diff --git a/app_home.py b/app_home.py
index 4badc36..a8df6c9 100644
--- a/app_home.py
+++ b/app_home.py
@@ -1,3 +1,35 @@
 import streamlit as st
 
-st.switch_page("pages/2_📒_Notebooks.py")
+from open_notebook.domain.base import ObjectModel
+from open_notebook.exceptions import NotFoundError
+from pages.components import (
+    note_panel,
+    source_embedding_panel,
+    source_insight_panel,
+    source_panel,
+)
+from pages.stream_app.utils import setup_page
+
+setup_page("📒 Open Notebook", sidebar_state="collapsed")
+
+if "object_id" not in st.query_params:
+    st.switch_page("pages/2_📒_Notebooks.py")
+    st.stop()
+
+object_id = st.query_params["object_id"]
+try:
+    obj = ObjectModel.get(object_id)
+except NotFoundError:
+    st.switch_page("pages/2_📒_Notebooks.py")
+    st.stop()
+
+obj_type = object_id.split(":")[0]
+
+if obj_type == "note":
+    note_panel(object_id)
+elif obj_type == "source":
+    source_panel(object_id)
+elif obj_type == "source_insight":
+    source_insight_panel(object_id)
+elif obj_type == "source_embedding":
+    source_embedding_panel(object_id)
diff --git a/open_notebook/domain/base.py b/open_notebook/domain/base.py
index 146589d..8514e89 100644
--- a/open_notebook/domain/base.py
+++ b/open_notebook/domain/base.py
@@ -1,5 +1,5 @@
 from datetime import datetime
-from typing import Any, ClassVar, Dict, List, Optional, Type, TypeVar
+from typing import Any, ClassVar, Dict, List, Optional, Type, TypeVar, cast
 
 from loguru import logger
 from pydantic import BaseModel, ValidationError, field_validator
@@ -29,15 +29,26 @@ class ObjectModel(BaseModel):
     @classmethod
     def get_all(cls: Type[T], order_by=None) -> List[T]:
         try:
+            # If called from a specific subclass, use its table_name
+            if cls.table_name:
+                target_class = cls
+                table_name = cls.table_name
+            else:
+                # This path is taken if called directly from ObjectModel
+                raise InvalidInputError(
+                    "get_all() must be called from a specific model class"
+                )
+
             if order_by:
                 order = f" ORDER BY {order_by}"
             else:
                 order = ""
-            result = repo_query(f"SELECT * FROM {cls.table_name} {order}")
+
+            result = repo_query(f"SELECT * FROM {table_name} {order}")
             objects = []
             for obj in result:
                 try:
-                    objects.append(cls(**obj))
+                    objects.append(target_class(**obj))
                 except Exception as e:
                     logger.critical(f"Error creating object: {str(e)}")
 
@@ -52,15 +63,44 @@ class ObjectModel(BaseModel):
         if not id:
             raise InvalidInputError("ID cannot be empty")
         try:
+            # Get the table name from the ID (everything before the first colon)
+            table_name = id.split(":")[0] if ":" in id else id
+
+            # If we're calling from a specific subclass and IDs match, use that class
+            if cls.table_name and cls.table_name == table_name:
+                target_class: Type[T] = cls
+            else:
+                # Otherwise, find the appropriate subclass based on table_name
+                found_class = cls._get_class_by_table_name(table_name)
+                if not found_class:
+                    raise InvalidInputError(f"No class found for table {table_name}")
+                target_class = cast(Type[T], found_class)
+
             result = repo_query(f"SELECT * FROM {id}")
             if result:
-                return cls(**result[0])
+                return target_class(**result[0])
             else:
-                raise NotFoundError(f"{cls.table_name} with id {id} not found")
+                raise NotFoundError(f"{table_name} with id {id} not found")
         except Exception as e:
-            logger.error(f"Error fetching {cls.table_name} with id {id}: {str(e)}")
+            logger.error(f"Error fetching object with id {id}: {str(e)}")
             logger.exception(e)
-            raise NotFoundError(f"{cls.table_name} with id {id} not found")
+            raise NotFoundError(f"Object with id {id} not found - {str(e)}")
+
+    @classmethod
+    def _get_class_by_table_name(cls, table_name: str) -> Optional[Type["ObjectModel"]]:
+        """Find the appropriate subclass based on table_name."""
+
+        def get_all_subclasses(c: Type["ObjectModel"]) -> List[Type["ObjectModel"]]:
+            all_subclasses: List[Type["ObjectModel"]] = []
+            for subclass in c.__subclasses__():
+                all_subclasses.append(subclass)
+                all_subclasses.extend(get_all_subclasses(subclass))
+            return all_subclasses
+
+        for subclass in get_all_subclasses(ObjectModel):
+            if hasattr(subclass, "table_name") and subclass.table_name == table_name:
+                return subclass
+        return None
 
