diff --git a/04-call-summaries/chunking.py b/04-call-summaries/chunking.py
new file mode 100644
index 0000000..5aa10bf
--- /dev/null
+++ b/04-call-summaries/chunking.py
@@ -0,0 +1,134 @@
+import datetime
+from langchain_text_splitters import (
+    RecursiveCharacterTextSplitter,
+    NLTKTextSplitter,
+    SpacyTextSplitter,
+)
+from langchain_core.prompts import PromptTemplate
+from langchain.docstore.document import Document
+from llm import LLM
+from run import get_transcript
+
+
+# split text into chunks
+def get_text_chunks(text, chunk_size, chunk_overlap, text_splitter_choice):
+    if text_splitter_choice == "2":
+        text_splitter = NLTKTextSplitter()
+    elif text_splitter_choice == "3":
+        text_splitter = SpacyTextSplitter()
+    else:
+        text_splitter = RecursiveCharacterTextSplitter(
+            chunk_size=chunk_size, chunk_overlap=chunk_overlap
+        )
+
+    texts = text_splitter.split_text(text)
+
+    docs = [
+        Document(
+            page_content=t,
+        )
+        for t in texts
+    ]
+
+    return docs
+
+
+CHUNKING_PROMPT = """
+You are a helpful AI assistant tasked with summarizing transcripts, however we can only process the transcripts in pieces.
+Please fill out and return the following template: {template} with data in the text: {text}
+If the following template already has the field filled out, do not overwrite this information.
+"""
+
+initial_temp = """
+1. Caller Information:
+- Name
+- Contact Information
+- Availability
+- Household Information
+2. Reason/Type of Call: e.g., Applying for benefits, Follow-ups
+3. Previous Benefits History:
+- Applied for
+- Receives
+- Denied
+4. Benefits Discussion: Prefix the discussed benefit with a hashtag (e.g., #SNAP, #LIHEAP)
+5. Discussion Points:
+- Key information points
+6. Documents Needed: e.g., Income verification, Housing documentation
+7. Next Steps for Client
+8. Next Steps for Agent
+"""
+
+
+def chunking_ingest(transcript, prompt):
+    text_chunks = get_text_chunks(
+        transcript, chunk_size=750, chunk_overlap=300, text_splitter_choice="2"
+    )
+    prompt_template = PromptTemplate.from_template(prompt)
+    template = initial_temp
+
+    print("""
+        Select an llm
+        1. openhermes (default)
+        2. dolphin
+        3. gemini
+        4. gpt 4
+        5. gpt 4o
+        6. claude 3
+        """)
+
+    llm = input() or "1"
+
+    if llm == "2":
+        client = LLM(client_name="ollama", model_name="dolphin-mistral")
+        print("""----------
+            Dolphin
+            """)
+
+    elif llm == "3":
+        client = LLM(client_name="gemini")
+        print("""----------
+            Gemini Flash 1.5
+            """)
+    elif llm == "4":
+        print("""----------
+        GPT 4
+        """)
+        client = LLM(client_name="gpt", model_name="gpt4")
+    elif llm == "5":
+        print("""----------
+        GPT 4o
+        """)
+        client = LLM(client_name="gpt", model_name="gpt-4o")
+    elif llm == "6":
+        print("""----------
+            Claude 3
+            """)
+        client = LLM(client_name="claude")
+    else:
+        print("""
+            Openhermes
+            """)
+        client = LLM(client_name="ollama", model_name="openhermes")
+
+    client.init_client()
+    ct = datetime.datetime.now()
+    print("current time:-", ct)
+    for text in text_chunks:
+        formatted_prompt = prompt_template.format(
+            text=text.page_content, template=template
+        )
+        print("Processing Text Chunk")
+        template = client.generate_text(prompt=formatted_prompt)
+    print("Complete")
+    return template
+
+
+if __name__ == "__main__":
+    print(
+        chunking_ingest(
+            transcript=get_transcript("./son_calling_behalf_mother_transcript.txt"),
+            prompt=CHUNKING_PROMPT,
+        )
+    )
+    ct = datetime.datetime.now()
+    print(ct)
diff --git a/04-call-summaries/llm.py b/04-call-summaries/llm.py
index 1975b1c..b9e68c9 100644
--- a/04-call-summaries/llm.py
+++ b/04-call-summaries/llm.py
@@ -9,79 +9,96 @@
 dotenv.load_dotenv()
 
