diff --git a/pyproject.toml b/pyproject.toml
index d7d7c08..5ea8411 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -47,6 +47,7 @@ simsimd = "^3.7.4"
 youtube-transcript-api = "^0.6.2"
 pytube = "^15.0.0"
 pydub = "^0.25.1"
+langchain-google-genai = "^0.0.9"
 
 [tool.poetry.dev-dependencies]
 #fasttext = {git = "https://github.com/cfculhane/fastText"} # FastText doesn't come with pybind11 and we need to use this workaround.
diff --git a/src/wandbot/adcopy/adcopy.py b/src/wandbot/adcopy/adcopy.py
index 3bc1607..a9c0a7f 100644
--- a/src/wandbot/adcopy/adcopy.py
+++ b/src/wandbot/adcopy/adcopy.py
@@ -1,6 +1,6 @@
 import json
-import random
 import logging
+import random
 from operator import itemgetter
 from typing import Any, Dict, List
 
@@ -8,6 +8,7 @@
 from langchain_core.prompts import ChatPromptTemplate
 from langchain_core.runnables import Runnable, RunnableParallel
 from langchain_openai import ChatOpenAI
+
 from wandbot.chat.chat import Chat
 from wandbot.chat.schemas import ChatRequest
 from wandbot.rag.utils import ChatModel
@@ -79,7 +80,9 @@ def build_prompt_input_variables(
         self, query: str, persona: str, action: str
     ) -> Dict[str, Any]:
         wandbot_response = self.query_wandbot(query)
-        additional_context = "\n".join(random.choices(self.contexts[action], k=2))
+        additional_context = "\n".join(
+            random.choices(self.contexts[action], k=2)
+        )
         persona_prompt = (
             TECHNICAL_PROMPT if persona == "technical" else EXECUTIVE_PROMPT
         )
@@ -122,7 +125,9 @@ def _load_chain(self, model: ChatOpenAI) -> Runnable:
         return chain
 
     def __call__(self, query: str, persona: str, action: str) -> str:
-        logging.info(f"Generating ad copy for {persona} {action} with query: '{query}'")
+        logging.info(
+            f"Generating ad copy for {persona} {action} with query: '{query}'"
+        )
         inputs = self.build_inputs_for_ad_formats(query, persona, action)
         outputs = self.chain.batch(inputs)
         str_output = ""
diff --git a/src/wandbot/api/app.py b/src/wandbot/api/app.py
index ec2ad45..d162fa7 100644
--- a/src/wandbot/api/app.py
+++ b/src/wandbot/api/app.py
@@ -33,9 +33,9 @@
 from datetime import datetime, timezone
 
 import pandas as pd
-import wandb
 from fastapi import FastAPI
 
+import wandb
 from wandbot.api.routers import adcopy as adcopy_router
 from wandbot.api.routers import chat as chat_router
 from wandbot.api.routers import content_navigator as content_navigator_router
diff --git a/src/wandbot/api/client.py b/src/wandbot/api/client.py
index 23fa1e9..d60715f 100644
--- a/src/wandbot/api/client.py
+++ b/src/wandbot/api/client.py
@@ -15,12 +15,13 @@
 
 import aiohttp
 import requests
+
+from wandbot.api.routers.adcopy import AdCopyRequest, AdCopyResponse
+from wandbot.api.routers.chat import APIQueryRequest, APIQueryResponse
 from wandbot.api.routers.content_navigator import (
     ContentNavigatorRequest,
     ContentNavigatorResponse,
 )
-from wandbot.api.routers.adcopy import AdCopyRequest, AdCopyResponse
-from wandbot.api.routers.chat import APIQueryRequest, APIQueryResponse
 from wandbot.api.routers.database import (
     APIFeedbackRequest,
     APIFeedbackResponse,
@@ -66,7 +67,9 @@ def __init__(self, url: str):
         )
         self.retrieve_endpoint = urljoin(str(self.url), "retrieve")
         self.generate_ads_endpoint = urljoin(str(self.url), "generate_ads")
-        self.generate_content_suggestions_endpoint = urljoin(str(self.url), "generate_content_suggestions")
+        self.generate_content_suggestions_endpoint = urljoin(
+            str(self.url), "generate_content_suggestions"
+        )
 
