Customers send emails. Suppliers share documents. Software reports changes. Brezel connects these inputs with ownership, context and the next authorised steps.
A signal is incoming information that may matter to your business: an email, file, message or event from connected software. Email and webhook signals provide the technical foundation. The sources included depend on your configured connections.
An incoming item is not permission to act. First establish what it concerns, which customer or case it belongs to and who is allowed to handle it.
This diagram explains the relationship. It is not a live view or a commitment to future features.
A relevant signal becomes a case. The enquiry, supporting documents, open questions and work results belong together. Later inputs can add to the same context.
Your team does not have to start from scratch with each reply. An assistant can use the approved history, identify missing details and prepare a next action. Classification and business content still need to be reviewable.
Brezel separates incoming information, decisions and actions. The responsible person can review, edit or reject a proposal. Software and tools are used only within the authorised scope.
Sending a binding proposal, changing a record and involving an external adviser are different actions. Each needs the appropriate permission and, where required, approval.
A customer asks for a proposal, but the scope and preferred date are missing. An assistant classifies the enquiry, summarises what is known and drafts follow-up questions. Your sales team reviews the draft and decides whether to send it.
The customer’s reply adds to the case and may lead to further tasks. This example explains the principle; specific automation and sending channels require configured integrations and approvals.
A decision dashboard focused on outstanding human decisions, and automatic approvals for clearly bounded trusted cases, are development topics. They are not blanket promises of full automation included in today’s package.
Learning from approvals requires clear quality and knowledge boundaries. The evolving learning loop is therefore distinct from the underlying signal and case model.
AI support for real tasks. Within your rules. With people remaining responsible.
brezel