 
-def get_transcript(file_path="./transcript.txt"):
-    file = open(file_path, encoding="utf-8")
-    content = file.read()
-    return content
-
-
-def ollama_client(
-    model_name=None,
-    prompt=None,
-    callbacks=None,
-    settings=None,
-):
-    if not settings:
-        settings = {
-            # "temperature": 0.1,
-            # "system": "",
-            # "template": "",
-            # See https://github.com/langchain-ai/langchain/blob/master/libs/community/langchain_community/llms/ollama.py
-            "stop": None
-        }
-
-    print("LLM settings:", model_name, settings)
-    # To connect via another URL: Ollama(base_url='http://localhost:11434', ...)
-    return Ollama(model=model_name, callbacks=callbacks, **settings).invoke(prompt)
-
-
-def google_gemini_client(
-    model_name="gemini-pro",
-    prompt=None,
-    settings=None,
-):
-    # Get a Google API key by following the steps after clicking on Get an API key button
-    # at https://ai.google.dev/tutorials/setup
-    GOOGLE_API_KEY = os.environ.get("GOOGLE_API_KEY")
-
-    print("LLM settings:", model_name, settings)
-
-    genai.configure(api_key=GOOGLE_API_KEY)
-    if settings:
-        genai.GenerationConfig(**settings)
-    model = genai.GenerativeModel(model_name)
-    return model.generate_content(prompt)
-
-
-def gpt3_5(prompt, model="gpt-3.5-turbo"):
-    # Get API key from https://platform.openai.com/api-keys
-    OPEN_AI_API_KEY = os.environ.get("OPEN_AI_API_KEY")
-    openai_client = OpenAI(api_key=OPEN_AI_API_KEY)  # Uses OPENAI_API_KEY
-    return (
-        openai_client.chat.completions.create(
-            model=model, messages=[{"role": "user", "content": prompt}]
-        )
-        .choices[0]
-        .message.content
-    )
-
-
-def gpt_4_turbo(prompt):
-    return gpt3_5(prompt, model="gpt-4-turbo")
-
-
-def claude(prompt, model="claude-3-opus-20240229", max_tokens=1024):
-    # Get API key from https://console.anthropic.com/settings/keys
-    ANTHROPIC_API_KEY = os.environ.get("ANTHROPIC_API_KEY")
-
-    client = anthropic.Anthropic(
-        api_key=ANTHROPIC_API_KEY,
-    )
-    generated_response = client.messages.create(
-        model=model,
-        max_tokens=max_tokens,
-        messages=[{"role": "user", "content": prompt}],
-    ).content
-    text_response = "\n".join([text_block.text for text_block in generated_response])
-
-    return text_response
+class LLM:
+    def __init__(
+        self,
+        client_name=None,
+        model_name=None,
+        max_tokens=1024,
+        settings=None,
+    ):
+        self.client_name = client_name
+        """Name of llm selection"""
+        self.model_name = model_name
+        """User friendly model name"""
+        self.model_version = model_name
+        """Exact model name being passed into the initializer"""
+        self.max_tokens = max_tokens
+        self.client = None
+        self.settings = settings
+
+    def init_client(self):
+        """Retrieves the llm client"""
+        if self.client_name == "ollama":
+            if self.settings is None:
+                self.settings = {
+                    # "temperature": 0.1,
+                    # "system": "",
+                    # "template": "",
+                    # See https://github.com/langchain-ai/langchain/blob/master/libs/community/langchain_community/llms/ollama.py
+                    "stop": None
+                }
+            if self.model_name is None:
+                self.model_name = "openhermes"
+            # To connect via another URL: Ollama(base_url='http://localhost:11434', ...)
+            self.client = Ollama(model=self.model_version, **self.settings)
+
+        elif self.client_name == "gemini":
+            # Get a Google API key by following the steps after clicking on Get an API key button
+            # at https://ai.google.dev/tutorials/setup
+            GOOGLE_API_KEY = os.environ.get("GOOGLE_API_KEY")
+            if self.model_name == "gemini-pro":
+                self.model_version = "gemini-1.5-pro-latest"
+            else:
+                self.model_version = "gemini-1.5-flash-latest"
+
+            genai.configure(api_key=GOOGLE_API_KEY)
+            if self.settings is not None:
+                genai.GenerationConfig(**self.settings)
+            self.client = genai.GenerativeModel(self.model_version)
+
+        elif self.client_name == "gpt":
+            if self.model_name == "gpt3":
+                self.model_version = "gpt-3.5-turbo"
+            elif self.model_name == "gpt4":
+                self.model_version = "gpt-4-turbo"
+            else:
+                self.model_version = "gpt-4o"
+
+            # Get API key from https://platform.openai.com/api-keys
+            OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY")
+            self.client = OpenAI(api_key=OPENAI_API_KEY)  # Uses OPENAI_API_KEY
+
+        elif self.client_name == "claude":
+            self.model_version = "claude-3-opus-20240229"
+            ANTHROPIC_API_KEY = os.environ.get("ANTHROPIC_API_KEY")
+            self.client = anthropic.Anthropic(
+                api_key=ANTHROPIC_API_KEY,
+            )
+
+    def generate_text(self, prompt=None):
+        """Generates response given prompt"""
+        if self.client_name == "ollama":
+            return self.client.invoke(prompt)
+        elif self.client_name == "gemini":
+            return self.client.generate_content(prompt).text
+        elif self.client_name == "gpt":
+            return (
+                self.client.chat.completions.create(
+                    model=self.model_version,
+                    messages=[{"role": "user", "content": prompt}],
+                )
+                .choices[0]
+                .message.content
+            )
+        elif self.client_name == "claude":
+            generated_response = self.client.messages.create(
+                model=self.model_version,
+                max_tokens=self.max_tokens,
+                messages=[{"role": "user", "content": prompt}],
+            ).content
+            text_response = "\n".join(
+                [text_block.text for text_block in generated_response]
+            )
+
+            return text_response
diff --git a/04-call-summaries/requirements.txt b/04-call-summaries/requirements.txt
index 5b8da85..c986355 100644
--- a/04-call-summaries/requirements.txt
+++ b/04-call-summaries/requirements.txt
@@ -3,4 +3,9 @@ google-generativeai
 tokenizers
 langchain
 langchain_community
-openai
\ No newline at end of file
+openai
+sentence-transformers
+nltk
+spacy==3.7.4
+langchain-text-splitters
+langchain_openai
\ No newline at end of file
diff --git a/04-call-summaries/run.py b/04-call-summaries/run.py
index 43c8eea..b271d7b 100644
--- a/04-call-summaries/run.py
+++ b/04-call-summaries/run.py
@@ -1,6 +1,6 @@
 from langchain_core.prompts import PromptTemplate
 