     def _get_chat_thread(
         self, request: APIGetChatThreadRequest
@@ -640,9 +643,9 @@ async def generate_ads(
         response = await self._generate_ads(request)
 
         return response
-    
+
     async def _generate_content_suggestions(
-            self, request: ContentNavigatorRequest
+        self, request: ContentNavigatorRequest
     ) -> ContentNavigatorResponse | None:
         """Call the content navigator API.
 
@@ -664,7 +667,7 @@ async def _generate_content_suggestions(
                     return None
 
     async def generate_content_suggestions(
-            self, user_id: str, query: str
+        self, user_id: str, query: str
     ) -> ContentNavigatorResponse:
         """Generates content suggestions given query.
 
@@ -683,4 +686,3 @@ async def generate_content_suggestions(
         response = await self._generate_content_suggestions(request)
 
         return response
-            
diff --git a/src/wandbot/api/routers/adcopy.py b/src/wandbot/api/routers/adcopy.py
index 34cdcef..360edf2 100644
--- a/src/wandbot/api/routers/adcopy.py
+++ b/src/wandbot/api/routers/adcopy.py
@@ -3,6 +3,7 @@
 from fastapi import APIRouter
 from pydantic import BaseModel
 from starlette import status
+
 from wandbot.adcopy.adcopy import AdCopyEngine
 
 
diff --git a/src/wandbot/api/routers/chat.py b/src/wandbot/api/routers/chat.py
index 735188d..310ba43 100644
--- a/src/wandbot/api/routers/chat.py
+++ b/src/wandbot/api/routers/chat.py
@@ -1,3 +1,5 @@
+from typing import Optional
+
 from fastapi import APIRouter
 from starlette import status
 
@@ -38,12 +40,14 @@ def query(
     Returns:
         The APIQueryResponse object containing the result of the query.
     """
+    logger.info(request.images[0][:10])
     result = chat(
         ChatRequest(
             question=request.question,
             chat_history=request.chat_history,
             language=request.language,
             application=request.application,
+            images=request.images,
         ),
     )
     result = APIQueryResponse(**result.model_dump())
diff --git a/src/wandbot/api/routers/content_navigator.py b/src/wandbot/api/routers/content_navigator.py
index 590ef43..22df118 100644
--- a/src/wandbot/api/routers/content_navigator.py
+++ b/src/wandbot/api/routers/content_navigator.py
@@ -1,10 +1,12 @@
-import  httpx
-
+import httpx
 from fastapi import APIRouter
 from pydantic import BaseModel
 from starlette import status
 
-CONTENT_NAVIGATOR_ENDPOINT = "https://wandb-content-navigator.replit.app/get_content"
+CONTENT_NAVIGATOR_ENDPOINT = (
+    "https://wandb-content-navigator.replit.app/get_content"
+)
+
 
 class ContentNavigatorRequest(BaseModel):
     """A user query to be used by the content navigator app"""
@@ -12,6 +14,7 @@ class ContentNavigatorRequest(BaseModel):
     user_id: str = None
     query: str
 
+
 class ContentNavigatorResponse(BaseModel):
     """Response from the content navigator app"""
 
@@ -26,7 +29,9 @@ class ContentNavigatorResponse(BaseModel):
 )
 
