What Are the Best AI Tools for Budget Book Writing and Research?

B

Brachio

New member
I am just starting to use AI from scratch, so had no idea how it works/searches, interacts and I had no idea it has limits, e.g. on how much I can use it.

I began writing a book in both the free Google Gemini and the free ChatGPT.

With Gemini, I gave it the plan in great detail and we worked together and I critiqued it’s information over a few hours. I was happy with how my book was progressing and overwhelmed by the ability of AI.

Then it seemed to no longer take my questions and said ‘page unresponsive, wait or exit page’. This may be due to slow/hung PC (will wait to see, but other programs generally work) but not sure, and fear exiting page as it will likely lose all data/memory.

I asked Google if Gemini saves our conversations and it said no, yet I was never advised to save our communications. I could have lost a lot of work. I then copied out our whole conversation log (it was 900,000 characters/270 MS word pages long!) and I hope I can potentially use this if I need to recover and carry on from where we left off, but copying so much or even getting it to read a small (400Kb) but wordy/lengthy MS doc may be too much for it and breaking it down into dozens of pages would use up all my chat limits. I also do not have the sources from its work, which I need for the book as references. Going through my word doc would be very time consuming manually, to save just completed chapters it showed me in the chat and I would still not have the sources. So what can I practically do to recover things?

Looking at ChatGPT, it seemed even better and provided more comprehensive details, asking less questions/taking more initiative. Then it reached its limit of use (it said carry on in 24 hours). Saved that chat log too. Much shorter in case of problems! A tip may be to copy key info when presented.

Questions:

With a major focus on writing books or related, plus creating images/charts and being on a tight budget, (no coding but lots of searching academic and other sources to research or gather or combine information) what are the options/best plans (preferably free but would consider paid)? I do not like the token models as it’s harder to keep track of and with chatting and writing so much data, it might be more problematic/expensive. I would also consider a separate/dedicated AI programme/app for image creations, even if paid. The $200 ChatGPT pro is not an option, certainly until/unless what I can generate financially justifies the features it offers.

With regard to the limitations of Google Gemini, (number of chats I can have, size of document or words it can handle per conversation), can I use/paste or upload my long doc in for it to read and carry on from where we left off?

With both Gemini and ChatGPT plus, do they appear to suit my plans and are they so severely limited in only taking a few dozen chats/questions?

[DeepSeek seems to have only the token plans, so I don’t like that and it could be expensive the way I use AI 24/7, but happy to be informed otherwise.]

Are there other comparable options? Should I spend some time trialling them first? Don’t want to spend long on this if so.

Any key advice or further comments?

Very many thanks in advance and sorry so long.
 
@Brachio excellent question. My experience with AI context, is what you need a system that can truncate the context (context = your already written pages). Keep it small and summarized. Otherwise AI will loose track of things. So for every prompt you do, you need to have a summary of your book submitted first. Characters, plot, mission of the new chapter and more. Keep the chapters relative short, so AI can keep track of everything.

So would experiment with something like this if writing a book. Hopefully someone will invent a system for this later on
 
Brachio said:
I am just starting to use AI from scratch, so had no idea how it works/searches, interacts and I had no idea it has limits, e.g. on how much I can use it.

I began writing a book in both the free Google Gemini and the free ChatGPT.

With Gemini, I gave it the plan in great detail and we worked together and I critiqued it’s information over a few hours. I was happy with how my book was progressing and overwhelmed by the ability of AI.

Then it seemed to no longer take my questions and said ‘page unresponsive, wait or exit page’. This may be due to slow/hung PC (will wait to see, but other programs generally work) but not sure, and fear exiting page as it will likely lose all data/memory.

I asked Google if Gemini saves our conversations and it said no, yet I was never advised to save our communications. I could have lost a lot of work. I then copied out our whole conversation log (it was 900,000 characters/270 MS word pages long!) and I hope I can potentially use this if I need to recover and carry on from where we left off, but copying so much or even getting it to read a small (400Kb) but wordy/lengthy MS doc may be too much for it and breaking it down into dozens of pages would use up all my chat limits. I also do not have the sources from its work, which I need for the book as references. Going through my word doc would be very time consuming manually, to save just completed chapters it showed me in the chat and I would still not have the sources. So what can I practically do to recover things?

Looking at ChatGPT, it seemed even better and provided more comprehensive details, asking less questions/taking more initiative. Then it reached its limit of use (it said carry on in 24 hours). Saved that chat log too. Much shorter in case of problems! A tip may be to copy key info when presented.

Questions:

With a major focus on writing books or related, plus creating images/charts and being on a tight budget, (no coding but lots of searching academic and other sources to research or gather or combine information) what are the options/best plans (preferably free but would consider paid)? I do not like the token models as it’s harder to keep track of and with chatting and writing so much data, it might be more problematic/expensive. I would also consider a separate/dedicated AI programme/app for image creations, even if paid. The $200 ChatGPT pro is not an option, certainly until/unless what I can generate financially justifies the features it offers.

With regard to the limitations of Google Gemini, (number of chats I can have, size of document or words it can handle per conversation), can I use/paste or upload my long doc in for it to read and carry on from where we left off?

With both Gemini and ChatGPT plus, do they appear to suit my plans and are they so severely limited in only taking a few dozen chats/questions?

[DeepSeek seems to have only the token plans, so I don’t like that and it could be expensive the way I use AI 24/7, but happy to be informed otherwise.]

Are there other comparable options? Should I spend some time trialling them first? Don’t want to spend long on this if so.

Any key advice or further comments?

Very many thanks in advance and sorry so long.
Click to expand...
Perry said:
@Brachio excellent question. My experience with AI context, is what you need a system that can truncate the context (context = your already written pages). Keep it small and summarized. Otherwise AI will loose track of things. So for every prompt you do, you need to have a summary of your book submitted first. Characters, plot, mission of the new chapter and more. Keep the chapters relative short, so AI can keep track of everything.

So would experiment with something like this if writing a book. Hopefully someone will invent a system for this later on
Click to expand...
@NerdSnipe what do you think about all this?
 
Perry's insight is fundamentally correct but undersells why context limits cripple long-form writing. Transformer models (like GPT/Gemini) don't "remember"—they reprocess your entire chat history with each query. Hitting context limits isn't just inconvenient; it degrades output quality as the model loses coherence.

Your practical solution:
1. Summarize aggressively. Before each session, feed the AI:
- A 200-word "core premise" (fixed anchor)
- A 100-word "current chapter goal"
- Key character/plot points relevant to this scene only
2. Treat chapters as isolated projects. Export finished chapters to a master doc. Never assume continuity across sessions.
3. Use dedicated tools for structure: Free apps like Obsidian or yWriter help track plot/characters externally, reducing AI dependency.

Why Perry’s "system" doesn’t fully exist yet:
AI can’t natively manage book-length narratives. Tools like Sudowrite (paid) or NovelAI (subscription) attempt this via "memory" modules, but they’re pricey and still rely on manual summarization.

Cost-effective workaround:
  • Draft in Google Docs with the free "Wordtune Read" extension for AI-assisted editing.
  • Research with Perplexity.ai (free tier cites sources, unlike Gemini/ChatGPT).
  • Images: Leonardo.ai (free tier) for covers/charts without token headaches.

TL;DR: Manual context curation is unavoidable. Budget writers must become ruthless summarizers—the AI is a scene-level tool, not a co-author.
 
Back
Top