Valeria_Chatbot!.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/chrisagon/valeria/blob/masterfr/Valeria_Chatbot!.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Y7JTDlXJLmKc"
      },
      "source": [
        "# Welcome to the Lenny Chatbot Colab!\n",
        "\n",
        "This Colab notebook contains all of the code you need to make a basic chatbot that will answer questions about a corpus of text. Colab is a cloud-based programming environment which will let you run all of this code from your browser.\n",
        "\n",
        "At each step, follow the written instructions and press the \"play\" button next to the code sample in order to run it.\n",
        "\n",
        "**Important Note:** This is a basic chatbot running on a limited selection of articles. It's only a starting point to show you what's possible!\n",
        "\n",
        "If you have questions, feel free to reach out to me on Twitter at [@danshipper](https://www.twitter.com/danshipper)."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JwdhZA9KPRRP"
      },
      "source": [
        "## 1. Download our text corpus\n",
        "\n",
        "The first thing we need to do is download the text our chatbot is going to use as reference material for answering questions.\n",
        "\n",
        "In the Lenny Chatbot, I used every article he's written as the text corpus. But for this public codebase, I've collected two articles from his archive that we can use as a starting point.\n",
        "\n",
        "These are the articles I'm using:\n",
        "\n",
        "- [What is good retention?](https://www.lennysnewsletter.com/p/what-is-good-retention-issue-29)\n",
        "- [How the biggest consumer apps got their first 1,000 users\n",
        "](https://www.lennysnewsletter.com/p/how-the-biggest-consumer-apps-got)\n",
        "\n",
        "You can replace these articles with any text corpus you want, however.\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1N5jdPr9P7SZ",
        "outputId": "a86275f7-701e-489c-ac8f-88fa8df69ea9"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "fatal: destination path 'Lenny-Newsletter-Corpus' already exists and is not an empty directory.\n"
          ]
        }
      ],
      "source": [
        "! git clone https://github.com/EveryInc/Lenny-Newsletter-Corpus"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8s2D2f7BQM_E"
      },
      "source": [
        "# 2. Install our dependencies and define our functions\n",
        "\n",
        "In this section we'll install GPT Index and Langchain. We'll also define the functions that we'll use later to construct our index and query it.\n",
        "\n",
        "First, let's install our dependencies."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "hwEIqQd3aL0r"
      },
      "outputs": [],
      "source": [
        "!pip install gpt-index\n",
        "!pip install langchain"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qiQSJU9baR4u"
      },
      "source": [
        "Now, we'll define the functions we're going to use later in order to construct our index and query it."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "J8qiIErLLiRX"
      },
      "outputs": [],
      "source": [
        "from gpt_index import SimpleDirectoryReader, GPTListIndex, readers, GPTSimpleVectorIndex, LLMPredictor, PromptHelper\n",
        "from langchain import OpenAI\n",
        "import sys\n",
        "import os\n",
        "from IPython.display import Markdown, display\n",
        "\n",
        "def construct_index(directory_path):\n",
        "    # set maximum input size\n",
        "    max_input_size = 4096\n",
        "    # set number of output tokens\n",
        "    num_outputs = 256\n",
        "    # set maximum chunk overlap\n",
        "    max_chunk_overlap = 20\n",
        "    # set chunk size limit\n",
        "    chunk_size_limit = 600\n",
        "\n",
        "    # define LLM\n",
        "    llm_predictor = LLMPredictor(llm=OpenAI(temperature=0, model_name=\"text-davinci-003\", max_tokens=num_outputs))\n",
        "    prompt_helper = PromptHelper(max_input_size, num_outputs, max_chunk_overlap, chunk_size_limit=chunk_size_limit)\n",
        " \n",
        "    documents = SimpleDirectoryReader(directory_path).load_data()\n",
        "    \n",