 
-@router.post("/", response_model=ContentNavigatorResponse, status_code=status.HTTP_200_OK)
+@router.post(
+    "/", response_model=ContentNavigatorResponse, status_code=status.HTTP_200_OK
+)
 async def generate_content_suggestions(request: ContentNavigatorRequest):
     async with httpx.AsyncClient(timeout=1200.0) as content_client:
         response = await content_client.post(
@@ -34,7 +39,7 @@ async def generate_content_suggestions(request: ContentNavigatorRequest):
             json={"query": request.query, "user_id": request.user_id},
         )
         response_data = response.json()
-    
+
     slack_response = response_data.get("slack_response", "")
     rejected_slack_response = response_data.get("rejected_slack_response", "")
     response_items_count = response_data.get("response_items_count", 0)
@@ -48,4 +53,3 @@ async def generate_content_suggestions(request: ContentNavigatorRequest):
         rejected_slack_response=rejected_slack_response,
         response_items_count=response_items_count,
     )
-
diff --git a/src/wandbot/api/routers/database.py b/src/wandbot/api/routers/database.py
index 7174af6..4a8f840 100644
--- a/src/wandbot/api/routers/database.py
+++ b/src/wandbot/api/routers/database.py
@@ -1,8 +1,8 @@
-import wandb
 from fastapi import APIRouter
 from starlette import status
 from starlette.responses import Response
 
+import wandb
 from wandbot.database.client import DatabaseClient
 from wandbot.database.database import engine
 from wandbot.database.models import Base
@@ -13,8 +13,8 @@
     FeedbackCreate,
     QuestionAnswer,
     QuestionAnswerCreate,
-    YoutubeAssistantThreadCreate,
     YoutubeAssistantThread,
+    YoutubeAssistantThreadCreate,
 )
 from wandbot.utils import get_logger
 
diff --git a/src/wandbot/apps/slack/__main__.py b/src/wandbot/apps/slack/__main__.py
index 6c2955e..5d066f2 100644
--- a/src/wandbot/apps/slack/__main__.py
+++ b/src/wandbot/apps/slack/__main__.py
@@ -15,6 +15,7 @@
 
 from slack_bolt.adapter.socket_mode.async_handler import AsyncSocketModeHandler
 from slack_bolt.async_app import AsyncApp
+
 from wandbot.api.client import AsyncAPIClient
 from wandbot.apps.slack.config import SlackAppEnConfig, SlackAppJaConfig
 from wandbot.apps.slack.handlers.ad_copy import (
@@ -62,6 +63,7 @@
 api_client = AsyncAPIClient(url=config.WANDBOT_API_URL)
 slack_client = app.client
 
+
 def get_init_block(user: str) -> List[Dict[str, Any]]:
     initial_block = [
         {
@@ -133,6 +135,7 @@ def get_init_block(user: str) -> List[Dict[str, Any]]:
     ]
     return initial_block
 
+
 # --------------------------------------
 # Main Wandbot Mention Handler
 # --------------------------------------
@@ -212,10 +215,11 @@ async def handle_mention(event: dict, say: callable) -> None:
     )
 )
 
+
 async def main():
     handler = AsyncSocketModeHandler(app, config.SLACK_APP_TOKEN)
     await handler.start_async()
 
 
 if __name__ == "__main__":
-    asyncio.run(main())
\ No newline at end of file
+    asyncio.run(main())
diff --git a/src/wandbot/apps/slack/config.py b/src/wandbot/apps/slack/config.py
index 5b903df..92692df 100644
--- a/src/wandbot/apps/slack/config.py
+++ b/src/wandbot/apps/slack/config.py
@@ -48,12 +48,12 @@
     "#support チャンネルにいるwandbチームに質問してください。この答えは役に立ったでしょうか？下のボタンでお知らせ下さい。"
 )
 
-JA_ERROR_MESSAGE = "「おっと、問題が発生しました。しばらくしてからもう一度お試しください。」"
-
-JA_FALLBACK_WARNING_MESSAGE = (
-    "**警告: {model}** にフォールバックします。これらの結果は **gpt-4** ほど良くない可能性があります\n\n"
+JA_ERROR_MESSAGE = (
+    "「おっと、問題が発生しました。しばらくしてからもう一度お試しください。」"
 )
 