-from llm import claude, google_gemini_client, gpt3_5, gpt_4_turbo, ollama_client
+from llm import LLM
 
 
 # download transcripts from https://drive.google.com/drive/folders/19r6x3Zep4N9Rl_x4n4H6RpWkXviwbxyw?usp=sharing
@@ -32,88 +32,109 @@ def get_transcript(file_path="./transcript.txt"):
 8. Next Steps for Agent
 """
 
-print("""
-      Select an llm
-      1. openhermes (default)
-      2. dolphin
-      3. gemini
-      4. gpt 3.5
-      5. gpt 4
-      6. claude 3
-      7. all
-      """)
-
 
 transcript = get_transcript()
-llm = input() or "1"
 prompt_template = PromptTemplate.from_template(PROMPT)
 formatted_prompt = prompt_template.format(transcript=transcript)
 
-if llm == "2":
-    test = ollama_client(model_name="dolphin-mistral", prompt=formatted_prompt)
-    print("""----------
-        Dolphin
-        """)
-elif llm == "3":
-    test = google_gemini_client(prompt=formatted_prompt).text
-    print("""----------
-        Gemini
-        """)
-elif llm == "4":
-    print("""----------
-      GPT 3.5
-      """)
-    test = gpt3_5(prompt=formatted_prompt)
-elif llm == "5":
-    print("""----------
-        GPT 4
-        """)
-    test = gpt_4_turbo(prompt=formatted_prompt)
-elif llm == "6":
-    print("""----------
-        Claude 3
-        """)
-    test = claude(prompt=formatted_prompt)
-elif llm == "7":
-    test_open_hermes = ollama_client(model_name="openhermes", prompt=formatted_prompt)
+
+def stuffing_summary(prompt=None):
     print("""
-        Openhermes
+        Select an llm
+        1. openhermes (default)
+        2. dolphin
+        3. gemini
+        4. gpt 4
+        5. gpt 4o
+        6. claude 3
+        7. all
         """)
-    print(test_open_hermes)
 
