feat: showcase text generation and thinking steps from suggested prompts

Author: zimeg

ai/llm_caller.py

@@ -1,27 +1,17 @@
 import os
-from typing import Dict, List
 
 import openai
 from openai import Stream
 from openai.types.responses import ResponseStreamEvent
 
-DEFAULT_SYSTEM_CONTENT = """
-You're an assistant in a Slack workspace.
-Users in the workspace will ask you to help them write something or to think better about a specific topic.
-You'll respond to those questions in a professional way.
-When you include markdown text, convert them to Slack compatible ones.
-When a prompt has Slack's special syntax like <@USER_ID> or <#CHANNEL_ID>, you must keep them as-is in your response.
-"""
-
 
 def call_llm(
-    messages_in_thread: List[Dict[str, str]],
-    system_content: str = DEFAULT_SYSTEM_CONTENT,
+    prompt: str,
 ) -> Stream[ResponseStreamEvent]:
     openai_client = openai.OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
-    messages = [{"role": "system", "content": system_content}]
-    messages.extend(messages_in_thread)
     response = openai_client.responses.create(
-        model="gpt-4o-mini", input=messages, stream=True
+        model="gpt-4o-mini",
+        input=prompt,
+        stream=True,
     )
     return response

listeners/assistant/assistant_thread_started.py

@@ -1,5 +1,4 @@
 from logging import Logger
-from typing import Dict, List
 
 from slack_bolt import Say, SetSuggestedPrompts
 
@@ -18,24 +17,19 @@ def assistant_thread_started(
         logger: Logger instance for error tracking
     """
     try:
-        say("How can I help you?")
-
-        prompts: List[Dict[str, str]] = [
-            {
-                "title": "What does Slack stand for?",
-                "message": "Slack, a business communication service, was named after an acronym. Can you guess what it stands for?",
-            },
-            {
-                "title": "Write a draft announcement",
-                "message": "Can you write a draft announcement about a new feature my team just released? It must include how impactful it is.",
-            },
-            {
-                "title": "Suggest names for my Slack app",
-                "message": "Can you suggest a few names for my Slack app? The app helps my teammates better organize information and plan priorities and action items.",
-            },
-        ]
-
-        set_suggested_prompts(prompts=prompts)
+        say("What would you like to do today?")
+        set_suggested_prompts(
+            prompts=[
+                {
+                    "title": "Prompt a task with thinking steps",
+                    "message": "Wonder a few deep thoughts.",
+                },
+                {
+                    "title": "Generate a release announcement",
+                    "message": "Please write detailed changelog notes for a feature that almost seems to be magic.",
+                },
+            ]
+        )
     except Exception as e:
         logger.exception(f"Failed to handle an assistant_thread_started event: {e}", e)
         say(f":warning: Something went wrong! ({e})")

listeners/assistant/message.py

@@ -1,8 +1,9 @@
+import time
 from logging import Logger
-from typing import Dict, List
 
 from slack_bolt import BoltContext, Say, SetStatus
 from slack_sdk import WebClient
+from slack_sdk.models.messages.chunk import MarkdownTextChunk, TaskUpdateChunk
 
 from ai.llm_caller import call_llm
 
@@ -13,6 +14,7 @@ def message(
     client: WebClient,
     context: BoltContext,
     logger: Logger,
+    message: dict,
     payload: dict,
     say: Say,
     set_status: SetStatus,
@@ -34,47 +36,96 @@ def message(
         thread_ts = payload["thread_ts"]
         user_id = context.user_id
 
-        set_status(
-            status="thinking...",
-            loading_messages=[
-                "Teaching the hamsters to type faster…",
-                "Untangling the internet cables…",
-                "Consulting the office goldfish…",
-                "Polishing up the response just for you…",
-                "Convincing the AI to stop overthinking…",
-            ],
-        )
-
-        replies = client.conversations_replies(
-            channel=context.channel_id,
-            ts=context.thread_ts,
-            oldest=context.thread_ts,
-            limit=10,
-        )
-        messages_in_thread: List[Dict[str, str]] = []
-        for message in replies["messages"]:
-            role = "user" if message.get("bot_id") is None else "assistant"
-            messages_in_thread.append({"role": role, "content": message["text"]})
-
-        returned_message = call_llm(messages_in_thread)
-
         streamer = client.chat_stream(
             channel=channel_id,
             recipient_team_id=team_id,
             recipient_user_id=user_id,
             thread_ts=thread_ts,
         )
 