+JA_FALLBACK_WARNING_MESSAGE = "**警告: {model}** にフォールバックします。これらの結果は **gpt-4** ほど良くない可能性があります\n\n"
+
 
 class SlackAppEnConfig(BaseSettings):
     APPLICATION: str = Field("Slack_EN")
diff --git a/src/wandbot/apps/slack/handlers/ad_copy.py b/src/wandbot/apps/slack/handlers/ad_copy.py
index 111c211..903b4f1 100644
--- a/src/wandbot/apps/slack/handlers/ad_copy.py
+++ b/src/wandbot/apps/slack/handlers/ad_copy.py
@@ -1,8 +1,9 @@
-import re
 import logging
+import re
 from typing import Any, Dict, List
 
 from slack_sdk.web.async_client import AsyncWebClient
+
 from wandbot.api.client import AsyncAPIClient
 from wandbot.apps.slack.utils import get_initial_message_from_thread
 
@@ -131,8 +132,11 @@ async def handle_adcopy_action(
     query = re.sub(r"\<@\w+\>", "", query).strip()
     logger.info(f"Initial message: {initial_message}")
 
-    await say(f"Working on generating ads for '{persona}' focussed on '{action}' \
-for the query: '{query}'...", thread_ts=thread_ts)
+    await say(
+        f"Working on generating ads for '{persona}' focussed on '{action}' \
+for the query: '{query}'...",
+        thread_ts=thread_ts,
+    )
 
     api_response = await api_client.generate_ads(
         query=query, action=action, persona=persona
diff --git a/src/wandbot/apps/slack/handlers/content_navigator.py b/src/wandbot/apps/slack/handlers/content_navigator.py
index 482e234..bef0560 100644
--- a/src/wandbot/apps/slack/handlers/content_navigator.py
+++ b/src/wandbot/apps/slack/handlers/content_navigator.py
@@ -1,6 +1,7 @@
 import logging
 
 from slack_sdk.web.async_client import AsyncWebClient
+
 from wandbot.api.client import AsyncAPIClient
 from wandbot.apps.slack.utils import get_initial_message_from_thread
 
@@ -32,15 +33,20 @@ async def handle_content_navigator_action(
     if api_response.response_items_count > 0:
         await say(api_response.slack_response, thread_ts=thread_ts)
     else:
-        await say("No content suggestions found. Try rephrasing your query, but note \
+        await say(
+            "No content suggestions found. Try rephrasing your query, but note \
 there may also not be any relevant pieces of content for this query. Add '--debug' to \
 your query and try again to see a detailed resoning for each suggestion.",
-                thread_ts=thread_ts)
-    
+            thread_ts=thread_ts,
+        )
+
     # if debug mode is enabled, send the rejected suggestions as well
     if len(api_response.rejected_slack_response) > 1:
-        await say("REJECTED SUGGESTIONS:\n{api_response.rejected_slack_response}", thread_ts=thread_ts)
-        
+        await say(
+            "REJECTED SUGGESTIONS:\n{api_response.rejected_slack_response}",
+            thread_ts=thread_ts,
+        )
+
 
 def create_content_navigator_handler(
     slack_client: AsyncWebClient, api_client: AsyncAPIClient
@@ -57,4 +63,4 @@ async def executive_signups_handler(
             logger=logger,
         )
 
-    return executive_signups_handler
\ No newline at end of file
+    return executive_signups_handler
diff --git a/src/wandbot/apps/slack/handlers/docsbot.py b/src/wandbot/apps/slack/handlers/docsbot.py
index 3626ad8..7901a46 100644
--- a/src/wandbot/apps/slack/handlers/docsbot.py
+++ b/src/wandbot/apps/slack/handlers/docsbot.py
@@ -2,6 +2,7 @@
 
