[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"blog.post.building-an-ai-shopping-co-pilot":3,"blog.related.building-an-ai-shopping-co-pilot":129,"blog.adjacent.building-an-ai-shopping-co-pilot":419},{"data":4,"meta":125},[5],{"id":6,"Title":7,"Content":8,"Description":9,"Category":10,"slug":11,"createdAt":12,"updatedAt":13,"publishedAt":14,"readTime":15,"documentId":16,"Image":17,"author":67,"Seo":112},19,"Build an AI shopping co-pilot","We recently [client-side power-ups and improvements to the SDK](https:\u002F\u002Fchatthing.ai\u002Fblog\u002Fclient-side-power-ups-and-sdk-improvements) that make it easy to create amazing integrations with your websites and web apps. In this tutorial, we will demonstrate how you can build an AI-powered shopping co-pilot that can perform actions on your user's behalf and generally be super helpful!\n\nYou can take a look at what we will be building here: [AI shopping copilot demo](https:\u002F\u002Fai-shopping-copilot-demo.pages.dev\u002F)\nand you can find all the code here: [Github AI shopping copilot demo](https:\u002F\u002Fgithub.com\u002Fpixelhop\u002Fai-shopping-copilot-demo)\n\nIf you prefer watching to reading, checkout the video version of this post:\n\n{% youtube  id=\"DobIEEFHAWo\" \u002F%}\n\n## Let's get started\n\nFirst, we need to create our bot in Chat Thing. We created a new bot, gave it a name, some example questions to guide the user on what it can do, and the most important part, the system message prompt. \n\nTake a look at our settings below: \n\n![shopping-co-pilot-settings.webp](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1720717806\u002Fshopping_co_pilot_settings_3fcbfce39e.webp)\n\nThis is a simple demo, so our prompt doesn't need to be complicated:\n\n>You are a helpful shopping assistant at a store that sells AI merchandise.\n>\n>You can navigate to the following URLs\n>* \"\u002F\" - The home page\n>* \"\u002Fproducts\u002F:productId\" - A product page, where \"productId\" is replaced with the actual id of a product\n>* \"\u002Fcheckout\" - The checkout page\n>\n>You can offer to navigate customers to the above page if you feel it useful, or if they would like more info about a product.\n>\n>Answer in a fun, informal, jokey, friendly tone.\n\nWe tell the bot it's a shopping assistant, we give it info about the URLs it can navigate to (more on that later), and we tell it the tone it should use. All pretty standard prompting!\n\nFor this demo, we are using GPT4-o as our model, but any model that supports power-ups will work.\n\n## Turn on advanced SDK features\n\nBefore embedding your bot on your site, you should turn on the advanced SDK features by going to your bot \"Channels\" page, and then the \"Web Channel Settings\" and enabling the toggle.\n\n![Xnapper July 11 Screenshot.jpg](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1720728221\u002FXnapper_July_11_Screenshot_57e9cc71d9.jpg)\n\nThis setting is optional, as enabling it makes your bot more susceptible to prompt hijacking by allowing you to override and extend the prompt from the frontend. This is a tradeoff that is necessary if you need the advanced client-side functionality.\n\n## Embed your bot onto your site\n\nNow that you have a bot and the advanced SDK functionality is turned on, it's time to grab its embed script and add it to your site. \n\nClick the \"Embed\" button in your bot settings, and then grab the \"Embed as chat widget\" script and add it to the HTML of your site.\n\n![Xnapper July 11.jpg](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1720728672\u002FXnapper_July_11_66982b6d60.jpg)\n\nOurs looks like this:\n\n![Xnapper July 11 Screenshot (1).jpg](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1720728804\u002FXnapper_July_11_Screenshot_1_577e800510.jpg)\n\nYou should now see your chat widget on the page and be able to open it.\n\n![Xnapper July 15 (6).webp](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1721059071\u002FXnapper_July_15_6_e657c7214f.webp)\n\n## Create AI co-pilots for your site with client-side power-ups\n\nWith the bot now on your site, the real fun of integration can begin. Our new Chat Thing SDK has lots of functionality that you can read about here: [Chat Thing AI SDK](https:\u002F\u002Fchatthing.ai\u002Fdocs\u002Fsdk), but for this tutorial, we will mainly be focusing on using client-side power-ups so our bot can take actions on behalf of the user right on the front-end of your website or app.\n\nBefore we can do that, we need to get hold of the Chat Thing SDK that is added to the global window object after our Chat Thing script loads. Due to the Chat Thing script being deferred, it might not be ready as soon as the page loads, so we need a bit of code that will wait for it to be ready and return the SDK when it is. (We will be improving how this works later, as well as releasing an NPM package that won't have this issue).\n\nWe create a simple composable that returns the SDK like below:\n\n```typescript\nexport const useChatThing = () => {\n    function getChatThing(resolve: (sdk: Window['chatThing']) => void) {\n        if (window.chatThing) {\n            return resolve(window.chatThing);\n        }\n\n        setTimeout(() => getChatThing(resolve), 200);\n    }\n\n    return new Promise\u003CWindow['chatThing']>((resolve) => {\n        getChatThing(resolve);\n    });\n};\n```\n\nWe can now get hold of the Chat Thing SDK ready to use like this:\n\n```typescript\nconst chatThing = await useChatThing();\n```\n\nNow that we have a reference to the Chat Thing SDK, we can use it to tightly integrate the bot with our app.\n\n### Letting out bot know about the available products\n\nBefore we can get the bot to do cool things like add a product to a cart for us, we need to let it know about the available products. For the purpose of keeping this tutorial short and simple, we have chosen to do this using the new \"Extend system message\" functionality of the SDK. \n\n```typescript\nchatThing.systemMessage('extend', `\n    \u003Cproducts>\n    ${JSON.stringify(products)}\n    \u003C\u002Fproducts>\n  `);\n```\n\nThis allows you to add additional information to your system message on the fly. Here we stringify our list of products and add them to the system message.\n\nIn case you are wondering, our fake products array looks a bit like this:\n\n```typescript\nexport const products: Product[] = [\n  {\n    id: \"AI-001\",\n    name: \"AI-Generated Fortune Cookies\",\n    price: 9.99,\n    image: window.location.origin + \"\u002Fimages\u002Fai-fortune-cookies.webp\",\n    description: `\u003Cp>Indulge in our \u003Cstrong>AI-Generated Fortune Cookies\u003C\u002Fstrong>, where every cookie contains a unique, humorous fortune created by artificial intelligence. Perfect for parties, gifts, or just a good laugh!