        "    index = GPTSimpleVectorIndex(\n",
        "        documents, llm_predictor=llm_predictor, prompt_helper=prompt_helper\n",
        "    )\n",
        "\n",
        "    index.save_to_disk('index.json')\n",
        "\n",
        "    return index\n",
        "\n",
        "def ask_lenny():\n",
        "    index = GPTSimpleVectorIndex.load_from_disk('index.json')\n",
        "    while True: \n",
        "        query = input(\"What do you want to ask Lenny? \")\n",
        "        response = index.query(query, response_mode=\"compact\")\n",
        "        display(Markdown(f\"Lenny Bot says: <b>{response.response}</b>\"))\n",
        "  "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pftgPf5rQqox"
      },
      "source": [
        "# 3. Set OpenAI API Key\n",
        "In order to run this notebook you'll need an API key from OpenAI. \n",
        "\n",
        "If you don't have one already, you can grab one by [signing up](https://platform.openai.com/overview). Then click your account icon on the top right of the screen and select \"View API Keys\". Create an API key.\n",
        "\n",
        "Then run the code below and paste it into the text input.\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "TZ6WSBhxQLbK"
      },
      "outputs": [],
      "source": [
        "os.environ[\"OPENAI_API_KEY\"] = input(\"Paste your OpenAI API key here and hit enter:\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gyIIBFSkRbVU"
      },
      "source": [
        "# 4. Construct Index\n",
        "\n",
        "Now we're going to construct our index. This will take every file in the folder 'Lenny-Newsletter-Corpus', split it into chunks, and embed it with OpenAI's embeddings API.\n",
        "\n",
        "**Important Note:** This step costs money. Running it on the text corpus we've given you by default should only cost $0.03 in total. But if you use other pieces of text be careful if they're really long.\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BqCTqL9jRiLm",
        "outputId": "105adb83-654c-4a93-90a2-85536a99d0f8"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "> Adding chunk: <h1>👉 <a href=\"https://www.lennysnewsletter.com...\n",
            "> Adding chunk: <em>every</em> startup confronts this question ...\n",
            "> Adding chunk: 424w, https://substackcdn.com/image/fetch/w_848...\n",
            "> Adding chunk: width=\"1456\" height=\"819\" data-attrs=\"{&quot;sr...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: to your user directly</strong> — online, offlin...\n",
            "> Adding chunk: class=\"image2-inset\"><picture><source type=\"ima...\n",
            "> Adding chunk: 1456w\" sizes=\"100vw\"><img src=\"https://bucketee...\n",
            "> Adding chunk: 424w, https://substackcdn.com/image/fetch/w_848...\n",
            "> Adding chunk: Wolfe and Justin Mateen would basically <strong...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: width=\"636\" height=\"474\" data-attrs=\"{&quot;src...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: Moreau</a></p></blockquote><h4><strong> Startup...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: width=\"612\" height=\"612\" data-attrs=\"{&quot;src...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: the companies and giving them Lyft credits.</st...\n",
            "> Adding chunk: class=\"image2-inset\"><picture><source type=\"ima...\n",
            "> Adding chunk: 1456w\" sizes=\"100vw\"><img src=\"https://bucketee...\n",
            "> Adding chunk: 424w, https://substackcdn.com/image/fetch/w_848...\n",
            "> Adding chunk: then, the founding team knew that it would only...\n",
            "> Adding chunk: 424w, https://substackcdn.com/image/fetch/w_848...\n",
            "> Adding chunk: width=\"1024\" height=\"683\" data-attrs=\"{&quot;sr...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: stores — Pinterest </strong></h4><blockquote><p...\n",
            "> Adding chunk: Popel</a></p></blockquote><h4>App Store — TikTo...\n",
            "> Adding chunk: geared toward DVD fanatics and movie buffs, bef...\n",
            "> Adding chunk: class=\"image-link is-viewable-img image2\" targe...\n",
            "> Adding chunk: 1272w, https://substackcdn.com/image/fetch/w_14...\n",
            "> Adding chunk: class=\"sizing-normal\" alt=\"\" srcset=\"https://su...\n",
