Before
Users likely had to copy questions into a separate AI tool, provide relevant Slack history manually, and then paste the response back into the channel or thread. Follow-up questions also depended on manually reconstructing the prior conversation and choosing the correct Slack location for the reply.
How it works
- A Slack command or message is classified as either a new request or a thread-based follow-up, while bot-generated events are excluded from the thread path.
- The workflow builds the GPT prompt from the submitted text and, for thread messages, retrieves and reuses the relevant conversation history.
- GPT generates the answer, after which the workflow posts a new Slack response or updates the existing thread message with the response and preserved context.
Workflow details
- Request intake: The workflow receives a Slack event or command, extracts the submitted text, and determines whether the request is a new command or a message in an existing thread.
- Conversation handling: For new commands, it creates an initial assistant context from the prompt. For thread messages, it retrieves the relevant Slack conversation and removes stored context markers so the prior exchange can be used as input.
- AI response generation: The assembled prompt and conversation context are sent to GPT, which returns a generated response. The workflow then adds that response to the maintained conversation context.
- Slack delivery: New command responses are posted to the specified Slack channel, while thread-based responses update the relevant Slack message with the original message, GPT response, and preserved context.
- Automated routing: The workflow distinguishes command requests from thread messages and avoids the thread-response path for events that contain a bot identifier, helping prevent automated replies from being treated as new user messages.
What changed
Slack becomes the request and response surface: the workflow automatically classifies the event, gathers relevant thread context, sends it to GPT, and delivers the result to the correct channel or thread without a manual tool handoff.
Work removed
Copying Slack questions into an AI tool, retrieving and assembling thread history, pasting generated answers back into Slack, selecting the correct channel or thread for delivery, and manually maintaining conversation context.
Expected outcome
Each request removes several recurring copy-and-paste and context-assembly actions, typically eliminating at least one manual AI-tool handoff per interaction.