\u003C\u002Fp>`,\n  },\n  ...\n```\n\nOf course, in a real e-commerce app, you will not have a single array containing all your products lying around, or at least I hope you don't, and even if you did, it's not a great idea to use your system message as a way of communicating these to the bot as you will waste thousands and thousands of tokens with each message. \n\nThe best implementation would be to create a client-side power-up that allows your bot to dynamically search through your products inventory. That's a tutorial in itself, so for the time being, let's put up with our basic implementation.\n\n### Registering your first custom client-side power-ups with the SDK\n\nIn our example Vue.js e-commerce store, we do this in the `onMounted` hook of `App.vue` so that power-ups are loaded as soon as the Vue.js app is mounted.\n\nLet's see how we can create a power-up to add a product to the basket. \n\n```typescript\n  chatThing.registerPowerUp({\n    name: 'Add to basket',\n    description: 'Add a product to the basket',\n    parameters: {\n      productId: {\n        type: 'string',\n        description: 'The unique product ID',\n        required: true\n      },\n      qty: {\n        type: 'number',\n        description: 'How many to add',\n        required: true\n      }\n    },\n    handler: (args) => {\n      return addToCart(args.productId, args.qty);\n    }\n  });\n```\n\nThere's a bit to break down here, so let's go through it step by step. We pass the `registerPowerUp` function a configuration object that contains a few things.\n\nFirst up, we have `name` and `description`. Name is hopefully very obviously the name of the power-up. The name will help the bot decide when to use this power-up and will also be shown in the chat UI while the power-up is running. \n\nDescription is very important and is the primary information the bot will use to determine when to use your power-up. In this case, adding an item to a cart is a relatively simple action, so a description of \"Add a product to the basket\" is enough.\n\nNext, we have the `parameters` object. This object lets you define the arguments your bot will need to collect before it can perform your action. So in this example, to add a product to the basket, we need a product ID and the quantity of items to be added.\n\nEach parameter object has the following:\n\n* `type` - The type of parameter: \"string\", \"number\", \"boolean\", or \"object\"\n* `description` - Similar to the power-up description, the parameter description lets the bot know what this parameter is and how it should fill it.\n* `required` - A boolean value, true if the parameter is always required by the action, false if not\n* \nFinally, we have the `handler` callback. This is where you add the implementation for this power-up. When the bot decides to use this action, the handler will be called and passed an arguments object containing the parameters you defined. \n\nIn this case we will receive an object containing the product ID and quantity looking like this:\n\n```typescript\n{\n  productId: 'some-product-id',\n  qty: 2\n}\n```\n\nYou can do anything you like in the handler. You could call an API, use an existing function in your app, trigger an update to a global store, or whatever else you would like to do. The only rule is that your handler must return something back to the bot. This could be a success or error message, or some useful information the bot can use for future answers.\n\nIn our example, we use our existing add to cart function that updates our fake cart, and just for reference, it looks something like this:\n\n```typescript\nexport function addToCart(productId: string, qty: number) {\n  const product = products.find((product) => product.id === productId);\n  const existingProductInCart = cart.value.find(\n    (product) => product.id === productId\n  );\n\n  if (existingProductInCart) {\n    existingProductInCart.qty += qty;\n  } else if (product) {\n    cart.value.push({\n      ...product,\n      qty,\n    });\n  }\n\n  return {\n    cart: cart.value,\n    totalPrice: totalPrice.value,\n  };\n}\n```\n\nRemember, this is a fake e-commerce implementation, but it's useful to see what the function returns. It returns an object that contains the current cart and all the items in it, and a total price property. This is returned back to the bot so it can update you on the new price of the cart as well as the items in it.\n\nAnd with all that done, we should now be able to add a product to the cart using the bot!\n\n![Xnapper July 15.jpg](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1721067556\u002FXnapper_July_15_2052889232.jpg)\n\n### Adding more functionality\n\nNow that you've added your first client-side power-up, hopefully it should be quite straightforward for you to add others. I won't go through how we implemented each of the others in our demo (although remember you can find all the code [here](https:\u002F\u002Fgithub.com\u002Fpixelhop\u002Fai-shopping-copilot-demo\u002Ftree\u002Fmain)), but we added additional power-ups to remove items, get the cart, check order status, and finally to navigate the user to other pages. \n\n### Wrapping up\n\nThe possibilities with client-side power-ups are absolutely huge, and we can't wait to see what you do with them! We will be posting more detailed tutorials like this for exciting use cases, so if there's anything in particular you would like to see, please let us know!","Learn how to build an AI-powered shopping co-pilot using Chat Thing's client-side power-ups. Create a smart assistant that enhances your e-commerce