BuildWithHQ
Canonical knowledge unit · ai-agents

AI Agents suggestion batches and human approval

AI Agents groups proposed AI actions into a suggestion batch tied to one Inbox item and its activity version, allowing BuildWithHQ to validate, review, approve, reject, or escalate the work before execution.

ID: BWHQ-GOCLAW-005Status: specifiedAuthority: highai-agentsactionsapprovalsuggestion-batchhuman-in-the-loop

AI Agents suggestion batches and human approval

Suggestion batch

AiActionSuggestionBatches is the human work package for one Inbox item. Proposed actions are validated against known action keys and input schemas before the batch is finalized.

Approval

Approval is a permissioned action in its own right. High-risk or low-confidence work can be routed to ReadyForApproval or NeedsHuman rather than executed automatically.

Canonical AI Agents actions

The Build Master includes deterministic action templates such as SendConversationReply, CreateFollowup, LinkRecords, PostMessage, and UpdateRecordStatus. New actions should enter through the same governed action registry rather than private ad-hoc mutation code.

Related knowledge

AI Agents planning and model callAI Agents planning uses routed AI models to classify the work, estimate risk/confidence, draft a response, and propose structured actions without performing application mutations during the planning step.Security invariantsSecurity is designed as a shared structural layer: server-derived identity, one authorization path, explicit tenant/application/location scope, and separate permissions for read, AI, export, and sensitive disclosure.
Machine-readable source: canonical Markdown. This HTML companion exists for human reading, search indexing, and AI retrieval.