Introducing the latest update to CustomGPT, our versatile ChatGPT-4 platform that let’s you build your own ChatGPT with all your business content – now armed with Realtime Data Indexing.
Yes, you read that right! Now your business can feed fresh-off-the-press content to your chatbot and get insights almost instantly. From hot market trends to the latest podcast episodes, your ChatGPT chatbot is now more up-to-date than ever.
Intrigued? Stick around as we dive into this real-time revolution and how it can take your business conversations to the next level. Let’s go!
The Power and Potential of Realtime Data Indexing
This is about embracing immediacy and agility in your business. With Realtime Data Indexing, you can instantly update your CustomGPT chatbot with the latest information relevant to your operations, keeping you at the cutting edge of your industry.
Imagine an event or data source (like a customer service form submission or a request for a quote) coming in to your business and then becoming instantly searchable via ChatGPT.
Don’t just keep up, outpace your competitors! With up-to-the-minute data at your fingertips, you can respond to customer queries, make informed decisions, and develop strategies based on the very latest information.
And best of all : This is all accessible via the CustomGPT API. So you can integrate it directly into your current workflows.
This feature isn’t limited to a single type of content. Whether it’s websites, helpdesks, YouTube videos, or audio podcasts, you can make your CustomGPT chatbot your business’s ultimate source of truth, ready to tackle queries with real-time knowledge.
Not just us, but our early beta adopters are raving about this feature too! It’s set to be a game-changer, revolutionizing the way businesses interact with their AI chatbots.
How It Works: The Magic Behind Realtime Data Indexing
Let’s unravel the magic that powers our Realtime Data Indexing feature. It all begins when you upload your files through our user-friendly no-code app or via our API.
As soon as you upload your files, our system swiftly places them into our multi-queue, multi-worker processing system. This system is designed to manage a high volume of files simultaneously, allowing you to ingest hundreds or even thousands of files at a go.
Once in the system, the files are instantly parsed and processed. They are then converted into “embeddings” using our robust large-scale processing platform. This process transforms your files into a format that can be easily analyzed and referenced.
The vector embeddings are then stored in a top-tier, commercial class vector database called Pinecone. And voila! Your files are now instantly searchable with ChatGPT.
The Result
All of this happens within an impressive average time of 36 seconds. So, in barely more than half a minute, you can ask ChatGPT questions about your newly ingested documents and be assured that the responses will be grounded in the content of those documents.
That’s the magic of Realtime Data Indexing! Your very own, always-up-to-date AI chatbot ready to provide insights based on the very latest data at a moment’s notice.
Using CustomGPT With Realtime Data Indexing: Step-By-Step Guide
Getting started with Realtime Data Indexing on CustomGPT is simple. We offer two options to get your files uploaded and processed.
The Upload Wizard is a user-friendly tool that allows you to manually upload your files. Here’s how:
1. Login to your account on app.customgpt.ai.
2. From your Project Settings, navigate to Data -> Upload -> Upload Files.
3. Click on “Add Files” to upload your documents.
With this option, you can upload up to 50 files at a time, and each file can be up to 100MB in size.
For advanced users looking to integrate real-time indexing into their workflows, our API provides a seamless solution. You can push files directly into the chatbot and they will be instantly indexed. This feature is included in all our paid plans.
For examples on how to use our API, refer to the Python cookbook at https://github.com/Poll-The-People/customgpt-cookbook .
36 seconds. Over the month of June 2023, the average processing time (over 40,000+ documents) was 36 seconds.
Yes. For example, in June 2023, if you uploaded a PDF on the platform, it was instantly searchable as part of your large document corpus in less than 36 seconds.
So you could technically upload hundreds of documents and have them be instantly searchable with ChatGPT.
No. We cannot guarantee how fast/slow your website is. So if you ingest your website pages or Youtube videos via a Sitemap, the indexing will be quite fast — but might be constrained due to the crawling speed of your website.
Just to be clear: While file upload is faster, website scraping could be minutes (or more) depending on the speed of your website (not our systems). We intentially slow down the crawler so as not to crash your website.
We have implemented a multi-queue system that will allocate different customers to different queues. So the load from other customers should usually not affect your indexing.
Yes. Customers on the Premium plan have access to our full indexing capacity. This significantly reduces the chances of being backed up by other customers in the queue.
6. Do you have an API I can use?
Yes. An API is included with all paid plans. You can also see the cookbook for examples of using the API or the Python SDK.
7. What is CustomGPT and how does it leverage ChatGPT-4?
CustomGPT is an AI chatbot platform that we’ve built on top of OpenAI’s ChatGPT-4. It ingests your business content and uses that data to answer questions. The system can process data from various sources including websites, helpdesks, YouTube videos, and audio podcasts, transforming it into usable data for ChatGPT.
Our multi-source data integration feature allows businesses to create a ChatGPT chatbot with their own data quickly and easily. It can ingest data from multiple sources, making it a versatile tool that adapts to your business’s unique data.
The responses provided by ChatGPT are bounded by a rock-solid context boundary so that ChatGPT does not hallucinate (aka: make up facts). This ensures that ChatGPT does not create responses outside of the provided business content. In other words, it will not talk about competitors or make up facts that aren’t present in your data.
Realtime data indexing is a powerful new feature that allows the system to ingest new information in real-time. For example, if a stock market story breaks at 4:00 PM, you can ingest it into your chatbot using the CustomGPT API and query it with ChatGPT within seconds. This feature brings the power of ChatGPT to your business’s real-time events and data.
When you upload your files through our no-code app or API, they are immediately queued in our multi-worker processing system. These files are then parsed, processed, and vectorized into embeddings. These embeddings are stored in a vector database and become instantly searchable with ChatGPT. This entire process occurs in less than 36 seconds on average.
So, we leave you with this thought: How could Realtime Data Indexing elevate your business? What untapped potential could it help you to uncover? How might it empower your teams to work more efficiently and to make better, more informed decisions?