     def needs_embedding(self) -> bool:
         return False
diff --git a/open_notebook/domain/notebook.py b/open_notebook/domain/notebook.py
index 86944f8..e473193 100644
--- a/open_notebook/domain/notebook.py
+++ b/open_notebook/domain/notebook.py
@@ -93,10 +93,42 @@ class Asset(BaseModel):
     url: Optional[str] = None
 
 
+class SourceEmbedding(ObjectModel):
+    table_name: ClassVar[str] = "source_embedding"
+    content: str
+
+    @property
+    def source(self) -> "Source":
+        try:
+            src = repo_query(f"""
+            select source.* from {self.id}                    fetch source
+
+            """)
+            return Source(**src[0]["source"])
+        except Exception as e:
+            logger.error(f"Error fetching source for embedding {self.id}: {str(e)}")
+            logger.exception(e)
+            raise DatabaseOperationError(e)
+
+
 class SourceInsight(ObjectModel):
+    table_name: ClassVar[str] = "source_insight"
     insight_type: str
     content: str
 
+    @property
+    def source(self) -> "Source":
+        try:
+            src = repo_query(f"""
+            select source.* from {self.id}                    fetch source
+
+            """)
+            return Source(**src[0]["source"])
+        except Exception as e:
+            logger.error(f"Error fetching source for insight {self.id}: {str(e)}")
+            logger.exception(e)
+            raise DatabaseOperationError(e)
+
 
 class Source(ObjectModel):
     table_name: ClassVar[str] = "source"
@@ -112,7 +144,7 @@ class Source(ObjectModel):
             return dict(
                 id=self.id,
                 title=self.title,
-                insights=self.insights,
+                insights=[insight.model_dump() for insight in self.insights],
                 full_text=self.full_text,
             )
         else:
diff --git a/pages/3_🔍_Ask_and_Search.py b/pages/3_🔍_Ask_and_Search.py
index d70a3e9..964f208 100644
--- a/pages/3_🔍_Ask_and_Search.py
+++ b/pages/3_🔍_Ask_and_Search.py
@@ -3,7 +3,7 @@ import streamlit as st
 from open_notebook.domain.models import Model
 from open_notebook.domain.notebook import text_search, vector_search
 from open_notebook.graphs.rag import graph as rag_graph
-from pages.stream_app.utils import setup_page
+from pages.stream_app.utils import convert_source_references, setup_page
 
 setup_page("🔍 Search")
 
@@ -40,7 +40,7 @@ with ask_tab:
                 messages=messages
             ),  # config=dict(configurable=dict(model_id=model.id))
         )
-        st.markdown(rag_results["messages"][-1].content)
+        st.markdown(convert_source_references(rag_results["messages"][-1].content))
         with st.expander("Details (for debugging)"):
             st.json(rag_results)
 