 from slack_sdk.web import SlackResponse
 from slack_sdk.web.async_client import AsyncWebClient
+
 from wandbot.api.client import AsyncAPIClient
 from wandbot.apps.slack.config import SlackAppEnConfig, SlackAppJaConfig
 from wandbot.apps.slack.formatter import MrkdwnFormatter
diff --git a/src/wandbot/apps/slack/handlers/youtube_chat.py b/src/wandbot/apps/slack/handlers/youtube_chat.py
index 0e6b8ed..d47bc88 100644
--- a/src/wandbot/apps/slack/handlers/youtube_chat.py
+++ b/src/wandbot/apps/slack/handlers/youtube_chat.py
@@ -3,6 +3,7 @@
 
 from pytube.exceptions import RegexMatchError
 from slack_sdk.web.async_client import AsyncWebClient
+
 from wandbot.apps.slack.utils import get_initial_message_from_thread
 from wandbot.youtube_chat.video_utils import YoutubeVideoInfo
 
@@ -201,7 +202,10 @@ async def handle_youtube_chat_input(
         await ack()
         logger.info(f"Received message: {body}")
         url = body["actions"][0]["value"]
-        await say("Working on in it...", thread_ts=body["message"]["thread_ts"],)
+        await say(
+            "Working on in it...",
+            thread_ts=body["message"]["thread_ts"],
+        )
         video_confirmation_block = get_video_confirmation_blocks(url)
         await say(
             blocks=video_confirmation_block,
diff --git a/src/wandbot/chat/chat.py b/src/wandbot/chat/chat.py
index ff0f0d7..a4f72c1 100644
--- a/src/wandbot/chat/chat.py
+++ b/src/wandbot/chat/chat.py
@@ -24,7 +24,10 @@
           print(f"WandBot: {response.answer}")
           print(f"Time taken: {response.time_taken}")
 """
-from typing import List
+
+from typing import List, Optional
+
+from weave.monitoring import StreamTable
 
 import wandb
 from wandbot.chat.config import ChatConfig
@@ -33,7 +36,6 @@
 from wandbot.database.schemas import QuestionAnswer
 from wandbot.retriever import VectorStore
 from wandbot.utils import Timer, get_logger
-from weave.monitoring import StreamTable
 
 logger = get_logger(__name__)
 
@@ -65,23 +67,27 @@ def __init__(
             job_type="chat",
         )
         self.run._label(repo="wandbot")
-        self.chat_table = StreamTable(
-            table_name="chat_logs",
-            project_name=self.config.wandb_project,
-            entity_name=self.config.wandb_entity,
-        )
+        # self.chat_table = StreamTable(
+        #     table_name="chat_logs",
+        #     project_name=self.config.wandb_project,
+        #     entity_name=self.config.wandb_entity,
+        # )
 
         self.rag_pipeline = RAGPipeline(vector_store=vector_store)
 
     def _get_answer(
-        self, question: str, chat_history: List[QuestionAnswer]
+        self,
+        question: str,
+        chat_history: List[QuestionAnswer],
+        images: Optional[List[str]] = None,
     ) -> RAGPipelineOutput:
         history = []
         for item in chat_history:
             history.append(("user", item.question))
             history.append(("assistant", item.answer))
+            # TODO: Add image prompts to history
 
-        result = self.rag_pipeline(question, history)
+        result = self.rag_pipeline(question, history, images=images)
 
         return result
 
@@ -95,8 +101,11 @@ def __call__(self, chat_request: ChatRequest) -> ChatResponse:
             An instance of `ChatResponse` representing the chat response.
         """
         try:
+            logger.info(chat_request.images[0][:10])
             result = self._get_answer(
-                chat_request.question, chat_request.chat_history or []
+                chat_request.question,
+                chat_request.chat_history or [],
+                images=chat_request.images,
             )
 
             result_dict = result.model_dump()
@@ -108,7 +117,7 @@ def __call__(self, chat_request: ChatRequest) -> ChatResponse:
             }
             result_dict.update({"application": chat_request.application})
             self.run.log(usage_stats)
-            self.chat_table.log(result_dict)
+            # self.chat_table.log(result_dict)
             return ChatResponse(**result_dict)
         except Exception as e:
             with Timer() as timer:
diff --git a/src/wandbot/chat/rag.py b/src/wandbot/chat/rag.py
index 10590ba..02f471a 100644
--- a/src/wandbot/chat/rag.py
+++ b/src/wandbot/chat/rag.py
@@ -1,8 +1,13 @@
 import datetime
-from typing import List, Tuple
+from typing import List, Optional, Tuple
 
