What it does
6 features
moinAI · AI chatbot for customer communication with configured actions
Product and access
moinAI is a proprietary product of knowhere GmbH in Hamburg. The provider calls it an "agentic AI platform"; at its core it is a knowledge-based chatbot with actions through webhooks.
Core features
6 features from provider sources · checked on 5 November 2026. Availability and limits are directly below.
AI agents for customer conversations
AI agents answer from assigned knowledge sources. Standard agents are meant for simple information, specialised agents get instructions, resources and AI actions.
Source 12AI actions
A webhook action calls a predefined webhook and processes the response, for example for availability, CRM lookup, shop search or ticket creation. There are also follow-up actions, web search and CSV actions.
Source 18Knowledge Base with RAG
You decide which sources the chatbot may learn from. According to the page nothing is generated without reliable information. Individual content can be switched on or off per agent.
Source 6Live chat and handover
According to the page agents forward complex or sensitive requests to responsible staff. The workflow catalogue mentions handovers in Microsoft Teams.
Source 2Guardrails and compliance check
The optional compliance check blocks non-compliant answers and can slow generation. A competitor list prevents mentions, communication rules control style.
Source 14Templates for processes
The page shows ready-made chatbot workflows such as address change, customer verification, appointment booking and contract information from SQL. It does not describe triggers, steps and error handling.
Source 3
Usable AI models
Fixed choice across providers
The product page names models from the GPT series or open-weight models, while the technology page speaks of proven models in Germany without naming them. The pricing page lists processing via Azure in Western Europe and an EU option at OVH. The pages checked do not say how you choose or which model runs on your plan.
Model families named
- OpenAI (GPT)
- Model choice
- Not stated
- Your own API key
- Not stated
- Router or custom endpoint
- Not stated
- Local models
- Not stated
The rating covers the generally available main access. Partly documented · Sources accessed on
Does it fit my task?
2 practical tasks · 6 requirements and limits
A good fit if …
- you want customer requests on the website or in messengers answered from your own knowledge sources
- individual queries such as delivery status or appointments should run into your systems through webhooks
- you want a guided rollout with handover to a live chat
Less suitable if …
- you want an agent that plans tasks itself or runs on a schedule without a customer conversation
- you want to check the host, model providers and sub-processors publicly before signing
- you need short terms or a self-service test without a conversation
Editorial assessment based on provider documentation. Not a product test.
Practical tasks
Editorial examples based on documented features. No product tests performed.
Check delivery status
- Your input
- A specialised agent with a webhook to the shipping system and an instruction for follow-up questions.
- Possible result
- The agent asks for the order number, fetches the status and states it in the chat.
- What to check
- Check the action used in the AI Playground and store texts for missing data.
Prepare an address change
- Your input
- A follow-up action that leads to the form after an answer.
- Possible result
- With a positive reply the form starts and a webhook transfers the data.
- What to check
- Test the instruction, because the action does not trigger without it, and check the transfer in the target system.
Have an agent fetch delivery status through a webhook and hand over to live chat when there are problems.
Your possible prompt
Handle this customer request using the approved knowledge sources. Prepare a sourced response and hand unclear cases to the service team. Check permissions before taking actions in customer systems.
Adjust the task, data and permissions.
Requirements and limits
Limitation
Actions are predefined
Each webhook belongs to one use case and is created and published in the hub. The help article recommends several clearly separated actions and calls complex search agents less reliable.
Source 18Limitation
Actions limited per plan
The pricing page limits "AI agent actions" to 1, 5 or 10 per plan and process integrations to 1 or 3.
Source 4Limitation
MCP only for the Knowledge Base
The moinAI MCP server opens the Knowledge Base to external AI tools such as Claude or Cursor. The reviewed pages do not describe connecting third-party MCP servers as agent tools.
Source 20Limitation
Unsuitable content
The help article names sensitive content such as dosages or allergens as well as granular figures as unsuitable for AI agents.
Source 11Not yet confirmed
Model providers
The agent page names the GPT series or open-weight models, the technology page OpenAI only for internal processes and tailored models in the product. No reviewed page names concrete providers for operation.
Source 6Not yet confirmed
Independent multi-step work
The marketing page speaks of step-by-step processes and direct execution. The help articles show webhooks, forms and follow-ups for this, but no step sequence planned by the agent itself.
Source 2
Can I use it?
Availability still unclear
Europe & regional availability
GDPR page, help article, technology page and linked privacy policy reviewed. The privacy policy mainly describes the website. The data processing agreement is available only on request.
Reviewed scope: Hosted platform; location statements without a named host
Reviewed on:
Availability still unclear
Location
The GDPR page says everything is hosted in Germany. The help article shows in a figure a data centre in Europe without US cloud providers. Neither page names a host.
Region, reason and source
Region covered: Germany or Europe
Documented reason: The two statements differ in scope; both come from the provider.
Official source · www.moin.ai · Official source · helpcenter.moin.ai
Availability still unclear
Model providers and training
The agent page names the GPT series or open-weight models. The technology page names OpenAI for internal processes. According to the article a model for voice recordings runs on its own infrastructure, without training. The pages say nothing about training on chat content.
Region, reason and source
Region covered: Open
Documented reason: No reviewed page says where models for chat answers run or who operates them.
Official source · www.moin.ai · Official source · www.moin.ai · Official source · helpcenter.moin.ai
Availability still unclear
Data processing agreement
According to the pricing page the data processing agreement is part of all plans. According to the GDPR page you can view it on request by email. The text was not reviewed.