diff --git a/pages/components/__init__.py b/pages/components/__init__.py
new file mode 100644
index 0000000..d6561b5
--- /dev/null
+++ b/pages/components/__init__.py
@@ -0,0 +1,11 @@
+from pages.components.note_panel import note_panel
+from pages.components.source_embedding_panel import source_embedding_panel
+from pages.components.source_insight import source_insight_panel
+from pages.components.source_panel import source_panel
+
+__all__ = [
+    "note_panel",
+    "source_embedding_panel",
+    "source_insight_panel",
+    "source_panel",
+]
diff --git a/pages/components/note_panel.py b/pages/components/note_panel.py
new file mode 100644
index 0000000..d1cba31
--- /dev/null
+++ b/pages/components/note_panel.py
@@ -0,0 +1,30 @@
+import streamlit as st
+from loguru import logger
+from streamlit_monaco import st_monaco  # type: ignore
+
+from open_notebook.domain.notebook import Note
+
+
+def note_panel(note_id, notebook_id=None):
+    note: Note = Note.get(note_id)
+    if not note:
+        raise ValueError(f"Note not fonud {note_id}")
+    t_preview, t_edit = st.tabs(["Preview", "Edit"])
+    with t_preview:
+        st.subheader(note.title)
+        st.markdown(note.content)
+    with t_edit:
+        note.title = st.text_input("Title", value=note.title)
+        note.content = st_monaco(
+            value=note.content, height="600px", language="markdown"
+        )
+        if st.button("Save", key=f"pn_edit_note_{note.id or 'new'}"):
+            logger.debug("Editing note")
+            note.save()
+            if not note.id and notebook_id:
+                note.add_to_notebook(notebook_id)
+            st.rerun()
+    if st.button("Delete", type="primary", key=f"delete_note_{note.id or 'new'}"):
+        logger.debug("Deleting note")
+        note.delete()
+        st.rerun()
diff --git a/pages/components/source_embedding_panel.py b/pages/components/source_embedding_panel.py
new file mode 100644
index 0000000..4ef4a29
--- /dev/null
+++ b/pages/components/source_embedding_panel.py
@@ -0,0 +1,17 @@
+import streamlit as st
+
+from open_notebook.domain.notebook import SourceEmbedding
+
+
+def source_embedding_panel(source_embedding_id):
+    si: SourceEmbedding = SourceEmbedding.get(source_embedding_id)
+    if not si:
+        raise ValueError(f"Embedding not found {source_embedding_id}")
+    with st.container(border=True):
+        url = f"Navigator?object_id={si.source.id}"
+        st.markdown("**Original Source**")
+        st.markdown(f"{si.source.title} [link](%s)" % url)
+    st.markdown(si.content)
+    if st.button("Delete", type="primary", key=f"delete_embedding_{si.id or 'new'}"):
+        si.delete()
+        st.rerun()
diff --git a/pages/components/source_insight.py b/pages/components/source_insight.py
new file mode 100644
index 0000000..ad38793
--- /dev/null
+++ b/pages/components/source_insight.py
@@ -0,0 +1,18 @@
+import streamlit as st
+
+from open_notebook.domain.notebook import SourceInsight
+
+
+def source_insight_panel(source, notebook_id=None):
+    si: SourceInsight = SourceInsight.get(source)
+    if not si:
+        raise ValueError(f"insight not found {source}")
+    st.subheader(si.insight_type)
+    with st.container(border=True):
+        url = f"Navigator?object_id={si.source.id}"
+        st.markdown("**Original Source**")
+        st.markdown(f"{si.source.title} [link](%s)" % url)
+    st.markdown(si.content)
+    if st.button("Delete", type="primary", key=f"delete_insight_{si.id or 'new'}"):
+        si.delete()
+        st.rerun()
diff --git a/pages/components/source_panel.py b/pages/components/source_panel.py
new file mode 100644
index 0000000..8fa130d
--- /dev/null
+++ b/pages/components/source_panel.py
@@ -0,0 +1,84 @@
+import streamlit as st
+import streamlit_scrollable_textbox as stx  # type: ignore
+import yaml
+from humanize import naturaltime
+
+from open_notebook.domain.notebook import Source
+from open_notebook.utils import surreal_clean
+from pages.stream_app.utils import run_patterns
+
+
+def source_panel(source_id: str, modal=False):
+    source: Source = Source.get(source_id)
+    if not source:
+        raise ValueError(f"Source not found: {source_id}")
+
+    current_title = source.title if source.title else "No Title"
+    source.title = st.text_input("Title", value=current_title)
+    if source.title != current_title:
+        st.toast("Saved new Title")
+        source.save()
+
+    process_tab, source_tab = st.tabs(["Process", "Source"])
+    with process_tab:
+        c1, c2 = st.columns([3, 1])
+        with c1:
+            title = st.empty()
+            if source.title:
+                title.subheader(source.title)
+            if source.asset and source.asset.url:
+                from_src = f"from URL: {source.asset.url}"
+            elif source.asset and source.asset.file_path:
+                from_src = f"from file: {source.asset.file_path}"
+            else:
+                from_src = "from text"
+            st.caption(f"Created {naturaltime(source.created)}, {from_src}")
+            for insight in source.insights:
+                with st.expander(f"**{insight.insight_type}**"):
+                    st.markdown(insight.content)
+                    if st.button(
+                        "Delete", type="primary", key=f"delete_insight_{insight.id}"
+                    ):
+                        insight.delete()
+                        st.rerun(scope="fragment" if modal else "app")
+
+        with c2:
+            with open("transformations.yaml", "r") as file:
+                transformations = yaml.safe_load(file)
+                for transformation in transformations["source_insights"]:
+                    if st.button(
+                        transformation["name"], help=transformation["description"]
+                    ):
+                        result = run_patterns(
+                            source.full_text, transformation["patterns"]
+                        )
+                        source.add_insight(
+                            transformation["insight_type"], surreal_clean(result)
+                        )
+                        st.rerun(scope="fragment" if modal else "app")
+
+            if st.button(
+                "Embed vectors",
+                icon="🦾",
+                disabled=source.embedded_chunks > 0,
+                help="This will generate your embedding vectors on the database for powerful search capabilities",
+            ):
+                source.vectorize()
+                st.success("Embedding complete")
+
+            chk_delete = st.checkbox(
+                "🗑️ Delete source", key=f"delete_source_{source.id}", value=False
+            )
+            if chk_delete:
+                st.warning(
+                    "Source will be deleted with all its insights and embeddings"
+                )
+                if st.button(
+                    "Delete", type="primary", key=f"bt_delete_source_{source.id}"
+                ):
+                    source.delete()
+                    st.rerun()
+
+    with source_tab:
+        st.subheader("Content")
+        stx.scrollableTextbox(source.full_text, height=300)
diff --git a/pages/stream_app/chat.py b/pages/stream_app/chat.py
index cb33852..71c0437 100644
--- a/pages/stream_app/chat.py
+++ b/pages/stream_app/chat.py
@@ -1,12 +1,18 @@
+from typing import Union
+
 import humanize
 import streamlit as st
 from langchain_core.runnables import RunnableConfig
 