+from langchain_anthropic import ChatAnthropic
 from langchain_community.callbacks import get_openai_callback
+from langchain_core.messages import HumanMessage, SystemMessage
+from langchain_core.prompts import ChatPromptTemplate
+from langchain_openai import ChatOpenAI
 from pydantic import BaseModel
+
 from wandbot.rag.query_handler import QueryEnhancer
 from wandbot.rag.response_synthesis import ResponseSynthesizer
 from wandbot.rag.retrieval import FusionRetrieval
@@ -58,15 +63,55 @@ def __init__(
         )
         self.response_synthesizer = ResponseSynthesizer()
 
+    def generate_multi_modal_initial_response(
+        self, question: str, images: List[str]
+    ) -> str:
+        # model = ChatOpenAI(model="gpt-4-vision-preview", max_tokens=500)
+        model = ChatAnthropic(model="claude-3-opus-20240229")
+        system_message = SystemMessage(
+            content="""You are a Weights & Biases support expert.
+            Your goal is to describe the attached screenshots in the context of the user query. You are provided with a support ticket and screenshots related to the issue.
+            Provide a detailed description of the image in the context of the query so that the ticket can be answered correctly while incorporating the image info."""
+        )
+        prompt = [
+            {
+                "type": "text",
+                "text": question,
+            }
+        ]
+        for img in images:
+            prompt += [
+                {
+                    "type": "image_url",
+                    "image_url": {"url": f"data:image/png;base64,{img}"},
+                }
+            ]
+        message = HumanMessage(content=prompt)
+        response = model.invoke([system_message, message])
+        return response.content
+
     def __call__(
-        self, question: str, chat_history: List[Tuple[str, str]] | None = None
+        self,
+        question: str,
+        chat_history: List[Tuple[str, str]] | None = None,
+        images: Optional[List[str]] = None,
     ):
         if chat_history is None:
             chat_history = []
 
+        multi_modal_response = (
+            self.generate_multi_modal_initial_response(question, images)
+            if images is not None
+            else ""
+        )
+
         with get_openai_callback() as query_enhancer_cb, Timer() as query_enhancer_tb:
             enhanced_query = self.query_enhancer.chain.invoke(
-                {"query": question, "chat_history": chat_history}
+                {
+                    "query": question,
+                    "chat_history": chat_history,
+                    "image_context": multi_modal_response,
+                }
             )
 
         with Timer() as retrieval_tb:
diff --git a/src/wandbot/chat/schemas.py b/src/wandbot/chat/schemas.py
index f952ded..bce7b6f 100644
--- a/src/wandbot/chat/schemas.py
+++ b/src/wandbot/chat/schemas.py
@@ -17,7 +17,7 @@
 """
 
 from datetime import datetime
-from typing import List
+from typing import List, Optional
 
 from pydantic import BaseModel
 
@@ -46,6 +46,7 @@ class ChatRequest(BaseModel):
     chat_history: List[QuestionAnswer] | None = None
     application: str | None = None
     language: str = "en"
+    images: Optional[List[str]] = None
 
 
 class ChatResponse(BaseModel):
diff --git a/src/wandbot/database/client.py b/src/wandbot/database/client.py
index 06813ac..0e527bb 100644
--- a/src/wandbot/database/client.py
+++ b/src/wandbot/database/client.py
@@ -15,7 +15,7 @@
 
 import json
 from datetime import datetime, timedelta
-from typing import Any, List, Collection
+from typing import Any, Collection, List
 