            "> Adding chunk: width=\"16\" height=\"16\" viewBox=\"0 0 24 24\" fill...\n",
            "> Adding chunk: other startups, and other entrepreneurs</strong...\n",
            "> Adding chunk: Silbermann</a></p></blockquote><p class=\"button...\n",
            "> Adding chunk: when’d do I get to play with it?” and that’s th...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: 5&quot;,&quot;type&quot;:null,&quot;href&quot;:...\n",
            "> Adding chunk: 1456w\" sizes=\"100vw\" loading=\"lazy\"></picture><...\n",
            "> Adding chunk: class=\"image2-inset\"><picture><source type=\"ima...\n",
            "> Adding chunk: 1456w\" sizes=\"100vw\"><img src=\"https://bucketee...\n",
            "> Adding chunk: 424w, https://substackcdn.com/image/fetch/w_848...\n",
            "> Adding chunk: the email inbox management app for iPhone that ...\n",
            "> Adding chunk: class=\"image2-inset\"><picture><source type=\"ima...\n",
            "> Adding chunk: 1456w\" sizes=\"100vw\"><img src=\"https://bucketee...\n",
            "> Adding chunk: 424w, https://substackcdn.com/image/fetch/w_848...\n",
            "> Adding chunk: last thing from our minds when we launched Robi...\n",
            "> Adding chunk: each having 5 invites at first to share with th...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: width=\"613\" height=\"320\" data-attrs=\"{&quot;src...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: Malik's post on the 15th</a> really pushed it o...\n",
            "> Adding chunk: 424w, https://substackcdn.com/image/fetch/w_848...\n",
            "> Adding chunk: width=\"596\" height=\"318\" data-attrs=\"{&quot;src...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: Hoover</a></p></blockquote><h4>Instagram</h4><b...\n",
            "> Adding chunk: day of shopping it around. So four days all in....\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: width=\"1456\" height=\"1101\" data-attrs=\"{&quot;s...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: that launched on Product Hunt to reporters that...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: width=\"640\" height=\"526\" data-attrs=\"{&quot;src...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: as it sounds, the coverage by these bottom-feed...\n",
            "> Adding chunk: i.e. use whatever you've got going for you), th...\n",
            "> Adding chunk: the MVP and other respected product people, sha...\n",
            "> Adding chunk: are your early target users, and where they cur...\n",
            "> Adding chunk: <a href=\"https://www.shoutmeloud.com/popular-co...\n",
            "> Adding chunk: <p><em>Hello, and welcome to the&nbsp;<strong>f...\n",
            "> Adding chunk: of the triple-word-score.\"</em></p><p>— <a href...\n",
            "> Adding chunk: Transactional</strong>:  ~30% is GOOD, ~50% is ...\n",
            "> Adding chunk: 424w, https://substackcdn.com/image/fetch/w_848...\n",
            "> Adding chunk: width=\"1456\" height=\"1681\" data-attrs=\"{&quot;s...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: (SurveyMonkey, Miro), <strong><a href=\"https://...\n",
            "> Adding chunk: some, these retention benchmarks will seem high...\n",
            "> Adding chunk: (i.e. using the product, making a purchase, pos...\n",
            "> Adding chunk: 31% 3-month user retention, 22% 24-month (<a hr...\n",
            "> Adding chunk: includes companies such as <strong>Netflix, Spo...\n",
            "> Adding chunk: / Mid-Market SaaS: ~60% is GOOD, ~80% is GREAT<...\n",
            "> Adding chunk: 79% 12-month customer retention (<a href=\"http:...\n",
            "> Adding chunk: Johns</strong>: Over 70% is GOOD, over 90% is G...\n",
            "> Adding chunk: of Very Small business (VSB) is common because ...\n",
            "> Adding chunk: GREAT</p></li><li><p><strong>Fareed Mosavat: </...\n",
            "> Adding chunk: 139% (<a href=\"https://www.saastr.com/5-interes...\n",
            "> Adding chunk: Balfour</strong>: Over 100% is GOOD, over 120% ...\n",
            "> Adding chunk: 135% (<a href=\"https://docs.google.com/spreadsh...\n",
            "> Adding chunk: you’ve found this newsletter valuable, consider...\n",
