Getting started
How to use Storytell
Other Storytell products
About Storytell
How we manage Storytell user dataTroubleshooting
“SmartChat™”
Chat, select your tags and groups, or switch between your knowledge and world's knowledge.
Differentiating between "World’s Knowledge" and "All my Knowledge"
- World's Knowledge - you are chatting directly with Large Language Models (LLMs) without interrogating any content you've uploaded. When you ask a question or make a request, the response you get is based on all the data the LLM has been trained on. In the case of Open AI's GPT-3.5 and GPT-4 Omni, that's all of Wikipedia and much of the internet.
- All My Knowledge - you're chatting with all the knowledge that's been uploaded to your personal profile, or, if you're in org view, all the knowledge that's been uploaded to your entire organization and made accessible to you.
Tags & Groups
What about tags? Want to chat with a specific document? A tag? A group? Learn more about how to do that in this post.
Unselecting an active tag or group
In order to unselect or remove a tag or a group in a chat, all you need to do is to click on the X icon next to the Tag or Group under the chat section in your SmartChat™ dashboard like so:
Large Language Models (LLM)
Choose the best model (LLM) for you with SmartChat™
When you're chatting with the "World's Knowledge," you get to choose among several Large Language Models (LLMs). You can think of an LLM like a brain -- each brain has been trained differently, so they'll give you different results.
When you chat with the knowledge you uploaded, free subscribers have the option to choose between Claude-3.0 Haiku or GPT-3.5 Turbo with Haiku being the default option.
Pro subscribers, however, benefit from the enhanced capabilities of GPT-4 Omni, our most advanced and precise model to date.
Team subscribers get access to GPT-4 Omni, Claude-3.0 Opus and Claude-3.0 Sonnet.
How to choose an LLM:
- Go to SmartChat™
- Select which LLM model you’d like to use from the options found below the chat input.
- The availability of which LLM you can choose is based on your subscription tier. See the table below to see which LLM is available for you
GPT-4o Mini | Claude-0.3 Haiku | GPT-4o | Claude-3.0 Opus | Claude-3.5 Sonnet | |
Free Plan | ✅ | ✅ | ➖ | ➖ | ➖ |
Pro Plan | ✅ | ✅ | ✅ | ✅ | ✅ |
Team Plan | ✅ | ✅ | ✅ | ✅ | ✅ |
Enterprise users get a custom plan which includes:
- Many LLMs
- Model switching
- Fine-tuned models
- Single-tenant
- SSO Integration
- Projects
- RBAC
💡You can learn more about these different account tiers and upgrade your account by visiting https://storytell.ai/pricing
Unsure about which LLM is for you? Here's a brief rundown:
- GPT-4o Mini: This is an older model, developed by OpenAI. It's known for being faster and cheaper than GPT-4 with the tradeoff that the quality of responses are often more on the level of a high school student.
- GPT-4 Omni: Is built to give clearer and more accurate responses, even to complicated questions. It's better at understanding slight differences in meaning and can adjust to meet the particular needs of users or different businesses.
- Claude: This model was developed by Anthropic to communicate in a more natural, human-like manner. It may offer better output in the more creative fields.
- Claude-3.0 Haiku - a tailored iteration of the Claude-3.0 language model lineup, developed to capture the essence and brevity of haiku poetry. This model is adept at crafting short, evocative poems that follow the traditional 5-7-5 syllable structure and embody the simplicity and depth of imagery that haikus are known for. It is particularly useful for generating poetic content that conveys a sense of immediacy and natural beauty.
- Claude-3.0 Sonnet - a variant of the Claude-3.0 language model series, specifically fine-tuned for generating text that adheres to the structural and thematic elements of a sonnet. This model is designed to understand and produce the intricate rhyme schemes and iambic pentameter that characterize traditional sonnets, making it an ideal tool for creating poetic content with a classic flair.
- Claude-3.5 Opus - a specialized version of the Claude-3.0 language model family, optimized for composing longer and more complex pieces of writing, such as essays, research papers, or extensive reports. The Opus model is equipped with advanced capabilities to maintain coherence over extended passages, manage intricate subject matter, and adhere to sophisticated stylistic and formatting guidelines.