+from open_notebook.domain.base import ObjectModel
 from open_notebook.domain.notebook import ChatSession, Note, Notebook, Source
 from open_notebook.graphs.chat import graph as chat_graph
 from open_notebook.plugins.podcasts import PodcastConfig
 from open_notebook.utils import token_count
-from pages.stream_app.utils import create_session_for_notebook
+from pages.stream_app.utils import (
+    convert_source_references,
+    create_session_for_notebook,
+)
 
 from .note import make_note_from_chat
 
@@ -26,13 +32,7 @@ def build_context(notebook_id):
         if "not in" in status:
             continue
 
-        # todo: there is problably a better way to handle this
-        if item_type == "note":
-            item: Note = Note.get(id)
-        elif item_type == "source":
-            item: Source = Source.get(id)
-        else:
-            continue
+        item: Union[Note, Source] = ObjectModel.get(id)
 
         if not item:
             continue
@@ -48,9 +48,10 @@ def build_context(notebook_id):
     return st.session_state[notebook_id]["context"]
 
 
-def execute_chat(txt_input, current_session):
+def execute_chat(txt_input, context, current_session):
     current_state = st.session_state[current_session.id]
     current_state["messages"] += [txt_input]
+    current_state["context"] = context
     result = chat_graph.invoke(
         input=current_state,
         config=RunnableConfig(configurable={"thread_id": current_session.id}),
@@ -146,10 +147,10 @@ def chat_sidebar(current_notebook: Notebook, current_session: ChatSession):
         with st.container(border=True):
             request = st.chat_input("Enter your question")
             # removing for now since it's not multi-model capable right now
-            st.caption(f"Total tokens: {tokens}")
             if request:
                 response = execute_chat(
                     txt_input=request,
+                    context=context,
                     current_session=current_session,
                 )
                 st.session_state[current_session.id]["messages"] = response["messages"]
@@ -161,7 +162,7 @@ def chat_sidebar(current_notebook: Notebook, current_session: ChatSession):
                     continue
 
