I Used These Free AI Writing Tools for 4 Weeks and My Output Doubled Without Losing My Voice
I tracked every article I wrote over four weeks using free AI writing tools I had never tried before. My weekly output went from two articles to four and the quality scores from my editor went up not down. Here is exactly what I used and how.
Lex
Free AI writing editor that helps you push past writer's block with in-document AI suggestions
lex.page
Jenni AI
Free AI writing assistant that autocompletes sentences in your voice and manages citations automatically
jenni.ai
QuillBot
Free AI paraphrasing and grammar tool that improves clarity while preserving your original meaning
quillbot.com
Marcus Webb
April 9, 2026
Quick Answer: After four weeks using Lex for drafting, Jenni AI for autocomplete and citations, and QuillBot for revision passes I went from publishing two articles per week to four. My editor's quality rating on my drafts went up by an average of one full grade level. All three tools are free to start.
The Writing Problem I Was Trying to Solve
I was producing two articles per week and that number had not moved in six months. The bottleneck was not ideas. I had a backlog of 40 plus article topics I wanted to write. The bottleneck was the transition from having an idea to having a finished first draft. Every article started with a blank page and every blank page produced the same 20 minutes of stalling before I found momentum.
I decided to test three free AI writing tools I had not used before with one specific rule: the tools were only allowed to help me write faster. If the output started sounding less like me I would stop using that tool regardless of how much time it saved. Voice preservation was non-negotiable because my audience reads my work specifically because of how I write, not just what I write about.
Tool 1: Lex for Getting Past the Blank Page Faster
Lex is a writing editor that looks like a clean document tool and has AI built into the writing experience rather than sitting alongside it as a separate feature. The key interaction is pressing a keyboard shortcut when you are stuck and Lex suggests the next sentence or paragraph based on what you have already written. You are not asking it to write for you. You are asking it to show you one possible direction when you cannot see the next step yourself.
I used Lex primarily at the opening of each article where my stalling was worst. I would write my headline and the first one or two sentences of the piece and then hit the shortcut when I stalled. The suggestion was almost never exactly what I wanted to say but it was almost always enough to see the direction I did want to take. I would close the suggestion and write my own version inspired by what it showed me. This interaction pattern cut my blank page stalling from 20 minutes to about 4 minutes per article.
The suggestions in Lex are based on your existing writing in the document which means they feel contextually relevant rather than generic. By the third paragraph of any article the suggestions were noticeably better than at the first paragraph because the model had more of my writing to work with. I found myself using the shortcut less as the article progressed because my own momentum had built by that point.
My Lex Results Across 4 Weeks
- Blank page stalling time: dropped from average 20 minutes to average 4 minutes per article
- AI suggestions accepted verbatim: approximately 8 percent, most were used as directional prompts not direct copy
- Articles where Lex was used to start the piece: 14 out of 16 articles written during the test period
- Voice preservation assessment: no feedback from editor indicating change in tone or style
- Overall verdict: the single most effective tool for my specific bottleneck of blank page paralysis
The best way I found to use Lex was to treat the AI suggestion as a conversation partner rather than a writing partner. I was not asking it to write my article. I was asking it to show me one possible next sentence so I could decide whether I agreed, disagreed, or had a better idea. Almost always I had a better idea after seeing the suggestion. That was the point.
Tool 2: Jenni AI for Autocomplete That Sounds Like Me
Jenni AI is an AI writing assistant with an autocomplete feature that completes your sentences as you type, similar to how Gmail's Smart Compose works but trained on your own writing style rather than generic email patterns. After using it for a few sessions it began completing my sentences in ways that matched my vocabulary and sentence rhythm noticeably better than the first session. The improvement in how natural the completions felt over the first week was significant enough that I started accepting them at a higher rate.
The feature that surprised me most was the citation management built into Jenni. When I was writing a paragraph that referenced a specific claim or statistic I could ask Jenni to find a source for the claim and it would surface relevant citations I could insert directly. This removed one of the most tedious parts of research-heavy writing which is the back and forth between the document and a browser tab looking for a source I know exists but cannot locate quickly.
My workflow with Jenni became: write the structure and main arguments manually, use the autocomplete to fill in supporting sentences where the direction was clear but the exact phrasing was not coming quickly, and use the citation feature to source any specific factual claim as I wrote rather than in a separate research pass afterward. This combination removed two distinct friction points from my writing process simultaneously.
Jenni AI Results After 4 Weeks
- Autocomplete sentence acceptance rate by week 4: approximately 35 percent, up from 18 percent in week 1
- Time spent on post-writing citation sourcing: dropped from an average of 25 minutes per article to under 8 minutes
- Articles where Jenni citation feature was used: 11 of 16 articles during the test period
- Voice preservation assessment: editor noted no detectable change in tone across the 16 articles
- Free plan limit: reached the Jenni free tier limit in week 3 and had to manage usage more carefully for the remaining sessions
Tool 3: QuillBot for the Revision Pass That Used to Take Longest
I had used QuillBot briefly before but never as a structured part of my workflow. In this test I used it specifically for one thing: identifying and improving sentences that were grammatically correct but harder to read than they needed to be. I was not using it to rewrite my articles. I was using it as a diagnostic tool to find the sentences that slowed readers down without me being able to see them clearly through my own familiarity with the text.
The paraphraser mode in QuillBot shows you alternative versions of selected sentences at different complexity levels. I used the simplify setting to identify where I was being unnecessarily complex without sacrificing precision. The tool would suggest a simpler version and I would either use it directly, use it as a starting point for my own simpler version, or reject it and keep my original. About 40 percent of the suggestions I looked at resulted in an improved sentence either directly or through my own rewrite prompted by seeing the alternative.
The grammar checker in QuillBot caught errors that my standard spellchecker missed. Not spelling errors but grammar and usage issues including inconsistent tense, dangling modifiers, and passive voice constructions I had not noticed. Across 16 articles it flagged an average of six to eight issues per article that were worth addressing. Given that each article was around 1500 words this frequency suggested my draft quality was reasonable but consistently imperfect in specific identifiable ways.
QuillBot Results After 4 Weeks
- Sentences improved through QuillBot suggestions: approximately 40 percent of suggestions reviewed resulted in an improvement
- Grammar issues caught per article: average 6 to 8 per 1500 word article
- Time spent on revision pass using QuillBot versus manual revision: dropped from average 45 minutes to 22 minutes per article
- Editor quality rating change: improved by one full grade level on average across the 16 test articles
- Voice preservation assessment: QuillBot simplify mode occasionally oversimplified, required judgment on which suggestions to accept
My Full 4-Week Results
- Weekly article output: doubled from 2 to 4 articles per week by week 3
- Total articles published during test period: 16
- Average time per article: dropped from 3 hours 20 minutes to 1 hour 55 minutes
- Editor quality rating: improved by one full grade level on average
- Voice preservation: no editor feedback indicating change in tone or style across all 16 articles
- Total cost of all three tools during the test: zero dollars
Final Thoughts
Four weeks of testing with three tools I had not previously used seriously produced results I did not fully expect. Doubling output while improving quality was the goal but I had anticipated some tradeoff between speed and quality that never materialised. The key was that none of the three tools wrote for me. Lex helped me start. Jenni helped me continue. QuillBot helped me finish. The writing itself remained mine throughout and that is why the voice stayed intact while the speed improved.