            "> Adding chunk: <a class=\"image-link is-viewable-img image2\" ta...\n",
            "> Adding chunk: 1272w, https://substackcdn.com/image/fetch/w_14...\n",
            "> Adding chunk: class=\"sizing-normal\" alt=\"\" srcset=\"https://su...\n",
            "> Adding chunk: viewBox=\"0 0 24 24\" fill=\"none\" stroke=\"#FFFFFF...\n",
            "> Adding chunk: the company scaled to thousands of global emplo...\n",
            "> Adding chunk: move quickly&nbsp;when&nbsp;<a href=\"https://ww...\n",
            "> Adding chunk: Human tunnels. Fun facts.&nbsp;Silly but regula...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: width=\"1100\" height=\"1467\" data-attrs=\"{&quot;s...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: a problem&nbsp;statement&nbsp;is the single&nbs...\n",
            "> Adding chunk: to 10 times the goal.&nbsp;This&nbsp;ambitious ...\n",
            "> Adding chunk: you don’t hit the goal but get close, congratul...\n",
            "> Adding chunk: class=\"image-link is-viewable-img image2\" targe...\n",
            "> Adding chunk: 1272w, https://substackcdn.com/image/fetch/w_14...\n",
            "> Adding chunk: class=\"sizing-normal\" alt=\"\" srcset=\"https://su...\n",
            "> Adding chunk: width=\"16\" height=\"16\" viewBox=\"0 0 24 24\" fill...\n",
            "> Adding chunk: use it?)&nbsp;and “want” problems (do I want to...\n",
            "> Adding chunk: lot has been said about goals (for example,&nbs...\n",
            "> Adding chunk: class=\"image2-inset\"><picture><source type=\"ima...\n",
            "> Adding chunk: 1456w\" sizes=\"100vw\"><img src=\"https://bucketee...\n",
            "> Adding chunk: 424w, https://substackcdn.com/image/fetch/w_848...\n",
            "> Adding chunk: Your Voice”, by&nbsp;<a href=\"https://shantellm...\n",
            "> Adding chunk: not as obvious as you think.&nbsp;Get feedback ...\n",
            "> Adding chunk: users to focus on the task at hand is one of th...\n",
            "> Adding chunk: class=\"image-link is-viewable-img image2\" targe...\n",
            "> Adding chunk: 1272w, https://substackcdn.com/image/fetch/w_14...\n",
            "> Adding chunk: class=\"sizing-normal\" alt=\"\" srcset=\"https://su...\n",
            "> Adding chunk: width=\"16\" height=\"16\" viewBox=\"0 0 24 24\" fill...\n",
            "> Adding chunk: <a class=\"image-link is-viewable-img image2\" ta...\n",
            "> Adding chunk: 1272w, https://substackcdn.com/image/fetch/w_14...\n",
            "> Adding chunk: class=\"sizing-normal\" alt=\"\" srcset=\"https://su...\n",
            "> Adding chunk: stroke=\"#FFFFFF\" stroke-width=\"2\" stroke-lineca...\n",
            "> Adding chunk: class=\"image2-inset\"><picture><source type=\"ima...\n",
            "> Adding chunk: 1456w\" sizes=\"100vw\"><img src=\"https://bucketee...\n",
            "> Adding chunk: 424w, https://substackcdn.com/image/fetch/w_848...\n",
            "> Adding chunk: early design of the product we launched, giving...\n",
            "> Adding chunk: class=\"image2-inset\"><picture><source type=\"ima...\n",
            "> Adding chunk: 1456w\" sizes=\"100vw\"><img src=\"https://bucketee...\n",
            "> Adding chunk: 424w, https://substackcdn.com/image/fetch/w_848...\n",
            "> Adding chunk: 1: Crystallize the&nbsp;problem&nbsp;you are so...\n",
            "> Adding chunk: class=\"image-link is-viewable-img image2\" targe...\n",
            "> Adding chunk: 1272w, https://substackcdn.com/image/fetch/w_14...\n",
            "> Adding chunk: class=\"sizing-normal\" alt=\"\" srcset=\"https://su...\n",
            "> Adding chunk: 0 24 24\" fill=\"none\" stroke=\"#FFFFFF\" stroke-wi...\n",
            "> Adding chunk: detail on question.</p><a class=\"image-link ima...\n",
            "> Adding chunk: 1272w, https://substackcdn.com/image/fetch/w_14...\n",
            "> Adding chunk: class=\"sizing-normal\" alt=\"\" srcset=\"https://su...\n",
            "> Adding chunk: description of what you’re thinking, so that fo...\n",
            "> Adding chunk: are dropping off at too high a rate at the fina...\n",
            "> Adding chunk: over quantity.&nbsp;</strong>Three to five stro...\n",
            "> Adding chunk: think a metric makes sense for your goal (think...\n",
