Having a great writing partner can be a game-changer. That’s why we set out to harness the latest AI technology. Not
to replace writers, but to empower them. With NovelForge, you stay in control, using AI as a tool to spark ideas,
refine prose, and bring stories to life.
AI Assistants for you, not the other way around. They help you write better, faster, and more creatively.
Overcome Writer’s Block
Stuck on a scene or can’t find the right
words? AI Chat Bot can provide suggestions,
alternative phrasings, or even generate ideas
to keep your writing flowing.
Brainstorm Unique Plot Ideas
NovelForge AI Assistant can help develop
plot twists, character arcs, and unexpected
developments, making it easier to craft
engaging stories.
Edit and Polish
Get instant feedback on grammar, structure,
and clarity, making self-editing smoother and
more efficient.
Chatbot - not all chatbots are the same
Oh, you’re in for a treat. Sure, there are plenty of powerful AI tools out there, but most of them limit how much control you actually have. If an
AI gives you an unsatisfactory answer, or just flat-out refuses, you’re usually stuck repeating your request, hoping for a better response.
With NovelForge, we’ve built something different: Free-Form AI Chat. Here, you’re in control—you can edit anything, including what the
Assistant wrote! No more wrestling with prompts or retrying responses. Just tweak, refine, and let the AI adapt seamlessly.
Why Would You Need That?
Because when you edit the Assistant’s
response, whether it’s tweaking a word,
rephrasing a sentence, or even deleting
an entire passage, the AI believes it
wrote it that way!
This way, the next AI response naturally
adapts to your style, preferences, and
intent, making the Assistant truly work
for you, not the other way around. Your
corrections seamlessly refine future
responses without unnecessary back-
and-forth chit-chat.
Here are the unique features
•
It can run 100% locally, but if you don’t have the necessary hardware, you can use cloud servers like
openrouter while still keeping everything saved locally.
•
Every conversation can have its own document, and you can keep them indefinitely as they are always
storred on your own computer not on cloud
•
You can modify anything. The AI’s responses, your own previous messages, and even the entire
conversation history making AI think it wrote it itself.
•
Need to explore a character’s personality, dialogue style, or motivations? Just start a roleplay, and let
them respond in their own voice.
•
Want to see how your AI assistant would handle being in your shoes? Reverse roles and experience the
conversation from a new perspective.
•
Swap User and Assistant responses. Want the AI to learn your writing style? Write a paragraph in your
own voice, let the assistant rewrite it, then swap the messages - making the AI think it wrote your
original text. This helps the assistant adapt to your unique style, making its responses feel more
natural and in tune with your writing.
100% local:
•
WebUI,
•
Ollama
•
LM Studio
Support
Cloud-based with local save
•
OpenRouter,
•
OpenAi
•
Anything else that supports openAI API such as togetherai, featherless etc..
A full page of step-by-step instructions is available for setting up local or cloud AI without any prior knowledge.
Unlike standard AI chats, NovelForge lets you edit AI-generated text
directly. When you modify the AI’s response, the system treats those
changes as if they were part of the original output. AI simply assumes
it wrote them itself.
Example:
You ask: "What happens when the detective enters the warehouse?"
AI responds: "The detective finds an empty room, save for a single
flickering lightbulb."
You edit the response to add: "and a bloodstained note pinned to the
wall."
Now, the AI continues the conversation as if it had mentioned the
note itself!
In future responses, it will naturally reference the note without
needing additional prompts, something you wouldn’t get in a typical
AI chat.
Free-Form AI Chat and how it works
For your personal notes only, the AI ignores anything written in a comment.
Defines how the AI should behave, setting up its "roleplay" and personality.
Marks the text you submit.
Represents the AI’s response.
Activates roleplay mode, allowing both you and the AI to have character names. This is useful for example for
chatting with your novel's characters or roleplay. While this tag follows a specific syntax, you can also set it up
easily using the Roleplay button below the chat window.
Embedding can insert conversations or large system message without cluttering the Chat Window. A great place to
put an example of conversation exchange.
With Documents, you can link other documents from your project, such as Characters, Settings, or Manuscript
pages. You can even reference other chats you may have had.