experience.","blog","building-an-ai-shopping-co-pilot","2024-07-11T16:36:06.203Z","2024-07-23T15:10:02.199Z","2024-07-23T15:10:02.190Z",null,"tvo2vw4hnfeiayjrsp9rxuqd",{"id":18,"name":19,"alternativeText":15,"caption":15,"width":20,"height":21,"formats":22,"hash":58,"ext":24,"mime":27,"size":59,"url":60,"previewUrl":15,"provider":61,"provider_metadata":62,"createdAt":63,"updatedAt":64,"documentId":65,"publishedAt":66,"focalPoint":15},420,"build_and_ai_shopping_co_pilot.webp",1920,1080,{"large":23,"small":34,"medium":42,"thumbnail":50},{"ext":24,"url":25,"hash":26,"mime":27,"name":28,"path":15,"size":29,"width":30,"height":31,"provider_metadata":32},".webp","https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1721069499\u002Flarge_build_and_ai_shopping_co_pilot_7a58c01202.webp","large_build_and_ai_shopping_co_pilot_7a58c01202","image\u002Fwebp","large_build_and_ai_shopping_co_pilot.webp",33.09,1000,562,{"public_id":26,"resource_type":33},"image",{"ext":24,"url":35,"hash":36,"mime":27,"name":37,"path":15,"size":38,"width":39,"height":40,"provider_metadata":41},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1721069499\u002Fsmall_build_and_ai_shopping_co_pilot_7a58c01202.webp","small_build_and_ai_shopping_co_pilot_7a58c01202","small_build_and_ai_shopping_co_pilot.webp",14.13,500,281,{"public_id":36,"resource_type":33},{"ext":24,"url":43,"hash":44,"mime":27,"name":45,"path":15,"size":46,"width":47,"height":48,"provider_metadata":49},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1721069499\u002Fmedium_build_and_ai_shopping_co_pilot_7a58c01202.webp","medium_build_and_ai_shopping_co_pilot_7a58c01202","medium_build_and_ai_shopping_co_pilot.webp",23.73,750,422,{"public_id":44,"resource_type":33},{"ext":24,"url":51,"hash":52,"mime":27,"name":53,"path":15,"size":54,"width":55,"height":56,"provider_metadata":57},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1721069499\u002Fthumbnail_build_and_ai_shopping_co_pilot_7a58c01202.webp","thumbnail_build_and_ai_shopping_co_pilot_7a58c01202","thumbnail_build_and_ai_shopping_co_pilot.webp",5.3,245,138,{"public_id":52,"resource_type":33},"build_and_ai_shopping_co_pilot_7a58c01202",73.67,"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1721069499\u002Fbuild_and_ai_shopping_co_pilot_7a58c01202.webp","cloudinary",{"public_id":58,"resource_type":33},"2024-07-15T18:51:40.165Z","2024-07-15T18:52:48.524Z","ivhb48yxfpsf69kiudadfa73","2026-05-15T15:25:35.762Z",{"id":68,"Name":69,"createdAt":70,"updatedAt":71,"publishedAt":72,"documentId":73,"Image":74},2,"Zef","2024-03-20T13:35:42.255Z","2024-03-20T13:35:43.988Z","2024-03-20T13:35:43.980Z","qpnfq2shgxc6b1too3b96idh",{"id":75,"name":76,"alternativeText":15,"caption":15,"width":77,"height":77,"formats":78,"hash":106,"ext":80,"mime":83,"size":107,"url":108,"previewUrl":15,"provider":61,"provider_metadata":109,"createdAt":110,"updatedAt":110,"documentId":111,"publishedAt":66,"focalPoint":15},255,"IMG_2278.jpg",1972,{"large":79,"small":87,"medium":93,"thumbnail":99},{"ext":80,"url":81,"hash":82,"mime":83,"name":84,"path":15,"size":85,"width":30,"height":30,"provider_metadata":86},".jpg","https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1710941716\u002Flarge_IMG_2278_d3b5e4fa69.jpg","large_IMG_2278_d3b5e4fa69","image\u002Fjpeg","large_IMG_2278.jpg",194.09,{"public_id":82,"resource_type":33},{"ext":80,"url":88,"hash":89,"mime":83,"name":90,"path":15,"size":91,"width":39,"height":39,"provider_metadata":92},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1710941717\u002Fsmall_IMG_2278_d3b5e4fa69.jpg","small_IMG_2278_d3b5e4fa69","small_IMG_2278.jpg",54.08,{"public_id":89,"resource_type":33},{"ext":80,"url":94,"hash":95,"mime":83,"name":96,"path":15,"size":97,"width":47,"height":47,"provider_metadata":98},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1710941716\u002Fmedium_IMG_2278_d3b5e4fa69.jpg","medium_IMG_2278_d3b5e4fa69","medium_IMG_2278.jpg",114.8,{"public_id":95,"resource_type":33},{"ext":80,"url":100,"hash":101,"mime":83,"name":102,"path":15,"size":103,"width":104,"height":104,"provider_metadata":105},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1710941716\u002Fthumbnail_IMG_2278_d3b5e4fa69.jpg","thumbnail_IMG_2278_d3b5e4fa69","thumbnail_IMG_2278.jpg",6.59,156,{"public_id":101,"resource_type":33},"IMG_2278_d3b5e4fa69",641.66,"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1710941716\u002FIMG_2278_d3b5e4fa69.jpg",{"public_id":106,"resource_type":33},"2024-03-20T13:35:18.234Z","b2ylr2lzfx0rstj4qs9sh8b7",{"id":113,"metaTitle":7,"metaDescription":9,"keywords":15,"metaRobots":15,"structuredData":15,"metaViewport":15,"canonicalURL":15,"metaImage":114},70,{"id":18,"name":19,"alternativeText":15,"caption":15,"width":20,"height":21,"formats":115,"hash":58,"ext":24,"mime":27,"size":59,"url":60,"previewUrl":15,"provider":61,"provider_metadata":124,"createdAt":63,"updatedAt":64,"documentId":65,"publishedAt":66,"focalPoint":15},{"large":116,"small":118,"medium":120,"thumbnail":122},{"ext":24,"url":25,"hash":26,"mime":27,"name":28,"path":15,"size":29,"width":30,"height":31,"provider_metadata":117},{"public_id":26,"resource_type":33},{"ext":24,"url":35,"hash":36,"mime":27,"name":37,"path":15,"size":38,"width":39,"height":40,"provider_metadata":119},{"public_id":36,"resource_type":33},{"ext":24,"url":43,"hash":44,"mime":27,"name":45,"path":15,"size":46,"width":47,"height":48,"provider_metadata":121},{"public_id":44,"resource_type":33},{"ext":24,"url":51,"hash":52,"mime":27,"name":53,"path":15,"size":54,"width":55,"height":56,"provider_metadata":123},{"public_id":52,"resource_type":33},{"public_id":58,"resource_type":33},{"pagination":126},{"page":127,"pageSize":128,"pageCount":127,"total":127},1,25,{"data":130,"meta":415},[131,236,317],{"id":132,"Title":133,"Content":134,"Description":135,"Category":10,"slug":136,"createdAt":137,"updatedAt":138,"publishedAt":139,"readTime":140,"documentId":141,"Image":142,"author":185},58,"Lead Capture is live: meet your visitors before the first message","Every chat on your website starts the same way.\n\nA stranger arrives. They ask their question. Your agent answers. They leave.\n\nYou never find out who they were. They might have been your ideal customer. They might have been ready to buy. You will never know, because nothing asked.\n\nThat changes today.\n\n![The visitor-facing pre-chat form rendered on a bot page, with intro copy above the fields](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1777051094\u002Fdocs\u002Fprechat-form-visitor.png)\n\n### Meet your visitors before the first message\n\nLead Capture shows a short form to visitors before the chat opens. They fill in their name and email, or whatever your team actually needs to know, and then the conversation starts.\n\nBy the time they send their first message, you already know who they are.\n\n{% youtube id=\"ygNJiEvskZ4\" \u002F%}\n\n### Your team knows who called\n\nEvery form submission is saved to the conversation. Open any chat in Chat Thing and the visitor's details are right there at the top, before the first message.\n\nIf you want those details in your CRM, your inbox, or your team's Slack channel, you can connect that too. It is optional, but it means your team can follow up the next morning knowing exactly who they spoke to, what they asked, and what they came in for.