-    test_dolphin = ollama_client(model_name="dolphin-mistral", prompt=formatted_prompt)
-    print("""----------
-        Dolphin
-        """)
-    print(test_dolphin)
+    llm = input() or "1"
 
-    test_gemini = google_gemini_client(prompt=formatted_prompt).text
-    print("""----------
-        Gemini
+    if llm == "2":
+        client = LLM(client_name="ollama", model_name="dolphin-mistral")
+        print("""----------
+            Dolphin
+            """)
+    elif llm == "3":
+        client = LLM(client_name="gemini")
+        print("""----------
+            Gemini Flash 1.5
+            """)
+    elif llm == "4":
+        print("""----------
+        GPT 4
+        """)
+        client = LLM(client_name="gpt", model_name="gpt4")
+    elif llm == "5":
+        print("""----------
+        GPT 4o
         """)
-    print(test_gemini)
+        client = LLM(client_name="gpt", model_name="gpt4o")
+    elif llm == "6":
+        print("""----------
+            Claude 3
+            """)
+        client = LLM(client_name="claude")
+    elif llm == "7":
+        print("""
+            Openhermes
+            """)
+        ollama_openhermes = LLM(client_name="ollama", model_name="openhermes")
+        ollama_openhermes.init_client()
+        ollama_openhermes_response = ollama_openhermes.generate_text(prompt=prompt)
+        print(ollama_openhermes_response)
 
-    print("""----------
-      GPT 3.5
-      """)
-    test_gpt3_5 = gpt3_5(prompt=formatted_prompt)
-    print(test_gpt3_5)
+        print("""----------
+            Dolphin
+            """)
+        ollama_dolphin = LLM(client_name="ollama", model_name="dolphin-mistral")
+        ollama_dolphin.init_client()
+        dolphin_response = ollama_dolphin.generate_text(prompt=prompt)
+        print(dolphin_response)
 
-    print("""----------
+        print("""----------
+            Gemini Flash 1.5
+            """)
+        gemini = LLM(client_name="gemini")
+        gemini.init_client()
+        gemini_response = gemini.generate_text(prompt=prompt)
+        print(gemini_response)
+
+        print("""----------
         GPT 4
         """)
-    test_gpt4 = gpt_4_turbo(prompt=formatted_prompt)
-    print(test_gpt4)
+        gpt_4 = LLM(client_name="gpt", model_name="gpt4")
+        gpt_4.init_client()
+        gpt_4_response = gpt_4.generate_text(prompt=prompt)
+        print(gpt_4_response)
 
-    print("""----------
-        Claude 3
-        """)
-    test_claude = claude(prompt=formatted_prompt)
-    print(test_claude)
-else:
-    test = ollama_client(model_name="openhermes", prompt=formatted_prompt)
-    print("""
-        Openhermes
+        print("""----------
+        GPT 4o
         """)
-if test:
-    print(test)
+        gpt_4o = LLM(client_name="_4o", model_name="gpt4o")
+        gpt_4o.init_client()
+        gpt_4o_response = gpt_4o.generate_text(prompt=prompt)
+        print(gpt_4o_response)
+
+        print("""----------
+            Claude 3
+            """)
+        claude = LLM(client_name="claude")
+        claude.init_client()
+        claude_response = claude.generate_text(prompt=prompt)
+        print(claude_response)
+    else:
+        client = LLM(client_name="ollama", model_name="openhermes")
+        print("""
+            Openhermes
+            """)
+    client.init_client()
+    response = client.generate_text(prompt=prompt)
+    if response:
+        print(response)
+
+
+if __name__ == "__main__":
+    stuffing_summary(prompt=formatted_prompt)