The Tags
You'll notice tags in the chat window. These can be added manually or automatically by the AI. To add a tag, right-click anywhere in the
chat window, select Add Tag, and choose the appropriate tag. You can also copy and paste tags as needed.
Writing your message in the response window
Normally, you type your message in the bottom message box, just like in a regular chat. However, if you need more space, you can also
type directly in the top response window, provided that:
•
Your message starts with the User tag.
•
It is the last block, with no other tags following it.
•
The bottom message box is empty.
Reverse Roles
You can easily reverse messages, making it appear as if the AI wrote what you originally typed and vice versa. Suddenly, the AI is the
one asking you questions, reacting to what it now believes it wrote.
This opens up creative possibilities. Turn the AI into an interviewer, a curious student, or even an in-world character engaging with you
dynamically. By flipping roles, you can guide the conversation in unexpected directions. And then just like that flip back at any time.
Swap messages while keepig the roles
Here, you can swap messages between Assistant and User without changing their roles. At first, this might seem like a pointelss trick, but
it’s actually a powerful way to train the AI to match your writing style.
For example, you can have the AI rewrite a few paragraphs from your book, then swap the messages. Now, instead of your text being
rewritten in the AI’s voice, it’s the AI’s response being rewritten in your voice. This conditions the AI to naturally mimic your writing style,
making its suggestions feel more like your own.
Character Roleplay
With Roleplay Mode, you can interact with the AI as if it were one of your characters, making it a powerful tool for refining dialogue, testing
character interactions, and deepening personality development.
Roleplay mode can be riggered by using
tag with name of the characters using this syntax:
{User = Elizabeth}, {Assistant = Jane}
Or simply use Roleplay Button and fill the character names.
To get started, simply provide a character description in the SYSTEM prompt, detailing their background, personality, and speech style.
Then, include a few sample dialogue exchanges between two characters, alternating lines to showcase their unique way of speaking.
When implemented correctly, the AI will adopt not just
surface-level speech patterns but deeper elements of
the character's persona, creating responses that feel
genuinely authentic rather than merely imitative.
The true measure of success comes when the AI
surprises you with responses that are perfectly in
character yet novel, revealing new dimensions of your
character you hadn't explicitly defined but that
emerge naturally from the personality foundation
you've built.
Optimizing Character Roleplay with Embeddings
Rather than cluttering your main chat window
with numerous examples, you can utilize the
Embedding feature to efficiently organize and
store your character dialogue samples.
Embeddings act like containers, neatly tucking
away example interactions while still keeping
them readily accessible to the AI.
This allows the AI to recall character traits,
speech patterns, and personalities without
overwhelming your chat history, keeping your
workspace clean while maintaining depth and
consistency in roleplay.
As a bonus, you can save these example
clippings for future use in other chats.
Creating Authentic Character Roleplay with AI
To create truly immersive and authentic character roleplay experiences with AI, you need to focus on two key elements:
1.Character Definition
You can link one or more of your characters from the Character cards to the AI Chat using the Document Function feature.
When setting up your character, include:
•
Core personality traits and values
•
Speech patterns (formal/informal, verbose/concise, dialect)
•
Typical mannerisms and reactions
•
Background information that shapes their worldview
•
Emotional tendencies and triggers
For example, instead of just "John is a detective," try "John is a world-weary detective with a dry sense of humor who speaks in short,
clipped sentences and often references obscure film noir movies."
2. Dialogue Examples
Provide multiple conversation examples showing:
•
How your character responds to different emotional situations
•
Their unique phrases and verbal tics
•
How they handle conflict versus casual conversation
•
Their typical thought process
Quality matters more than quantity. A few deeply characteristic examples are better than many generic ones.
NovelForge AI Assistants – Fully Local or Cloud-Based with Local Save
Linking to your documents
You can dynamically connecting relevant
documents from the project to the chat.
Currently these documents are supported:
•
Character Cards
•
Settings (places/lore)
•
Manuscripts
•
Notes
•
Other Chats
By linking these documents, the AI will get to see
it and can respond accordingly, without the need
to copy and paste text to the chat.
1
Direct editing of AI response eliminates back-and-forth corrections and gives you immediate control over narrative direction.
It had to be done by someone: renaming hundreds of files as easily as editing lines in a
text editor, with powerful column and block selection, keyboard macros and more.