\n\n### Ask for what you actually need\n\nThe default form asks for a name and an email. That is usually enough.\n\nBut if your team needs more, the company, the product they are asking about, a callback date, you can add those fields too. Drag to reorder. Mark the ones that matter as required.\n\nThere is also a short intro message you can put above the form. Use it to explain what the chat is for, set the right expectation, or just make people feel welcome before they fill anything in.\n\n![The \"Require a pre-chat form\" toggle and intro copy field in the web channel settings](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1777051081\u002Fdocs\u002Fprechat-form-toggle.png)\n\n### Why asking up front works\n\nMost teams try to capture emails at the end of a chat. A popup. A nudge. A \"before you go\" moment.\n\nThe conversion on those is brutal. The visitor got their answer and they are already on their way out.\n\nAsking at the start works because the visitor is already here. They came to get help. Filling in a short form is not friction, it is an introduction. And when they do it, the whole conversation that follows is warmer, more personal, and easier to follow up on.\n\n### Where this makes the biggest difference\n\nIf you sell anything with a sales process, demos, pricing questions, onboarding, Lead Capture means your team goes into every follow-up with context instead of a cold email address.\n\nIf you run a B2B site, knowing the company before the chat starts is the difference between a useful conversation and a polite one.\n\nIf you do lead generation, this replaces the popup form entirely. Same details. Better experience. And the visitor gets their questions answered on the way in.\n\n### The honest caveat\n\nLead Capture is on the web channel for now. It does not work on WhatsApp, Slack, or Discord, those channels do not have a natural place for a form before the conversation starts. We will think about how this could work on other channels in the future.\n\n### Privacy, kept simple\n\nVisitors see the form before they share anything. Required fields are marked. The intro copy is yours to write, including any consent line your jurisdiction needs. Their details stay in your Chat Thing account and only go to the tools you choose to connect.\n\n### Getting started\n\nLive today on Pro and Business plans.\n\nGo to your agent, open Channels, find the web channel settings, and turn on \"Require a pre-chat form.\" Pick your fields. Save. The next visitor who opens your chat will see it.","Lead Capture is a native pre-chat form on the web channel. Visitors fill it in, the chat opens, and their answers flow into your systems via webhook so your assistant knows who it is talking to from message one.","launch-week-05-lead-capture","2026-05-01T13:34:00.052Z","2026-05-15T09:40:45.108Z","2026-05-08T07:54:14.803Z",3,"hztpimum2ubw64jvkf1s6pzn",{"id":143,"name":144,"alternativeText":145,"caption":146,"width":20,"height":21,"formats":147,"hash":179,"ext":149,"mime":152,"size":180,"url":181,"previewUrl":15,"provider":61,"provider_metadata":182,"createdAt":183,"updatedAt":183,"documentId":184,"publishedAt":66,"focalPoint":15},838,"day-05-lead-capture-feature-card-v3-1920x1080","Lead Capture — Day 05","",{"large":148,"small":158,"medium":165,"thumbnail":172},{"ext":149,"url":150,"hash":151,"mime":152,"name":153,"path":15,"size":154,"width":30,"height":155,"sizeInBytes":156,"provider_metadata":157},".png","https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1778228678\u002Flarge_day_05_lead_capture_feature_card_v3_1920x1080_791009ded2.png","large_day_05_lead_capture_feature_card_v3_1920x1080_791009ded2","image\u002Fpng","large_day-05-lead-capture-feature-card-v3-1920x1080",368.13,563,368134,{"public_id":151,"resource_type":33},{"ext":149,"url":159,"hash":160,"mime":152,"name":161,"path":15,"size":162,"width":39,"height":40,"sizeInBytes":163,"provider_metadata":164},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1778228678\u002Fsmall_day_05_lead_capture_feature_card_v3_1920x1080_791009ded2.png","small_day_05_lead_capture_feature_card_v3_1920x1080_791009ded2","small_day-05-lead-capture-feature-card-v3-1920x1080",103.14,103137,{"public_id":160,"resource_type":33},{"ext":149,"url":166,"hash":167,"mime":152,"name":168,"path":15,"size":169,"width":47,"height":48,"sizeInBytes":170,"provider_metadata":171},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1778228678\u002Fmedium_day_05_lead_capture_feature_card_v3_1920x1080_791009ded2.png","medium_day_05_lead_capture_feature_card_v3_1920x1080_791009ded2","medium_day-05-lead-capture-feature-card-v3-1920x1080",215.41,215406,{"public_id":167,"resource_type":33},{"ext":149,"url":173,"hash":174,"mime":152,"name":175,"path":15,"size":176,"width":55,"height":56,"sizeInBytes":177,"provider_metadata":178},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1778228678\u002Fthumbnail_day_05_lead_capture_feature_card_v3_1920x1080_791009ded2.png","thumbnail_day_05_lead_capture_feature_card_v3_1920x1080_791009ded2","thumbnail_day-05-lead-capture-feature-card-v3-1920x1080",33.65,33650,{"public_id":174,"resource_type":33},"day_05_lead_capture_feature_card_v3_1920x1080_791009ded2",201.83,"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1778228678\u002Fday_05_lead_capture_feature_card_v3_1920x1080_791009ded2.png",{"public_id":179,"resource_type":33},"2026-05-08T08:24:39.310Z","kvyyazu3r6xxzw6qgrbu6c8m",{"id":186,"Name":187,"createdAt":188,"updatedAt":189,"publishedAt":190,"documentId":191,"Image":192},5,"Ella","2025-09-16T11:48:50.521Z","2025-09-16T12:07:31.210Z","2025-09-16T12:07:31.205Z","g8rg6x370xgt055caa9e1irh",{"id":193,"name":194,"alternativeText":15,"caption":15,"width":195,"height":196,"formats":197,"hash":230,"ext":199,"mime":83,"size":231,"url":232,"previewUrl":15,"provider":61,"provider_metadata":233,"createdAt":234,"updatedAt":234,"documentId":235,"publishedAt":66,"focalPoint":15},681,"IMG_5631.JPG",3648,5472,{"large":198,"small":207,"medium":215,"thumbnail":222},{"ext":199,"url":200,"hash":201,"mime":83,"name":202,"path":15,"size":203,"width":204,"height":30,"sizeInBytes":205,"provider_metadata":206},".JPG","https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1758024414\u002Flarge_IMG_5631_d9b99d485a.jpg","large_IMG_5631_d9b99d485a","large_IMG_5631.JPG",54.45,667,54447,{"public_id":201,"resource_type":33},{"ext":199,"url":208,"hash":209,"mime":83,"name":210,"path":15,"size":211,"width":212,"height":39,"sizeInBytes":213,"provider_metadata":214},