                 with st.chat_message(name=msg.type):
-                    st.write(msg.content)
+                    st.markdown(convert_source_references(msg.content))
                     if msg.type == "ai":
                         if st.button("💾 New Note", key=f"render_save_{msg.id}"):
                             make_note_from_chat(
diff --git a/pages/stream_app/note.py b/pages/stream_app/note.py
index 0516a1c..3dc96d4 100644
--- a/pages/stream_app/note.py
+++ b/pages/stream_app/note.py
@@ -3,11 +3,11 @@ from typing import Optional
 import streamlit as st
 from humanize import naturaltime
 from loguru import logger
-from streamlit_monaco import st_monaco  # type: ignore
 
 from open_notebook.domain.notebook import Note
 from open_notebook.graphs.multipattern import graph as pattern_graph
 from open_notebook.utils import surreal_clean
+from pages.components import note_panel
 
 from .consts import context_icons
 
@@ -25,29 +25,8 @@ def add_note(notebook_id):
 
 
 @st.dialog("Add a Source", width="large")
-def note_panel(notebook_id=None, note: Optional[Note] = None):
-    if not note:
-        note: Note = Note(note_type="human")
-
-    t_preview, t_edit = st.tabs(["Preview", "Edit"])
-    with t_preview:
-        st.subheader(note.title)
-        st.markdown(note.content)
-    with t_edit:
-        note.title = st.text_input("Title", value=note.title)
-        note.content = st_monaco(
-            value=note.content, height="600px", language="markdown"
-        )
-        if st.button("Save", key=f"pn_edit_note_{note.id or 'new'}"):
-            logger.debug("Editing note")
-            note.save()
-            if not note.id:
-                note.add_to_notebook(notebook_id)
-            st.rerun()
-    if st.button("Delete", type="primary", key=f"delete_note_{note.id or 'new'}"):
-        logger.debug("Deleting note")
-        note.delete()
-        st.rerun()
+def note_panel_dialog(note: Optional[Note] = None, notebook_id=None):
+    note_panel(note_id=note.id, notebook_id=notebook_id)
 
 
 def make_note_from_chat(content, notebook_id=None):
@@ -88,7 +67,7 @@ def note_card(note, notebook_id):
         st.caption(f"Updated: {naturaltime(note.updated)}")
 
         if st.button("Expand", icon="📝", key=f"edit_note_{note.id}"):
-            note_panel(notebook_id=notebook_id, note=note)
+            note_panel_dialog(notebook_id=notebook_id, note=note)
 
     st.session_state[notebook_id]["context_config"][note.id] = context_state
 
@@ -105,4 +84,4 @@ def note_list_item(note_id, score=None):
     ):
         st.write(note.content)
         if st.button("Edit Note", icon="📝", key=f"x_edit_note_{note.id}"):
-            note_panel(note=note)
+            note_panel_dialog(note=note)
diff --git a/pages/stream_app/source.py b/pages/stream_app/source.py
index c249b36..fdc56c5 100644
--- a/pages/stream_app/source.py
+++ b/pages/stream_app/source.py
@@ -2,8 +2,6 @@ import os
 from pathlib import Path
 
 import streamlit as st
-import streamlit_scrollable_textbox as stx  # type: ignore
-import yaml
 from humanize import naturaltime
 from loguru import logger
 
@@ -11,17 +9,13 @@ from open_notebook.config import UPLOADS_FOLDER
 from open_notebook.domain.notebook import Asset, Source
 from open_notebook.exceptions import UnsupportedTypeException
 from open_notebook.graphs.content_processing import graph
-from open_notebook.graphs.multipattern import graph as transform_graph
 from open_notebook.utils import surreal_clean
+from pages.components import source_panel
+from pages.stream_app.utils import run_patterns
 
 from .consts import context_icons
 
 
-def run_patterns(input_text, patterns):
-    output = transform_graph.invoke(dict(content_stack=[input_text], patterns=patterns))
-    return output["output"]
-
-
 # moved it here to replace it with the pipeline on 0.1.0
 def generate_toc_and_title(source) -> "Source":
     try:
@@ -43,80 +37,8 @@ def generate_toc_and_title(source) -> "Source":
 
