Short answer: How the Helpdesk Knowledge Base helps Aamu.app answer support questions from maintained sources instead of generic AI guesses.
Support answers need current product instructions, policies and known limitations. Fluent writing alone cannot supply those facts.
That is the role of the Helpdesk Knowledge Base in Aamu.app. It gives Aamu AI a way to retrieve relevant support knowledge before it answers or drafts. Instead of treating AI as a generic chatbot beside the work, Aamu connects Helpdesk AI to the maintained sources selected for that project.
This article focuses on the Helpdesk Knowledge Base as a curated support knowledge layer. For broader permission-aware workspace search, see Company Brain in Aamu.app. For the more specific customer-support workflow with Helpdesk drafts, Livechat, and human review, see the related article: AI customer support with Aamu.app: Helpdesk Knowledge Base, drafts, and human review.
Why company knowledge matters for AI
Most business questions are not answered well from general knowledge alone. A customer might ask whether a feature supports a certain setup. A teammate might need the current wording for a policy. A support person might need the safest way to explain a limitation. A visitor in live chat might ask how the product works before they decide to try it.
Without a knowledge layer, teams often work around the problem manually. Someone copies a policy into a prompt. Someone pastes product documentation into a chat. Someone asks AI a broad question and then checks the answer against old docs, tickets, and memory. That can help once or twice, but it does not become a reliable workflow.
Helpdesk Knowledge Base is meant to reduce that manual context work. The team maintains source knowledge in Aamu, and AI can retrieve the relevant parts when it needs to answer or draft.
What Helpdesk Knowledge Base is
Helpdesk Knowledge Base retrieves maintained support material selected for a project.
That knowledge can include material such as product explanations, support answers, policy notes, FAQ snippets, public pages, sitemap content, and Aamu Docs selected as knowledge sources. The useful pattern is simple: write down what the team knows, keep it current, and let AI retrieve it when a question needs grounded context.
In Aamu today, Helpdesk Knowledge Base knowledge sources can be added as Page URL sources, Sitemap sources, Manual FAQ snippets, and selected Aamu Doc sources. Helpdesk projects can also optionally use resolved Helpdesk tickets as knowledge sources when that setting is enabled. Resolved tickets are different from manually selected sources: they come from the support history, so they should be used only when old resolved answers are still useful and accurate.
What can become company knowledge
Good Helpdesk Knowledge Base source material is explicit and reusable. It should describe the answer clearly enough that another person on the team could use it too.
Useful source material includes:
Product how-to material that explains how features work and where settings live.
Support policies for cancellations, refunds, billing, data retention, account ownership, or escalation.
Setup instructions for email domains, Helpdesk, Livechat, forms, databases, API use, or project configuration.
Known limitations that should be explained consistently instead of improvised every time.
Tone guidance for how the team wants answers to sound.
Reusable support answers that come up often enough to deserve a maintained source.
Example: answering a product how-to question
Imagine someone asks:
Can Aamu create reply drafts for support emails without sending them automatically?
The knowledge base can retrieve the documented behavior: Helpdesk and Email AI prepare drafts, while sending is a separate action.
A grounded answer can then explain:
Yes. Aamu can prepare reply drafts for Helpdesk tickets and email workflows without sending them automatically. The draft appears in the normal reply editor, where a person can review, edit, and send it. Draft generation and sending are separate actions.
Example: answering a support policy question
Policy questions are where grounded knowledge becomes especially important. A customer might ask:
If we cancel today, will we be charged again next month?
AI should not improvise a billing policy. The answer should come from the team's actual rules. If the cancellation policy is written as Helpdesk Knowledge Base source material, Aamu can retrieve that knowledge before drafting an answer.
The team can then decide the right level of automation. A simple informational answer might be safe to draft quickly. An account-specific billing issue may need human review, because someone must verify the customer's actual subscription state.
Example: live chat and asking for a human
Live chat is different from a draft workflow because the visitor may see the answer immediately. When Allow AI to answer live chat messages is enabled for a project, Aamu Livechat can use Helpdesk Knowledge Base to answer common questions from company knowledge.
That works well for questions such as:
Does this feature exist?
Where do I find this setting?
How do I start using Helpdesk?
What does this product term mean?
Can Aamu help with this kind of workflow?
But not every live chat message is a knowledge question. A visitor may ask:
Can I talk to a real person?
A request for a person calls for a support handoff rather than another knowledge answer.
A useful response could be:
Sure. I can ask a human support person to take over this chat. Please add any details that would help them understand the issue.
Separately from Helpdesk Knowledge Base answers, the live chat widget may also have an email fallback when live support is not available. That is a widget or workflow behavior, not the same as an AI intent where the visitor asks the assistant to email them later.
How to prepare knowledge for better answers
Helpdesk Knowledge Base works best when source material is written in a way that can actually answer questions. A vague note is hard to retrieve and easy to misread. A clear answer, policy, or instruction is much more useful.
A practical preparation checklist:
Write policies as direct answers, including exceptions and escalation rules.
Keep product instructions current when settings, names, or workflows change.
Turn repeated support replies into maintained Docs or snippets.
Separate public-facing answers from internal-only notes when the distinction matters.
Include examples of the tone and level of detail the team wants.
Review weak AI answers as signs that the source knowledge may be incomplete.
Where human review still matters
Grounded knowledge does not make every answer safe to send automatically. Some questions involve account-specific facts, money, legal commitments, security, data deletion, customer emotion, or unclear context. Helpdesk Knowledge Base can help retrieve relevant material, but a person may still need to decide what to say.
This is why Aamu separates different levels of AI assistance. In Helpdesk and Emails, AI can prepare drafts for human review. In Livechat, AI answers can be enabled for real-time questions when the team has prepared the knowledge and accepts that level of automation. Asking for a human should be recognized as a different intent from asking a knowledge question.