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1758024414\u002Fsmall_IMG_5631_d9b99d485a.jpg","small_IMG_5631_d9b99d485a","small_IMG_5631.JPG",19.71,333,19714,{"public_id":209,"resource_type":33},{"ext":199,"url":216,"hash":217,"mime":83,"name":218,"path":15,"size":219,"width":39,"height":47,"sizeInBytes":220,"provider_metadata":221},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1758024414\u002Fmedium_IMG_5631_d9b99d485a.jpg","medium_IMG_5631_d9b99d485a","medium_IMG_5631.JPG",35.5,35499,{"public_id":217,"resource_type":33},{"ext":199,"url":223,"hash":224,"mime":83,"name":225,"path":15,"size":226,"width":227,"height":104,"sizeInBytes":228,"provider_metadata":229},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1758024414\u002Fthumbnail_IMG_5631_d9b99d485a.jpg","thumbnail_IMG_5631_d9b99d485a","thumbnail_IMG_5631.JPG",4.16,104,4156,{"public_id":224,"resource_type":33},"IMG_5631_d9b99d485a",1769.34,"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1758024415\u002FIMG_5631_d9b99d485a.jpg",{"public_id":230,"resource_type":33},"2025-09-16T12:06:55.571Z","of6gbjbn038al5hum1gsxvjc",{"id":237,"Title":238,"Content":239,"Description":240,"Category":10,"slug":241,"createdAt":242,"updatedAt":243,"publishedAt":244,"readTime":68,"documentId":245,"Image":246,"author":285},51,"Discord V2: Mentions, DMs, and Slash Commands","Discord users, this one is for you.\n\n{% youtube id=\"gyTe7D1Ur0A\" \u002F%}\n\nWe've rebuilt our Discord integration from the ground up. The result is an agent that feels like it actually belongs in Discord, rather than one that was stitched on as an afterthought.\n\n## What changed and why\n\nThe original Discord integration (v1) was built around a single slash command. Simple, focused, and it works well for servers that want exactly that: one clean command, minimal permissions, maximum predictability.\n\nBut a lot of you wanted more. You wanted your agent in the flow of conversation, not off to the side waiting to be formally invoked. So we built v2.\n\n## What's new in v2\n\n### Respond to @mentions\n\nYour agent now responds when someone mentions it directly in a channel. A team member can type `@YourAgent what's the returns policy?` mid-conversation and get an answer right there, without switching context or typing a slash command.\n\n### Direct messages\n\nUsers can now DM your agent privately. Useful for sensitive questions, personal account lookups, or extended back-and-forth that would clutter a public channel.\n\n### Slash commands\n\nSlash commands are still here, now with Discord's native autocomplete. Type `\u002Fask` (or whatever you configure) and Discord surfaces it instantly.\n\n### Threaded conversations\n\nV2 keeps conversations organised in threads, so channels stay clean even when your agent is active across multiple conversations at once.\n\n## V1 or V2: which is right for you?\n\nBoth versions are available.\n\nV2 is the better fit for most teams. It brings your agent into the natural flow of conversation, supports multiple interaction types, and uses Discord's full feature set.\n\nV1 is a good fit if your server is privacy-sensitive and you'd prefer the smallest possible permission footprint, or if a single slash command is genuinely all you need.\n\n## Setting it up\n\nSwitching to V2 takes three steps:\n\n1. **Switch to V2** in your Discord channel settings inside Chat Thing\n2. **Enable Message Content Intent** in your Discord agent settings (this allows the agent to read messages it's mentioned in)\n3. **Update the interaction endpoint URL** in your Discord agent settings\n\nNew to Discord? Our setup guide walks you through connecting to a server.\n\n---\n\n**Discord V2 is available now** on all plans with Discord channels.","Our Discord integration just got a major upgrade. Your agent now responds to @mentions, works in DMs, and supports slash commands.","discord-v2-mentions-dms-slash-commands","2026-04-03T14:04:07.209Z","2026-05-07T14:26:21.653Z","2026-05-07T14:06:11.426Z","glglrazs0t6l7v9tyqv73faa",{"id":247,"name":248,"alternativeText":249,"caption":146,"width":20,"height":21,"formats":250,"hash":279,"ext":149,"mime":152,"size":280,"url":281,"previewUrl":15,"provider":61,"provider_metadata":282,"createdAt":283,"updatedAt":283,"documentId":284,"publishedAt":66,"focalPoint":15},836,"day-04-discord-feature-card-v5-1920x1080","Discord V2 Gateway — Day 04",{"large":251,"small":258,"medium":265,"thumbnail":272},{"ext":149,"url":252,"hash":253,"mime":152,"name":254,"path":15,"size":255,"width":30,"height":155,"sizeInBytes":256,"provider_metadata":257},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1778163972\u002Flarge_day_04_discord_feature_card_v5_1920x1080_dd50c4434a.png","large_day_04_discord_feature_card_v5_1920x1080_dd50c4434a","large_day-04-discord-feature-card-v5-1920x1080",487.81,487807,{"public_id":253,"resource_type":33},{"ext":149,"url":259,"hash":260,"mime":152,"name":261,"path":15,"size":262,"width":39,"height":40,"sizeInBytes":263,"provider_metadata":264},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1778163972\u002Fsmall_day_04_discord_feature_card_v5_1920x1080_dd50c4434a.png","small_day_04_discord_feature_card_v5_1920x1080_dd50c4434a","small_day-04-discord-feature-card-v5-1920x1080",135.86,135861,{"public_id":260,"resource_type":33},{"ext":149,"url":266,"hash":267,"mime":152,"name":268,"path":15,"size":269,"width":47,"height":48,"sizeInBytes":270,"provider_metadata":271},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1778163972\u002Fmedium_day_04_discord_feature_card_v5_1920x1080_dd50c4434a.png","medium_day_04_discord_feature_card_v5_1920x1080_dd50c4434a","medium_day-04-discord-feature-card-v5-1920x1080",283.55,283550,{"public_id":267,"resource_type":33},{"ext":149,"url":273,"hash":274,"mime":152,"name":275,"path":15,"size":276,"width":55,"height":56,"sizeInBytes":277,"provider_metadata":278},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1778163972\u002Fthumbnail_day_04_discord_feature_card_v5_1920x1080_dd50c4434a.png","thumbnail_day_04_discord_feature_card_v5_1920x1080_dd50c4434a","thumbnail_day-04-discord-feature-card-v5-1920x1080",41.8,41800,{"public_id":274,"resource_type":33},"day_04_discord_feature_card_v5_1920x1080_dd50c4434a",384.64,"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1778163972\u002Fday_04_discord_feature_card_v5_1920x1080_dd50c4434a.png",{"public_id":279,"resource_type":33},"2026-05-07T14:26:13.064Z","f703sw9jn2bui9nanspjh57q",{"id":286,"Name":287,"createdAt":288,"updatedAt":289,"publishedAt":290,"documentId":291,"Image":292},11,"Chris","