Region, reason and source
Region covered: EU/EEA
Documented reason: Existence is documented, content and sub-processors are not public.
Official source · www.moin.ai · Official source · www.moin.ai
Availability still unclear
Sub-processors and certificates
The reviewed pages name no sub-processors and no certificates for the product. The privacy policy names website service providers such as HubSpot, Cloudflare and Zapier, and no AI provider. "GDPR compliant" is a provider claim.
Region, reason and source
Region covered: EU/EEA
Documented reason: Not publicly documented; the help article refers to the Customer Success team.
Official source · www.moin.ai · Official source · helpcenter.moin.ai · Official source · www.moin.ai
Accounts & permissions
Based on provider documentation. Permissions depend on the account, integration and organization policies. Review the permissions actually requested before use.
Configuration covered: Hosted platform with administration in the hub
Sources reviewed:
Show all 3 points
Documented with conditions
Account used
SSO uses OpenID Connect with Google ID and Microsoft Entra ID. Availability depends on the licence, and Customer Success must enable it. The reviewed pages do not mention SCIM.
EvidenceDocumented with conditions
Data access & team sharing
Agents reach only what webhooks and integrations release. Credentials sit in the webhook's headers or Basic Auth. According to the article the MCP server's API key grants wide access to the Knowledge Base; deleting there is final.
Evidence · Evidence
Memory and conversation continuity
4 of 6 criteria documented
Contexts are the memory of a conversation and end with the session. Storage across sessions is not provided.
Context only within a session; no memory feature documented
Sources accessed:
Documentation by criterion
StoragePartly documented
Storage and scope
Information is held as named variables in the session context, such as user_name. The article does not name the physical storage location.
SelectionDocumented
Selection and timing
Variables come through the widget API, from forms or from webhooks. They are set automatically or as configured. The article does not describe selection by the model.
Edit & deleteNot publicly established
Inspect, edit and delete
The article describes no view for editing or deleting individual contexts. Managers can delete conversations.
Access boundariesDocumented
Agent and channel boundaries
Context applies per session and can be used in agents, forms, webhooks and text placeholders. A level across sessions or channels is not provided.
RetrievalDocumented
Retrieval
Variables are available through placeholders such as {{user_name}}. Agents know date and time automatically. Semantic retrieval for conversation knowledge is not described.
New conversationsDocumented
Session reset and compaction
When the session expires, the information is deleted automatically according to the article. The page says nothing about compaction.
Cost and effort
3 plans · Prices as of 5 November 2026
Prices and included usage
Pricing checked
Three plans with "from" prices. The page names no trial, but a free chatbot draft and a demo. According to the page there is no setup fee.
| Plan | Price | Included |
|---|---|---|
| Business | From EUR 750 paid annually, from EUR 880 paid monthly, each per month | 15 AI agents, 1 AI agent action, 1 process integration, 20 MB Knowledge Base, 3-month minimum term. |
| Growth | From EUR 1,350 paid annually, from EUR 1,580 paid monthly, each per month | 35 AI agents, 5 AI agent actions, 3 process integrations, 50 MB Knowledge Base, 6-month minimum term. |
| Professional | From EUR 2,500 paid annually, from EUR 2,930 paid monthly, each per month | 75 AI agents, 10 AI agent actions, 3 process integrations, 100 MB Knowledge Base, 12-month minimum term. |
- According to the page all plans include unlimited conversations, a data processing agreement, HubSpot, Salesforce and Zendesk connections and webhooks.
- Third-party price information was not used.
Setup
Request a free demo or chatbot draft. According to the FAQ a chatbot can go live within a few weeks; you set up webhooks and AI actions in the hub.
Operation & oversight
Based on public provider documentation. Assessments are editorial inferences from the stated properties. Actual time requirements and reliability have not been measured.
Deployment covered: Hosted operation by the provider
Sources reviewed:
Editorial assessment
moinAI is well documented for supervised customer conversations but narrow in terms of independent agent work: actions are set up in advance, and runs depend on a conversation. For the host, model providers and sub-processors you need the data processing agreement or a request.
Show all 6 points
Documented
Files and working state
According to the help article conversations stay in the hub until the end of the contract by default. Managers set a deletion period from 30 days to "never", whereas the glossary says "never to 1 year". The Knowledge Base is limited to 20 to 100 MB per plan.
Evidence · Evidence · EvidenceDocumented with conditions
Resuming after failures
On errors a form leads the person back to re-enter data, and conditions and fallback texts catch missing data. The article does not describe retry logic.
Evidence
Evidence and open questions
21 sources · 6 open points
Open points
The reviewed sources contain no reliable information on these points. A gap says nothing about the product.
Europe: Availability still unclear
Requirements and limits: not yet confirmed
Sources and evidence
Official website, pricing page, privacy pages and help articles reviewed. Feature descriptions come from the provider; example tasks are editorial. The data processing agreement is available only on request and was not seen. Our own hands-on tests are still pending.
Editorial team: Michael Meierhoff · sources reviewed on . Provider information is not independent evidence of performance.
Provider
knowhere GmbH, based in Hamburg, offers the chatbot platform moinAI.
- Operator
- knowhere GmbH
- Organization
- Provider
- Website
- www.moin.ai
Changes to this profile 1
AI chatbot platform from Hamburg for customer service, marketing and sales, with AI agents that fetch data through webhooks and hand over to people.
These notes describe editorial changes. Source and pricing review dates appear in their respective sections.