Dynamic Auto Painter is the most advanced algorithmic software to automatically repaint
photos in the style of famous world masters such as Van Gogh, Monet, Benson, Corot,
Cézanne and many others.
Dramatically increase resolutions and details of your photos, art and 3D graphics using
artificial intelligence.
With just a few clicks, you can elevate any previously made painting with Dynamic Auto
Painter to gallery print-worthy quality.
Editing Suite with 50+ AI assistants
From essential tools like grammar checking
and paraphrasing to advanced features like
dialect conversion or tone adjustment,
NovelForge has it all.
AI can assist in two key areas. First, AI Assistants act as intelligent agents, specializing in tasks like grammar checking, rewriting, and
paraphrasing your text with precision. Second, AI chatbots serve as creative sounding boards, helping you refine and expand your ideas
through interactive dialogue.
AI Assistants
We've added 50+ AI assistants to help with everything from grammar checking and paraphrasing to precise dialect refinement—even by
decade! But that's not all—you can create your own custom assistants tailored to your unique needs!
Context Length
As you start feeding your previous documents to Chat or Assistants, you may soon hit the context length limit. When this happens, the
server may either return an error or simply refuse to process your request without any errors.
Here is a typical error message returned by LM Studio when the context length limit is reached:
There are two parts to increasing the context length:
1. CTX Length in NovelForge (under the CTX button)
This acts as a soft ceiling. Messages exceeding this limit will be cropped before being sent to the LLM.
2. Context Length in the Server (the most important factor which is responsible for this error)
Many newer models support large context lengths, but to improve speed and reduce memory usage, most local servers impose a lower default
limit. For example, in LM Studio, models like LLaMA 3.1 can support up to 130K tokens, but LM Studio caps the context at 4K by default.
(To see this limit in LM Studio, go to the Developer Options, then click on the loaded model)
Now, let's look at the bottom part of the NovelForge Manuscript editor. You’ll see the context length (budget) of the entire text in tokens.
Depending on the Assistant, it may either use only the selected text or include previous or full text. This is indicated in the Assistant with
colored tags.
If <prev> or <full> is shown, it means that either previous text (from the cursor or selection) or the full text of the editor window will be
submitted to the LLM.
Here’s the problem: if NovelForge indicates that your document needs 24K tokens budget, you need to:
•
Increase the token crop in the NovelForge CTX button to ensure the text won’t be cropped.
•
Increase the Context Length in the local server to allow it to process that length and not bail out
However, it’s not a good idea to increase the context length on the server to the full model capabilities, as it will slow down every exchange.
You shouldn’t expect local models to handle such large contexts efficiently (e.g., 130K tokens).
In my case, since my context length is around 24K (about 60 pages of text, which is like a mini novella), I can set the CTX in NovelForge to
32K (this is only a soft text cropping and won’t affect memory usage) and set the local server’s Context Length in LM Studio to 26K, which
should be sufficient.
It’s not advisable to push the context too far. If you really need the LLM to understand all your previous story, it’s better to summarize
the chapters first and then place the summary at the top of the current chapter you’re working on. (In the future, NovelForge will add a
summary place for each document, so it can pull all the summaries, but that’s not available yet.)
This mainly applies to your local server installation, as different servers have different methods for setting the context length.
If you’re using OpenRouter or another OpenAI cloud server, there wouldn’t be such hard ceiling in place, and you can likely use the maximum
context the model supports. In this case, just increase the CTX in NovelForge so it doesn’t crop the text, and you should be fine.
Ollama
Ollama has no GUI settings, and by default, it sets the context length to 2K for every model to prioritize speed. If you push beyond this, it
doesn’t even return an error, just refuse to respod.
To handle this, use the second option in the CTX NovelForge settings.
This will clone the current Ollama model using a "modelfile" and set the context length to your desired value. It doesn't duplicate the model
itself (no additional disk space is used) but configures the model loading settings for the new context length.
You can create as many modelfiles as needed and select the one that fits your required context length.
WebUI
WebUI doesn’t load models dynamically, so you need to manually load the model you plan to use in the Model tab before use. Depending on the
model loader (such as Transformers, Llamacpp, or ExLLAMA), you must set the context length before loading the model.
Some of our other funky projects