2024-09-26T14:45:01.162Z","2026-05-19T11:51:46.202Z","2026-05-19T11:51:46.263Z","lm3bpo0a4f8dlhl99y0bztzn",{"id":293,"name":294,"alternativeText":15,"caption":15,"width":295,"height":295,"formats":296,"hash":311,"ext":149,"mime":152,"size":312,"url":313,"previewUrl":15,"provider":61,"provider_metadata":314,"createdAt":315,"updatedAt":315,"documentId":316,"publishedAt":315,"focalPoint":15},841,"avatar.png",512,{"small":297,"thumbnail":304},{"ext":149,"url":298,"hash":299,"mime":152,"name":300,"path":15,"size":301,"width":39,"height":39,"sizeInBytes":302,"provider_metadata":303},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1779191498\u002Fsmall_avatar_d040acbf49.png","small_avatar_d040acbf49","small_avatar.png",243.02,243024,{"public_id":299,"resource_type":33},{"ext":149,"url":305,"hash":306,"mime":152,"name":307,"path":15,"size":308,"width":104,"height":104,"sizeInBytes":309,"provider_metadata":310},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1779191498\u002Fthumbnail_avatar_d040acbf49.png","thumbnail_avatar_d040acbf49","thumbnail_avatar.png",33.15,33147,{"public_id":306,"resource_type":33},"avatar_d040acbf49",66.53,"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1779191498\u002Favatar_d040acbf49.png",{"public_id":311,"resource_type":33},"2026-05-19T11:51:38.933Z","gg7jrmstvpwjzszr0gtzrhnk",{"id":318,"Title":319,"Content":320,"Description":321,"Category":10,"slug":322,"createdAt":323,"updatedAt":324,"publishedAt":325,"readTime":140,"documentId":326,"Image":327,"author":365},59,"Your agent noticed something your team missed, AI analytics are here","While we were testing the new Chat Thing analytics with a real estate customer, the dashboard flagged something nobody on the team had spotted. The same complaint kept coming up in conversation after conversation: for-sale signs were being left up on houses long after the houses had sold, and customers were quietly getting fed up about it.\n\nThe recommendation the dashboard wrote back was as plain as the problem. Take the signs down sooner.\n\nThat isn't the kind of insight you find by staring at a pie chart. It's what a member of staff might pick up if they read every single chat from the past week, and even then only if they were paying close attention.\n\nToday we're shipping the new Analytics dashboard for Chat Thing. It does the reading for you, and it tells you what to do.\n\n{% youtube id=\"1NWRzG6YW5k\" \u002F%}\n\n### What's in it\n\nOpen the Overview tab on any agent, pick a month range, and you get five sections that answer the questions you opened the dashboard to ask.\n\n### Topic Analysis\n\n![Topic Analysis — Top Conversation Topics ranked list and Trending Topics Over Time chart](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1773247246\u002Fdocs\u002Fanalytics\u002Fanalytics-topics.png)\n\nEvery conversation grouped by what it was actually about. Not raw keywords, themes.\n\nTop Conversation Topics gives you a ranked list with counts and percentages. Trending Topics Over Time is a weekly line chart that shows which subjects are climbing and which are falling.\n\nYou can manage up to ten custom topics per agent, rename them, colour-code them, or have Chat Thing generate them from your conversation data.\n\n### Sentiment Analysis\n\n![Sentiment Analysis — Positive \u002F Neutral \u002F Negative donut, trend line, and emotion-detection breakdown](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1773247248\u002Fdocs\u002Fanalytics\u002Fanalytics-sentiment.png)\n\nA donut chart breaks every conversation into Positive, Neutral, or Negative. A trend line shows whether your users are getting happier or more frustrated week by week.\n\nEmotion detection goes one level deeper. How often users are Curious, Satisfied, Frustrated, Confused, and the rest, with percentages for each.\n\n### Engagement Insights\n\n![Engagement Insights — messages per conversation, daily volume, resolution rate, and duration with trend chart](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1773247249\u002Fdocs\u002Fanalytics\u002Fanalytics-engagement.png)\n\nThe conversation quality metrics most teams want, in one place.\n\nAverage messages per conversation. Average daily chat volume. Resolution rate. Average conversation duration. The trend chart plots messages-per-conversation against resolution rate, so you can tell whether longer chats are actually helping anyone.\n\n### Common Questions\n\n![Common Questions — top ten clustered questions with topic tags and ask-counts](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1773247251\u002Fdocs\u002Fanalytics\u002Fanalytics-questions.png)\n\nAn AI clusters every question your users asked into the top ten things they actually wanted to know.\n\nEach cluster has a topic tag, an ask-count, and an expandable view of the individual conversations behind it. This is where you find the article you should write next, the FAQ entry that's missing, the prompt instruction your agent needs.\n\n### Actionable Recommendations\n\n![Actionable Recommendations — ranked suggestions with type, priority, impact, and action steps](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1773247253\u002Fdocs\u002Fanalytics\u002Fanalytics-recommendations.png)\n\nThe dashboard generates specific suggestions ranked by priority. High, Medium, Low, Info.\n\nEach recommendation has a type (Prompt Update, Knowledge Gap, Business Insight, Feature Suggestion, or Success), a description, an impact statement, action steps, and the count of conversations it affects. Some are about your agent. Others are about your business.\n\nThe for-sale-sign story came out of this section. The dashboard noticed the same complaint surfacing across many separate conversations, classified it as a Business Insight, and wrote a one-line action: take the signs down sooner, your customers are noticing.\n\n### It catches what you would miss\n\nThis is the bit we keep coming back to. The dashboard isn't only summarising agent performance. It's listening to what your customers are saying about the business behind the agent, and giving you a quiet nudge when something's off.\n\nA delivery window that's frustrating people. A returns step that nobody can find. A booking flow that loses customers halfway through. A sign that's been up too long.\n\nYou don't need to know the question in advance. The dashboard finds it for you.\n\n### Weekly email digest\n\nIf you'd rather not open the dashboard at all, opt in to a weekly email with the headline metrics, sentiment overview, top questions, AI insights, and links straight back into the dashboard. Configurable per user in Account settings.