-        # Loop over OpenAI response stream
-        # https://platform.openai.com/docs/api-reference/responses/create
-        for event in returned_message:
-            if event.type == "response.output_text.delta":
-                streamer.append(markdown_text=f"{event.delta}")
-            else:
-                continue
+        # This first example shows a generated text response for the provided prompt
+        if message["text"] != "Wonder a few deep thoughts.":
+            set_status(
+                status="thinking...",
+                loading_messages=[
+                    "Teaching the hamsters to type faster…",
+                    "Untangling the internet cables…",
+                    "Consulting the office goldfish…",
+                    "Polishing up the response just for you…",
+                    "Convincing the AI to stop overthinking…",
+                ],
+            )
+
+            # Loop over OpenAI response stream
+            # https://platform.openai.com/docs/api-reference/responses/create
+            for event in call_llm(message["text"]):
+                if event.type == "response.output_text.delta":
+                    streamer.append(markdown_text=f"{event.delta}")
+                else:
+                    continue
+
+            feedback_block = create_feedback_block()
+            streamer.stop(
+                blocks=feedback_block,
+            )
+
+        # The second example shows detailed thinking steps similar to tool calls
+        else:
+            streamer.append(
+                chunks=[
+                    MarkdownTextChunk(
+                        text="Hello.\nI have received the task. ",
+                    ),
+                    MarkdownTextChunk(
+                        text="This task appears manageable.\nThat is good.",
+                    ),
+                    TaskUpdateChunk(
+                        id="001",
+                        title="Understanding the task...",
+                        status="in_progress",
+                        details="- Indentify the goal\n- Identify constraints\n- Pretending this is obvious",
+                    ),
+                    TaskUpdateChunk(
+                        id="002",
+                        title="Performing acrobatics...",
+                        status="pending",
+                    ),
+                ],
+            )
+            time.sleep(4)
+
+            streamer.append(
+                chunks=[
+                    TaskUpdateChunk(
+                        id="001",
+                        title="Understanding the task...",
+                        status="complete",
+                        details="- Indentied the goal\n- Identied constraints\n- Pretended this was obvious",
+                        output="We'll continue to ramble now",
+                    ),
+                    TaskUpdateChunk(
+                        id="002",
+                        title="Performing acrobatics...",
+                        status="in_progress",
+                        details="- Jumping atop ropes\n- Juggling bowling pins\n- Riding a single wheel too",
+                    ),
+                ],
+            )
+            time.sleep(4)
 
-        feedback_block = create_feedback_block()
-        streamer.stop(blocks=feedback_block)
+            streamer.stop(
+                chunks=[
+                    TaskUpdateChunk(
+                        id="002",
+                        title="Performing acrobatics...",
+                        status="complete",
+                        details="- Jumped atop ropes\n- Juggled bowling pins\n- Rode a single wheel too",
+                    ),
+                    MarkdownTextChunk(
+                        text="The crowd appears to be astouned and applauds :popcorn:"
+                    ),
+                ],
+            )
 
     except Exception as e:
         logger.exception(f"Failed to handle a user message event: {e}")

listeners/events/app_mentioned.py

@@ -4,6 +4,7 @@
 from slack_sdk import WebClient
 
 from ai.llm_caller import call_llm
+
 from ..views.feedback_block import create_feedback_block
 
 
@@ -38,7 +39,7 @@ def app_mentioned_callback(client: WebClient, event: dict, logger: Logger, say:
             ],
         )
 
-        returned_message = call_llm([{"role": "user", "content": text}])
+        returned_message = call_llm(text)
 
         streamer = client.chat_stream(
             channel=channel_id,