 
 @st.dialog("Source", width="large")
-def source_panel(source_id):
-    source: Source = Source.get(source_id)
-    if not source:
-        st.error("Source not found")
-        return
-    current_title = source.title if source.title else "No Title"
-    source.title = st.text_input("Title", value=current_title)
-    if source.title != current_title:
-        st.toast("Saved new Title")
-        source.save()
-
-    process_tab, source_tab = st.tabs(["Process", "Source"])
-    with process_tab:
-        c1, c2 = st.columns([3, 1])
-        with c1:
-            title = st.empty()
-            if source.title:
-                title.subheader(source.title)
-            if source.asset.url:
-                from_src = f"from URL: {source.asset.url}"
-            elif source.asset.file_path:
-                from_src = f"from file: {source.asset.file_path}"
-            else:
-                from_src = "from text"
-            st.caption(f"Created {naturaltime(source.created)}, {from_src}")
-            for insight in source.insights:
-                with st.expander(f"**{insight.insight_type}**"):
-                    st.markdown(insight.content)
-                    if st.button(
-                        "Delete", type="primary", key=f"delete_insight_{insight.id}"
-                    ):
-                        insight.delete()
-                        st.rerun(scope="fragment")
-
-        with c2:
-            with open("transformations.yaml", "r") as file:
-                transformations = yaml.safe_load(file)
-                for transformation in transformations["source_insights"]:
-                    if st.button(
-                        transformation["name"], help=transformation["description"]
-                    ):
-                        result = run_patterns(
-                            source.full_text, transformation["patterns"]
-                        )
-                        source.add_insight(
-                            transformation["insight_type"], surreal_clean(result)
-                        )
-                        st.rerun(scope="fragment")
-
-            if st.button(
-                "Embed vectors",
-                icon="🦾",
-                disabled=source.embedded_chunks > 0,
-                help="This will generate your embedding vectors on the database for powerful search capabilities",
-            ):
-                source.vectorize()
-                st.success("Embedding complete")
-
-            chk_delete = st.checkbox(
-                "🗑️ Delete source", key=f"delete_source_{source.id}", value=False
-            )
-            if chk_delete:
-                st.warning(
-                    "Source will be deleted with all its insights and embeddings"
-                )
-                if st.button(
-                    "Delete", type="primary", key=f"bt_delete_source_{source.id}"
-                ):
-                    source.delete()
-                    st.rerun()
-
-    with source_tab:
-        st.subheader("Content")
-        stx.scrollableTextbox(source.full_text, height=300)
+def source_panel_dialog(source_id):
+    source_panel(source_id)
 
 
 @st.dialog("Add a Source", width="large")
@@ -207,7 +129,7 @@ def source_card(source, notebook_id):
             f"Updated: {naturaltime(source.updated)}, **{len(source.insights)}** insights"
         )
         if st.button("Expand", icon="📝", key=source.id):
-            source_panel(source.id)
+            source_panel_dialog(source.id)
 
     st.session_state[notebook_id]["context_config"][source.id] = context_state
 
@@ -226,4 +148,4 @@ def source_list_item(source_id, score=None):
             st.markdown(f"**{insight.insight_type}**")
             st.write(insight.content)
         if st.button("Edit source", icon="📝", key=f"x_edit_source_{source.id}"):
-            source_panel(source_id=source.id)
+            source_panel_dialog(source_id=source.id)
diff --git a/pages/stream_app/utils.py b/pages/stream_app/utils.py
index b790db0..2caae54 100644
--- a/pages/stream_app/utils.py
+++ b/pages/stream_app/utils.py
@@ -1,3 +1,4 @@
+import re
 from datetime import datetime
 from typing import List, Union
 
@@ -8,6 +9,7 @@ from open_notebook.database.migrate import MigrationManager
 from open_notebook.domain.models import model_manager
 from open_notebook.domain.notebook import ChatSession, Notebook
 from open_notebook.graphs.chat import ThreadState, graph
+from open_notebook.graphs.multipattern import graph as transform_graph
 from open_notebook.utils import (
     compare_versions,
     get_installed_version,
@@ -15,6 +17,11 @@ from open_notebook.utils import (
 )
 
 
+def run_patterns(input_text, patterns):
+    output = transform_graph.invoke(dict(content_stack=[input_text], patterns=patterns))
+    return output["output"]
+
+
 def version_sidebar():
     with st.sidebar:
         try:
@@ -76,8 +83,6 @@ def setup_stream_state(current_notebook: Notebook) -> ChatSession:
             logger.debug("Getting last updated session")
             chat_session = sessions[0]
 