\n\n### Why we built it this way\n\nThe data was always there. What was missing was someone to read it.\n\nAn LLM can read every conversation an agent had this month, cluster the themes, surface the friction, rank the priorities, and tell you what to do about it. So that's what the dashboard does.\n\n### Getting started\n\nLive today on Pro and Enterprise plans.\n\nOpen any agent, click Overview, pick a month range. Topics, sentiment, engagement metrics, common questions, and recommendations all populate automatically.\n\nTomorrow: if your community runs on Discord, you'll want to be here.\n","Chat Thing Analytics reads every conversation your agent has and tells you what to do. Topic analysis, sentiment, engagement, common questions, and ranked recommendations on every agent overview tab.","launch-week-03-ai-powered-analytics","2026-05-01T13:36:54.654Z","2026-05-06T07:42:18.442Z","2026-05-06T07:42:18.435Z","e3xh09z4zgdvu1rpx1k4n1g5",{"id":328,"name":329,"alternativeText":15,"caption":15,"width":20,"height":21,"formats":330,"hash":359,"ext":149,"mime":152,"size":360,"url":361,"previewUrl":15,"provider":61,"provider_metadata":362,"createdAt":363,"updatedAt":363,"documentId":364,"publishedAt":66,"focalPoint":15},832,"Day 03 Analytics Feature Card.png",{"large":331,"small":338,"medium":345,"thumbnail":352},{"ext":149,"url":332,"hash":333,"mime":152,"name":334,"path":15,"size":335,"width":30,"height":155,"sizeInBytes":336,"provider_metadata":337},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1778052374\u002Flarge_Day_03_Analytics_Feature_Card_df18a577dc.png","large_Day_03_Analytics_Feature_Card_df18a577dc","large_Day 03 Analytics Feature Card.png",592.64,592641,{"public_id":333,"resource_type":33},{"ext":149,"url":339,"hash":340,"mime":152,"name":341,"path":15,"size":342,"width":39,"height":40,"sizeInBytes":343,"provider_metadata":344},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1778052374\u002Fsmall_Day_03_Analytics_Feature_Card_df18a577dc.png","small_Day_03_Analytics_Feature_Card_df18a577dc","small_Day 03 Analytics Feature Card.png",162.23,162227,{"public_id":340,"resource_type":33},{"ext":149,"url":346,"hash":347,"mime":152,"name":348,"path":15,"size":349,"width":47,"height":48,"sizeInBytes":350,"provider_metadata":351},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1778052374\u002Fmedium_Day_03_Analytics_Feature_Card_df18a577dc.png","medium_Day_03_Analytics_Feature_Card_df18a577dc","medium_Day 03 Analytics Feature Card.png",339.16,339164,{"public_id":347,"resource_type":33},{"ext":149,"url":353,"hash":354,"mime":152,"name":355,"path":15,"size":356,"width":55,"height":56,"sizeInBytes":357,"provider_metadata":358},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1778052374\u002Fthumbnail_Day_03_Analytics_Feature_Card_df18a577dc.png","thumbnail_Day_03_Analytics_Feature_Card_df18a577dc","thumbnail_Day 03 Analytics Feature Card.png",47.69,47689,{"public_id":354,"resource_type":33},"Day_03_Analytics_Feature_Card_df18a577dc",595.61,"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1778052374\u002FDay_03_Analytics_Feature_Card_df18a577dc.png",{"public_id":359,"resource_type":33},"2026-05-06T07:26:15.045Z","k4lwd9xpyuyxokquhphazs8x",{"id":127,"Name":366,"createdAt":367,"updatedAt":368,"publishedAt":369,"documentId":370,"Image":371},"Gem","2023-08-25T14:44:02.325Z","2026-04-03T14:53:07.377Z","2023-08-25T14:44:05.581Z","rs8ltnu1mkqi6j2hvepdj8ym",{"id":372,"name":373,"alternativeText":15,"caption":15,"width":374,"height":375,"formats":376,"hash":409,"ext":80,"mime":83,"size":410,"url":411,"previewUrl":15,"provider":61,"provider_metadata":412,"createdAt":413,"updatedAt":413,"documentId":414,"publishedAt":66,"focalPoint":15},732,"me2026.jpg",2288,1978,{"large":377,"small":385,"medium":393,"thumbnail":401},{"ext":80,"url":378,"hash":379,"mime":83,"name":380,"path":15,"size":381,"width":30,"height":382,"sizeInBytes":383,"provider_metadata":384},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1775227982\u002Flarge_me2026_022d4321e3.jpg","large_me2026_022d4321e3","large_me2026.jpg",170.04,865,170040,{"public_id":379,"resource_type":33},{"ext":80,"url":386,"hash":387,"mime":83,"name":388,"path":15,"size":389,"width":39,"height":390,"sizeInBytes":391,"provider_metadata":392},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1775227982\u002Fsmall_me2026_022d4321e3.jpg","small_me2026_022d4321e3","small_me2026.jpg",43.94,432,43943,{"public_id":387,"resource_type":33},{"ext":80,"url":394,"hash":395,"mime":83,"name":396,"path":15,"size":397,"width":47,"height":398,"sizeInBytes":399,"provider_metadata":400},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1775227982\u002Fmedium_me2026_022d4321e3.jpg","medium_me2026_022d4321e3","medium_me2026.jpg",95.7,648,95701,{"public_id":395,"resource_type":33},{"ext":80,"url":402,"hash":403,"mime":83,"name":404,"path":15,"size":405,"width":406,"height":104,"sizeInBytes":407,"provider_metadata":408},"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1775227982\u002Fthumbnail_me2026_022d4321e3.jpg","thumbnail_me2026_022d4321e3","thumbnail_me2026.jpg",6.75,181,6747,{"public_id":403,"resource_type":33},"me2026_022d4321e3",732.7,"https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1775227982\u002Fme2026_022d4321e3.jpg",{"public_id":409,"resource_type":33},"2026-04-03T14:53:02.980Z","fzyymo1uivslt1rep9c83edj",{"pagination":416},{"page":127,"pageSize":140,"pageCount":417,"total":418},8,24,{"prev":420,"next":431},{"id":421,"Title":422,"Content":423,"Description":424,"Category":425,"slug":426,"createdAt":427,"updatedAt":428,"publishedAt":429,"readTime":15,"documentId":430},21,"GPT-4o mini live on Chat Thing","We're excited to announce that Chat Thing has integrated OpenAI's latest model, GPT-4o mini, into our platform! This powerful new AI brings enhanced capabilities and cost-efficiency to your chatbots and AI-driven solutions.\n\nYou can now choose GPT-4o mini from your bot  model settings.\n\n![Xnapper-2024-07-20-14.12.04.png](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1721482456\u002FXnapper_2024_07_20_14_12_04_05d5131d0f.png)\n\nKey benefits of GPT-4o mini on Chat Thing:\n* **Superior intelligence**: Outperforms GPT-3.5 Turbo on various benchmarks, including reasoning, math, and coding tasks.\n* **Multimodal capabilities**: Supports both text and vision inputs, with future expansions planned for video and audio.