talk to users?

Accepted Answer

Channel connectors and routes cover Slack, Microsoft Teams, Discord, WhatsApp, inbound webhooks, and a custom type for anything else — a connector holds the credentials, and routes map a specific location (a Slack channel ID, a Teams channel) to one of your apps, with per-route policy bindings and routing overrides. Inbound messages are acknowledged instantly while the agent runs in the background, and the same app can serve several routes at once. The embeddable web chat widget is modelled as a channel type but is not shipped yet.

Question 22

What is MCP and what can agents do with it?

Accepted Answer

MCP (Model Context Protocol) is the open standard through which agents will act, not just answer — and Botmanor ships the governance layer for it first, deliberately. Register an MCP server in your workspace registry and Botmanor tracks its connection status, runs a connection test on demand, and discovers the tools and resources it exposes; refresh the catalogue when the server’s surface changes. Policy definitions classify each tool by risk level — low, medium, or high — so the argument about which tools an agent should ever be allowed to reach happens in the open, before anything can call them. Invoking those tools inside a live conversation is part of the agent runtime upgrade and is not live yet; what you build today is the vetted, testable tool inventory that runtime will obey.

Question 23

Does Botmanor integrate with my existing business tools?

Accepted Answer

Through a generic framework rather than a catalogue of pre-built named connectors — which is the honest answer, and the more flexible one. The platform integrations service lets you register a provider under a category (CRM, communication, productivity, cloud storage, development, calendar, email, payment, analytics, accounting, or custom) with OAuth2, API-key, or basic auth, and vaults the credentials so agents and services resolve them at execution instead of holding them. On top of that you have inbound webhooks, an API-first surface, and the MCP registry for tool-based access. Named, ready-made connectors for specific SaaS products are on the roadmap; tell us which ones matter to you on the waitlist.

Question 24

Is there an API?

Accepted Answer

Yes — Botmanor is API-first. Every capability is exposed through a programmatic API, with webhooks for event-driven workflows and OAuth for secure third-party authentication, so you can manage apps, agents, knowledge bases, and conversations programmatically.

Question 25

Can agents work together on one question?

Accepted Answer

Not inside a live conversation yet — and the honest version is more useful than the marketing one. What ships today is the collaboration layer built on the open Agent-to-Agent (A2A) protocol rather than a proprietary scheme: a registry of known agents each carrying an explicit trust level (unverified, verified publisher, or internal), connections declaring which agents may talk to which, and structured messages with priorities, acknowledgements, and delivery status. That makes every hand-off a record instead of a hidden function call, and it lets architects draw the org chart for their agent workforce now. Live cross-agent delegation inside production conversations is on the runtime roadmap.

Question 26

How does Botmanor work with FluidGrids?

Accepted Answer

FluidGrids is Burdenoff’s visual workflow automation product, and Botmanor integrates with it for complex orchestration: conversations can trigger workflows, intent-classification nodes route to the right agents, conditional if/else branching reacts to agent responses, and action nodes create tickets, send notifications, or update your CRM — all assembled in a drag-and-drop canvas.

Question 27

Can I see what my agents are doing?

Accepted Answer

Yes, and this one is fully shipped. Every agent execution is recorded: which agent version ran, which model it called, the input and output, token counts, computed cost, status, and timing — rolled up into per-agent, per-app, and daily usage statistics. Conversation history from every channel is stored centrally in the shared team inbox, and the platform audit trail captures configuration changes. When someone asks what the bot said on Tuesday and why, you open the thread and then the execution behind it — no reconstruction.

Question 28

How is my data protected?

Accepted Answer

Botmanor runs on the Burdenoff multi-tenant platform. Every Botmanor entity — app, agent, knowledge base, provider, execution, conversation — is workspace-scoped, and access to it is governed by role-based access control enforced on every operation, so separation between tenants is an authorisation guarantee applied on every request rather than an honour system. Data is encrypted at rest and TLS-protected in transit, credentials and provider API keys live in managed secret stores and resolve at execution instead of being written into an agent definition, and OAuth secures third-party integrations.

Question 29

What compliance standards does Botmanor target?

Accepted Answer

Botmanor is built for GDPR alignment (data subject access requests, right to erasure, data portability) and CCPA. Our controls are designed against SOC 2 criteria, and a HIPAA Business Associate Agreement is planned for healthcare customers on the Enterprise plan. We deliberately label in-progress items as in progress rather than overclaiming.

Question 30

How do roles, permissions, and audit work?

Accepted Answer

Botmanor inherits enterprise governance from the Burdenoff platform: RBAC is enforced for every operation, workspace roles control who can create, edit, deploy, or delete bots, and a complete audit trail records all operations. SSO/SAML is part of the Enterprise tier.

Question 31

Are my conversations used to train AI models?

Accepted Answer

No. Botmanor orchestrates the LLM providers you configure; your knowledge bases and conversations are used to serve your agents, not to train foundation models. With multi-provider support you choose where inference runs — including Azure OpenAI for enterprise data boundaries, or a custom OpenAI-compatible endpoint pointed at a model server you host yourself for content that must not leave your infrastructure.

Question 32

When can I use Botmanor?

Accepted Answer

Botmanor is currently pre-launch. Join the waitlist at botmanor.com/waitlist and we’ll notify you the moment early access opens — waitlist members get first access to onboarding slots and launch announcements.

Question 33

How much will Botmanor cost?

Accepted Answer

Pricing is being finalised and will be announced at launch. We plan a free tier for evaluation alongside paid tiers for teams and enterprises, with transparent per-user pricing rather than opaque per-session metering. Join the waitlist and we’ll share the full pricing the moment it’s ready.

Question 34

What will the Enterprise tier include?

Accepted Answer

The Enterprise tier is designed for large teams and regulated industries: unlimited apps and agents, multi-agent orchestration, SSO/SAML, advanced RBAC and audit, custom MCP servers and LLM endpoints, priority support with a dedicated CSM, and stronger SLAs. Contact us via the waitlist or contact form to discuss enterprise requirements early.

Question 35

Will I be locked in?

Accepted Answer

Botmanor is explicitly designed against lock-in: multi-provider LLM support means you can switch or mix model vendors at any time, knowledge bases are built from your own documents and sources, and the API plus webhooks keep your data and configuration programmatically accessible.

Frequently asked questions

Still have questions?