 from sqlalchemy.future import create_engine
 from sqlalchemy.orm import sessionmaker
diff --git a/src/wandbot/evaluation/eval/__main__.py b/src/wandbot/evaluation/eval/__main__.py
index 3302f59..78de74c 100644
--- a/src/wandbot/evaluation/eval/__main__.py
+++ b/src/wandbot/evaluation/eval/__main__.py
@@ -95,9 +95,9 @@ def get_answer_correctness(row_str: str) -> str:
         reference_notes=row["reference_notes"],
     )
     result = parse_answer_eval("answer_correctness", result.dict())
-    result[
-        "answer_correctness_score_(ragas)"
-    ] = metrics.answer_correctness.score_single(row)
+    result["answer_correctness_score_(ragas)"] = (
+        metrics.answer_correctness.score_single(row)
+    )
     result = json.dumps(result)
     return result
 
@@ -113,9 +113,9 @@ def get_answer_relevancy(row_str: str) -> str:
         reference=row["ground_truths"],
     )
     result = parse_answer_eval("answer_relevancy", result.dict())
-    result[
-        "answer_relevancy_score_(ragas)"
-    ] = metrics.answer_relevancy.score_single(row)
+    result["answer_relevancy_score_(ragas)"] = (
+        metrics.answer_relevancy.score_single(row)
+    )
     result = json.dumps(result)
     return result
 
@@ -132,9 +132,9 @@ def get_answer_faithfulness(row_str: str) -> str:
     )
 
     result = parse_answer_eval("answer_faithfulness", result.dict())
-    result[
-        "answer_faithfulness_score_(ragas)"
-    ] = metrics.faithfulness.score_single(row)
+    result["answer_faithfulness_score_(ragas)"] = (
+        metrics.faithfulness.score_single(row)
+    )
     result = json.dumps(result)
     return result
 
diff --git a/src/wandbot/multi_modal/__init__.py b/src/wandbot/multi_modal/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/src/wandbot/multi_modal/multi_modal.py b/src/wandbot/multi_modal/multi_modal.py
new file mode 100644
index 0000000..200188f
--- /dev/null
+++ b/src/wandbot/multi_modal/multi_modal.py
@@ -0,0 +1,14 @@
+import base64
+
+from langchain.schema.messages import AIMessage, HumanMessage
+from langchain_openai import ChatOpenAI
+
+
+class MultiModalQueryEngine:
+
+    def __init__(self, model: str = "gpt-4-vision-preview"):
+        self.model = ChatOpenAI(model_name=model)
+
+    def encode_image(image_path):
+        with open(image_path, "rb") as image_file:
+            return base64.b64encode(image_file.read()).decode("utf-8")
diff --git a/src/wandbot/rag/query_handler.py b/src/wandbot/rag/query_handler.py
index e311fc7..851f5a4 100644
--- a/src/wandbot/rag/query_handler.py
+++ b/src/wandbot/rag/query_handler.py
@@ -192,7 +192,10 @@ def avoid_query(self) -> bool:
         )
 
     def parse_output(
-        self, query: str, chat_history: Optional[List[Tuple[str, str]]] = None
+        self,
+        query: str,
+        chat_history: Optional[List[Tuple[str, str]]] = None,
+        image_context: Optional[str] = None,
     ) -> Dict[str, Any]:
         """Parse the output of the model"""
         question = clean_question(query)
@@ -237,6 +240,7 @@ def parse_output(
             "avoid_query": self.avoid_query,
             "chat_history": chat_history,
             "all_queries": all_queries,
+            "image_context": image_context,
         }
 