\n* **Affordable pricing**: Up to 60% cheaper than GPT-3.5 Turbo, making AI more accessible for businesses of all sizes. Its now the cheapest model on Chat Thing meaning your message tokens go further!\n* **Expanded context window**: 128K tokens allow for processing larger amounts of information in a single request.\n* **Improved non-English language handling**: More cost-effective for multilingual applications.\n* **Enhanced safety features**: Built-in protections against jailbreaks and prompt injections.\n\nWhether you're building customer support chatbots, automating workflows, generating content, or developing lead generation tools, GPT-4o mini on Chat Thing offers the perfect balance of performance and affordability.","Chat Thing now integrates GPT-4o mini: Enhance your AI bots with superior intelligence, multimodal capabilities, and cost-efficiency. Upgrade today!","changelog","gpt-4o-mini-live-on-chat-thing","2024-07-20T13:13:21.576Z","2024-07-20T13:36:03.947Z","2024-07-20T13:36:03.938Z","fqbhre3duiekxw61yv5z8z8v",{"id":432,"Title":433,"Content":434,"Description":435,"Category":425,"slug":436,"createdAt":437,"updatedAt":438,"publishedAt":439,"readTime":15,"documentId":440},22,"Build bots with Gemini 1.5 Pro + more updates!","Hey there, AI enthusiasts and bot-building extraordinaires!\n\nBuckle up, because we're about to take you on a wild ride through our latest and greatest [Chat Thing](https:\u002F\u002Fchatthing.ai) updates, but before we start we got an answer to the question we know you have all been asking yourselves! \n\n**Does a 1 million long context window make you good at Chess?** \n\nWell, apparently not! Here's GPT-4o smoking Gemini 1.5 Pro on our silly [LLM chess benchmark](https:\u002F\u002Fllm-battle.chatthing.ai\u002Fgames\u002F11). \n\n![Xnapper Aug 07 2024.webp](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1723062239\u002FXnapper_Aug_07_2024_c352756ec7.webp)\n\nRead more about that later, but now, let's move on to our serious updates!\n\n## 🧠 Gemini 1.5 Models: Because Size (of Context) Matters (just not in Chess)!\n\nRemember when we thought 128k tokens were a lot? Well, hold our coffee, because Google has two absolute units of AI models, and guess what? They're now part of the Chat Thing family!\n\nWhy should you care? Well, imagine asking your bot to analyze \"War and Peace,\" and it says, \"Is that all you've got?\" That's the power we're talking about! They are both strong models and support vision and power-ups!\n\n### 💨 Gemini 1.5 Flash: The Budget-Friendly Brainiac\nThink of Flash as that friend who's always got a witty comeback, but never asks you to split the bill. It's quick, it's smart, and it won't break the bank. Perfect for high traffic bots that don't need to do complex reasoning tasks. \n\n### 🦸‍♂️ Gemini 1.5 Pro: The Superhero of AI\nIf Flash is your witty friend, Pro is like having the entire Avengers team at your disposal. It's not just smart; it's Google's flagship model and works out slightly cheaper than GPT-4o in terms of message tokens. And with a context window of 1 MILLION tokens, it's got a memory that would make an elephant jealous! 🐘\n\nAdding Gemini continues our commitment to integrating all the cutting edge models with Chat Thing so you can use all of them under one roof. Head over to the docs to view all our [available models](https:\u002F\u002Fchatthing.ai\u002Fdocs\u002Favailable-models).\n\nYou can now select the Gemini model from your bot's \"General settings\" within the \"Model settings\" section.\n\n![Xnapper Aug 07.webp](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1723063829\u002FXnapper_Aug_07_aec429ba3b.webp)\n\n## 📝 Conversation Summaries: Because Who Has Time to Read Everything?\n![Xnapper Aug 07 (3).webp](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1723064245\u002FXnapper_Aug_07_3_1baebdff2a.webp)\nWe heard you loud and clear: \"Reading is hard!\" (Okay, maybe that's not exactly what you said, but close enough). So, we've introduced Conversation Summaries!\nNow you can get the TL;DR of your bot convos right in the sidebar. It's like having a personal assistant for your AI assistant. How meta is that? 🤯\nAnd psst... we're cooking up something even cooler: email summaries! Soon, you'll be able to catch up on your bot's social life over your morning coffee. ☕\n\nYou can now enable this from your bot general settings:\n![Xnapper Aug 07 (1).webp](https:\u002F\u002Fres.cloudinary.com\u002Fdjyjvrw5u\u002Fimage\u002Fupload\u002Fv1723064286\u002FXnapper_Aug_07_1_f5cd6bbd41.webp)\n\n## 🏆 GPT-4o mini: The New Sheriff in Town\nMove over, GPT-3.5! There's a new default model in town, and its name is GPT-4o mini. \nIf you're still hanging out with GPT-3.5, it's time for an upgrade. Your bot will thank you (probably with a much wittier response).\n\n##♟️ LLM Chess Battle: May the Best Bot Win!\nEver wondered what would happen if we put AI models in a chess ring and let them duke it out? Well, wonder no more! Head over to [https:\u002F\u002Fllm-battle.chatthing.ai](https:\u002F\u002Fllm-battle.chatthing.ai) and watch the GPU-powered showdown live!\n\nWhy have we built this? Well we think its the most fun LLM benchmark around even if its not the most useful. As it turns our current LLM models are pretty rubbish at chess in comparison to humans, but even still the flagship models seem to be better than the older models so it will be interesting to see how this changes as newer models are released. \n\nCurrent reigning champ: GPT-4o! It's been serving checkmates like they're going out of style, even taking down the mighty Gemini 1.5 Pro. Will its reign last? Grab some popcorn and find out! 🍿\n\n## Wrapping It Up with a Bow (Because We're Fancy Like That)\n\nPhew! That was a lot, wasn't it? But wait, there's more! (Just kidding, that's all for now). We're constantly cooking up new ways to make Chat Thing the most awesome AI playground out there.\n\nHave you taken these new features for a spin yet? Did Gemini 1.5 Pro write you a sonnet about your cat? Did GPT-4o mini finally understand your sarcasm? We want to hear all about it!\n\nDrop us a line, send a carrier pigeon, or use that fancy AI bot you just built to let us know how these updates are rocking your world.\nStay cool, stay curious, and keep on bot-building, you magnificent AI tamers!\n\nCatch you on the flippity-flip! ✌️","Gemini 1.5 models with 1M tokens, conversation summaries, GPT-4o mini as default, and watch AI chess battles live! Upgrade your bots now!","build-bots-with-gemini-1-5-pro-more-updates","2024-08-07T20:26:38.464Z","2024-08-07T21:08:51.852Z","2024-08-07T21:08:51.845Z","jbeayfkrgru1z3nedk1x2v68"]