-    logger.debug(f"Chat session: {chat_session}")
-
     if not chat_session or chat_session.id is None:
         raise ValueError("Problem acquiring chat session")
     # sets the active session for the notebook
@@ -163,3 +168,36 @@ def setup_page(title: str, layout="wide", sidebar_state="expanded"):
     check_migration()
     check_models()
     version_sidebar()
+
+
+def convert_source_references(text):
+    """
+    Converts source references in brackets to markdown-style links.
+
+    Matches patterns like [source_insight:id], [note:id], [source:id], or [source_embedding:id]
+    and converts them to markdown links.
+
+    Args:
+        text (str): The input text containing source references
+
+    Returns:
+        str: Text with source references converted to markdown links
+
+    Example:
+        >>> text = "Here is a reference [source_insight:abc123]"
+        >>> convert_source_references(text)
+        'Here is a reference [source_insight:abc123](/?object_id=source_insight:abc123)'
+    """
+
+    # Pattern matches [type:id] where type can be source_insight, note, source, or source_embedding
+    pattern = r"\[((?:source_insight|note|source|source_embedding):[\w\d]+)\]"
+
+    def replace_match(match):
+        """Helper function to create the markdown link"""
+        source_ref = match.group(1)  # Gets the content inside brackets
+        return f"[[{source_ref}]](/?object_id={source_ref})"
+
+    # Replace all matches in the text
+    converted_text = re.sub(pattern, replace_match, text)
+
+    return converted_text
diff --git a/prompts/chat.jinja b/prompts/chat.jinja
index 0ba3471..dab2eda 100644
--- a/prompts/chat.jinja
+++ b/prompts/chat.jinja
@@ -5,9 +5,8 @@ You are a cognitive study assistant that helps users research and learn by engag
 - Access to project information and selected documents (CONTEXT)
 - Can engage in natural dialogue while maintaining academic rigor
 
-# FORMULATE YOUR DATA
-- Generate your answer based on the CONTEXT information
-- Ensure that your response is accurate and relevant to the user's query
+# YOUR OPERATING METHOD
+Whenever a user asks you a question, you need to identify the query context and the user intent. The user might be continuing a previous conversation or asking a new question. Looking at the CONTEXT will probably give you a hint of what the user is looking for. Once you identify the user intent, formulate your answer accordingly paying attention to the CITING INSTRUCTIONS below.
 
 {% if notebook %}
 # PROJECT INFORMATION
@@ -18,5 +17,26 @@ You are a cognitive study assistant that helps users research and learn by engag
 {% if context %}
 # CONTEXT
 
+The user has selected this context to help you with your response:
+
 {{context}}
-{% endif %}
\ No newline at end of file
+{% endif %}
+
+# CITING INSTRUCTIONS
+
+If your answer is based off of any item in the context, it's very important that your response contains references to the searched documents so the user can follow-up and read more about the topic. The way you do that is by adding the id of the specific document in between brackets like this: [document_id].
+
+## EXAMPLE
+
+User: Can you tell me more about the concept of "Deep Learning"?
+
+Assistant: Deep learning is a subset of machine learning in artificial intelligence (AI) that enables networks to learn unsupervised from unstructured or unlabeled data. [note:iuiodadalknda]. It can also be categorized into three main types: supervised, unsupervised, and reinforcement learning. [insight:adadadadadadad].
+
+Please note, "note:iuiodadalknda" and "insight:adadadadadadad" are examples of document IDs with different prefixes. You should not make up document IDs or copy the IDs from this example. You should use the IDs of the documents that you have access to through the search tool.
+
+## IMPORTANT
+
+- Do not make up documents or document ids. Only use the ids of the documents that you have access through the query you made.
+- The ID is composed of the type of document and a random string, such as "source:randomstring", "note:randomstring", or "insight:randomstring". There are various types of documents, including notes, insights, and sources. **Always use the complete ID exactly as it is provided, including its type prefix. Do not add, remove, or modify any part of the ID.**
+- Do not assume or change the type prefix of any document ID. If a document ID is "note:xyz", use it exactly as "note:xyz". Do not change it to "source:xyz" or any other variation.
+- **Use document IDs exactly as they are returned from the search tool. Do not add any prefixes or modify them in any way.**