Note: There is a character limit of up to 6,000 characters when using chat
How to copy a chat in SmartChat™
Like the response you're getting from SmartChat™? Hurrah!
Want to copy the response and paste it elsewhere? The best way to do that is to use our copy icon because it retains the formatting but not the footnotes.
Word to the wise: If you highlight the chat and copy it manually, the footnotes will remain (and you may lose the formatting).
Here's how text copied using our "copy icon" will appear when you paste it elsewhere:
Want to help us improve our algorithm and our accuracy? Give a solid response a thumbs up.
How to share SmartChat™
After you've:
- Created a Storytell account
- Uploaded content directly into SmartChat™ or via our Chrome extension
- Made some content accessible to folks in your org
- Created a tag or a group
It's time to share your SmartChat™ ! In fact, you can do this at any point after steps 1 & 2.
Share with your org
- Click the kebab icon on the upper right of the chat and click on Share SmartChat™ button.
- A modal will pop up. You can either share the qr code or share the direct link to the smartchat
Your coworker will need to:
- Create a Storytell account using the org domain
- Verify their email
- Then they can dive right in!
This is what a user will see once they’ve opened your shared SmartChat™:
Share with the world
Any tag that has content labeled "Anyone with the link" can be shared with the world. Once they create a Storytell account, they'll be able to chat with it, too.
Rate Storytell's output -- here's how
Yes, it's true! You can rate every single chat response from Storytell.
Why would you want to do that? Because we track your feedback and use it to improve our algorithm.
Rating Storytell chat responses is straightforward -- thumbs up or thumbs down:
- Hover your cursor over the chat response
- You'll see thumbs-up and thumbs-down buttons. Click the one that applies!
- You'll get a message in the line saying "Thanks for the feedback," which means we've received your rating.
Let's dive into the options
4a. Thumbs-up: the response was helpful!
- Want to say more? Select "Share More Feedback" to elaborate. This will open up an email and send it to feedback@Storytell.ai
4b. Thumbs-down: response was not helpful.
- Choose one of the options below to share why (was it irrelevant? didn't make sense? inaccurate?)
- You can always select "Share More Feedback" to elaborate.
What are Story Tile™ footnotes and how do they work?
Our footnotes, which links to Story Tiles™, are clickable numerical links following each sentence in chat.
The purpose of Story Tile™ footnotes is to build trust. When you click on them, they show you the original content our algorithm and vector database used in its response.
Why does it matter?
This solves the "black box" issue of AI, where we're often wondering how the Large Language Model (LLM) came up with its response. Unlike many other options, Storytell answers that question through its use of footnotes and Story Tiles™.
When do they appear?
Footnotes only appear when you're chatting with knowledge that's been uploaded to SmartChat™.
Can you make them go away?
You certainly can! Just click on the # button to make them invisible.
Enhance readability of SmartChat™ responses using Markdown mode
Markdown mode in SmartChat™ allows you to view responses with improved formatting, such as tables and structured text. To enable Markdown mode, follow these steps:
- Move your cursor over the chat response and a
#icon will appear. Click on this icon to switch the response to Markdown mode, which will present the content in your desired format, whether it's a table or other structured text. - Alternatively, click on the three dots located at the top-right corner of the chatbox to access the chat menu. Here, you can toggle the Markdown mode on. Once activated, all responses will be displayed with Markdown formatting applied.
- “SmartChat™”
- Differentiating between "World’s Knowledge" and "All my Knowledge"
- Tags & Groups
- Unselecting an active tag or group
- Large Language Models (LLM)
- Choose the best model (LLM) for you with SmartChat™
- How to choose an LLM:
- Unsure about which LLM is for you? Here's a brief rundown:
- How to copy a chat in SmartChat™
- How to share SmartChat™
- Rate Storytell's output -- here's how
- What are Story Tile™ footnotes and how do they work?
- Enhance readability of SmartChat™ responses using Markdown mode