            "> Adding chunk: class=\"image2-inset\"><picture><source type=\"ima...\n",
            "> Adding chunk: 1456w\" sizes=\"100vw\"><img src=\"https://bucketee...\n",
            "> Adding chunk: 424w, https://substackcdn.com/image/fetch/w_848...\n",
            "> Adding chunk: statements are like&nbsp;silver burritos.&nbsp;...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: width=\"274\" height=\"200\" data-attrs=\"{&quot;src...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: below, where Jerry and Elaine attempt to get a ...\n",
            "> Adding chunk: 1272w, https://substackcdn.com/image/fetch/w_14...\n",
            "> Adding chunk: class=\"sizing-normal\" alt=\"\" srcset=\"https://su...\n",
            "> Adding chunk: 0 24 24\" fill=\"none\" stroke=\"#FFFFFF\" stroke-wi...\n",
            "> Adding chunk: 424w, https://substackcdn.com/image/fetch/w_848...\n",
            "> Adding chunk: width=\"320\" height=\"200\" data-attrs=\"{&quot;src...\n",
            "> Adding chunk: 848w, https://substackcdn.com/image/fetch/w_127...\n",
            "> Adding chunk: not clear,&nbsp;ask “What problem are we trying...\n",
            "> Adding chunk: Subtle Art of Not Giving a F*ck</a></p></li><li...\n",
            "> [build_index_from_documents] Total LLM token usage: 0 tokens\n",
            "> [build_index_from_documents] Total embedding token usage: 82572 tokens\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "<gpt_index.indices.vector_store.simple.GPTSimpleVectorIndex at 0x7fe0a046c520>"
            ]
          },
          "execution_count": 6,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "construct_index('/content/Lenny-Newsletter-Corpus')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4gHdfdtsSGEW"
      },
      "source": [
        "# 5. Ask Questions!\n",
        "\n",
        "Now we'll run the \"ask_lenny\" function we defined above. \n",
        "\n",
        "This will prompt the you to input a question, and then it will find chunks of text that might answer the question, and summarize the answer from those text chunks using GPT-3.\n",
        "\n",
        "Remember, in this public Colab file we're only using two of Lenny's articles for our corpus. So it will only answer questions from:\n",
        "\n",
        "- [What is good retention?](https://www.lennysnewsletter.com/p/what-is-good-retention-issue-29)\n",
        "- [How the biggest consumer apps got their first 1,000 users\n",
        "](https://www.lennysnewsletter.com/p/how-the-biggest-consumer-apps-got)\n",
        "\n",
        "\n",
        "A few sample questions you can ask:\n",
        "\n",
        "- What is good retention for a consumer social product?\n",
        "\n",
        "- How did DoorDash get its first users?\n",
        "\n",
        "- How did LinkedIn get started?\n",
        "\n",
        "Again, this step costs money. So be aware!"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "background_save": true,
          "base_uri": "https://localhost:8080/",
          "height": 201
        },
        "id": "Pk7Gt-Kg_1dH",
        "outputId": "41879dd6-2e3f-4481-e610-4e1bd79fed7f"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "> [query] Total LLM token usage: 570 tokens\n",
            "> [query] Total embedding token usage: 9 tokens\n"
          ]
        },
        {
          "data": {
            "text/markdown": "Lenny Bot says: <b>\n~25% is GOOD</b>",
            "text/plain": [
              "<IPython.core.display.Markdown object>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "> [query] Total LLM token usage: 611 tokens\n",
            "> [query] Total embedding token usage: 7 tokens\n"
          ]
        },
        {
          "data": {
            "text/markdown": "Lenny Bot says: <b>\nReid intentionally seeded the product with successful friends and connections, recognizing that cultivating an aspirational brand was crucial to drive mainstream adoption. He shared the product with progressively larger groups and received feedback to amplify the adoption.</b>",
            "text/plain": [
              "<IPython.core.display.Markdown object>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "ask_lenny()"
      ]
    }
  ],
  "metadata": {
    "colab": {
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}