 
@@ -299,10 +303,13 @@ def _load_chain(self, model: ChatOpenAI) -> Runnable:
             query=itemgetter("query"),
             chat_history=itemgetter("chat_history"),
             enhanced_query=full_query_enhancer_chain,
+            image_context=itemgetter("image_context"),
         )
         chain = intermediate_chain | RunnableLambda(
             lambda x: x["enhanced_query"].parse_output(
-                x["query"], convert_to_messages(x["chat_history"])
+                x["query"],
+                convert_to_messages(x["chat_history"]),
+                image_context=x["image_context"],
             )
         )
 
diff --git a/src/wandbot/rag/retrieval.py b/src/wandbot/rag/retrieval.py
index 065c791..1a72ac9 100644
--- a/src/wandbot/rag/retrieval.py
+++ b/src/wandbot/rag/retrieval.py
@@ -3,6 +3,7 @@
 from langchain.retrievers.document_compressors import CohereRerank
 from langchain_core.documents import Document
 from langchain_core.runnables import Runnable, RunnablePassthrough
+
 from wandbot.rag.utils import get_web_contexts
 from wandbot.retriever.base import VectorStore
 from wandbot.retriever.web_search import YouSearch, YouSearchConfig
diff --git a/src/wandbot/rag/utils.py b/src/wandbot/rag/utils.py
index f2270a4..264de4c 100644
--- a/src/wandbot/rag/utils.py
+++ b/src/wandbot/rag/utils.py
@@ -47,6 +47,10 @@ def __set__(self, obj, value):
 Sub-queries to consider answering: 
 
 {sub_queries}
+
+Context from attached images:
+
+{image_context}
 """
 )
 
@@ -58,6 +62,7 @@ def create_query_str(enhanced_query, document_prompt=DEFAULT_QUESTION_PROMPT):
         "intents": enhanced_query["intents"],
         "sub_queries": "\t"
         + "\n\t".join(enhanced_query["sub_queries"]).strip(),
+        "image_context": enhanced_query["image_context"],
     }
     doc = Document(page_content=page_content, metadata=metadata)
     doc = clean_document_content(doc)
diff --git a/src/wandbot/retriever/base.py b/src/wandbot/retriever/base.py
index 3982ca6..8c08a47 100644
--- a/src/wandbot/retriever/base.py
+++ b/src/wandbot/retriever/base.py
@@ -1,18 +1,17 @@
 from operator import itemgetter
 from typing import List
 
-import wandb
+from chromadb.config import Settings as ChromaSettings
 from langchain_community.document_transformers import EmbeddingsRedundantFilter
 from langchain_community.vectorstores.chroma import Chroma
 from langchain_core.documents import Document
 from langchain_core.runnables import RunnableLambda, RunnableParallel
+
+import wandb
 from wandbot.ingestion.config import VectorStoreConfig
 from wandbot.retriever.reranking import CohereRerankChain
 from wandbot.retriever.utils import OpenAIEmbeddingsModel
 
-from chromadb.config import Settings as ChromaSettings
-
-
 
 class VectorStore:
     embeddings_model: OpenAIEmbeddingsModel = OpenAIEmbeddingsModel(
@@ -34,7 +33,7 @@ def __init__(
             collection_name=collection_name,
             embedding_function=self.embeddings_model,  # type: ignore
             persist_directory=persist_dir,
-            client_settings=ChromaSettings(anonymized_telemetry=False)
+            client_settings=ChromaSettings(anonymized_telemetry=False),
         )
 
     @classmethod
diff --git a/src/wandbot/youtube_chat/chat_utils.py b/src/wandbot/youtube_chat/chat_utils.py
index f9ce1b4..b546f6c 100644
--- a/src/wandbot/youtube_chat/chat_utils.py
+++ b/src/wandbot/youtube_chat/chat_utils.py
@@ -9,14 +9,12 @@
 from tenacity import (
     retry,
     retry_if_result,
-    wait_exponential,
     stop_after_attempt,
+    wait_exponential,
 )
 
 from wandbot.database.client import DatabaseClient
-from wandbot.database.schemas import (
-    YoutubeAssistantThreadCreate,
-)
+from wandbot.database.schemas import YoutubeAssistantThreadCreate
 from wandbot.utils import get_logger
 
 logger = get_logger(__name__)