Lex is a word processor with an AI editor built into it, from the team behind the Every newsletter. Lex launched in late 2022 and raised a $2.75M seed round from True Ventures in 2023. The interface looks like a stripped-down Google Docs. The AI stays one keystroke away: type +++ to continue a thought, hit Cmd+K for a command bar, or open the “Ask Lex” sidebar for feedback on a draft.
Most AI writing tools generate. Lex is built to edit. Its “AI checks” read your draft and tell you where the argument is weak, where the intro drags, where you buried the point. The feedback skews critical rather than flattering, which is rarer than it sounds and more useful than another tool praising your work.
So Lex is a tool for people whose job is words. If your work goes through a serious edit before it ships, Lex fits. If you need volume, templates, or anything past the page, it doesn’t.
What It Does Well
Critical feedback, not cheerleading. Ask a general chatbot to review your draft and it usually tells you the draft is strong. Lex’s checks are tuned to find the soft spots, the unsupported claim or the paragraph that says nothing. For anyone editing their own work, an honest second read beats another draft.
A calm place to actually write. The editor is deliberately plain. There’s no feature sidebar and no template gallery. You get focus mode, dark mode, version history, and folders. The AI holds back until you reach for it, so the tool matches your voice instead of flattening it into house-model prose.
Your choice of model, in one seat. Pro users can run GPT-4.1, Claude 4 Opus, or Claude 4 Sonnet on the same document and switch per task. Claude for a nuanced critique, GPT for a fast brainstorm. You get that flexibility without juggling three separate subscriptions.
How Each Role Puts It to Work
The Marketer: The strongest fit. Draft a blog post or a bylined thought-leadership piece in Lex, then run the AI checks to tighten a loose argument and cut the word count before it goes to review. What Lex won’t do is run the campaign around it. There are no ad-copy templates and no SEO workflow, and it won’t enforce a brand voice across a team. It sharpens the writing, not the marketing engine, and you still fact-check anything the model asserts before you publish.
The Researcher: Lex helps you write the deliverable, not run the analysis. Draft the findings narrative or the executive summary, then stress-test a shaky conclusion with the critique mode before a client sees it. Lex states it does not train on your writing and routes to enterprise model APIs, which is reassuring. But it has no grounding, no citations, and no transcript coding. It’s an editor, not a synthesis engine, and the security caveats below matter before any confidential transcript goes near it.
The Salesperson: A narrow but real use: the proposal narrative, or the long-form answer in an RFP. Lex earns its keep on the high-stakes written piece you want a demanding second read on. It is not a sales tool in any other sense. No CRM sync, no call recaps, no pipeline management.
Weaker fits: Finance, Operations, and HR get little here. There are no numbers, no audit trail, no integrations, no automation, and nothing built for hiring. You could draft a policy or a job description in Lex, but that’s writing in a text editor, not an HR capability.
Where It Could Be Better
The honest limit is breadth. Outside of writing, Lex does almost nothing. It isn’t a workspace or a project manager, and it won’t connect to the tools a team already runs. On vague prompts it can loop or repeat itself, so the input still needs a human with a point of view. There’s no middle tier between free and Pro, and the free plan caps how many AI checks you get. For a team that wants one tool to cover content, campaigns, and reporting, Lex is the wrong shape.
Why Not Just Use ChatGPT or Claude?
Lex runs on those same models, so part of what you’re paying for is a wrapper. The raw generation is available from a model you probably already own. What Lex adds is the workflow. The model lives inside a real editor with your whole document as context, so there’s no copy-paste loop between a chat window and your draft. The critique mode is purpose-built and less sycophantic than a raw chatbot. Version history and live collaboration come standard. If you write every day, that convenience is the value. If you write occasionally, you can get most of this from the model you already pay for.
Security & Compliance
What’s confirmed is reasonable for a small vendor. Lex hosts your data on its own database, states plainly that it does not train models on your writing, and says it would email you and offer an opt-out if that ever changed. It reaches the language models through enterprise APIs from OpenAI, Anthropic, and TogetherAI, which carry stronger privacy terms than the consumer tiers of those same tools.
What’s missing is the paperwork an approver needs. We found no public SOC 2 or ISO 27001 report and no stated HIPAA posture. There’s no trust page, and no visible detail on SSO, role-based access, or audit logs. For a solo writer that’s fine. For confidential client work, treat the absence as an open question to put to the vendor before anything sensitive goes in.
Data & AI Connectivity
Lex imports from Google Docs and lets you export your work, and that’s most of the story. It doesn’t connect to a CRM or a data warehouse, and we found no public API or webhooks. Its one real connectivity story is model connectivity: you can point it at several LLMs from one place. As a component that plugs into a wider stack, it’s a closed box.
Ratings
| Dimension | Rating | Rationale |
|---|---|---|
| Usability | 4.5 / 5 | Clean, Google-Docs-familiar, near-zero onboarding, AI one keystroke away. |
| Power | 4.0 / 5 | Genuinely strong at drafting and honest critique; model choice raises the ceiling. Deep, if narrow. |
| Flexibility | 2.5 / 5 | Writing only. No integrations, no API, no templates, no workspace. |
| Cost | 4.0 / 5 | Flat Pro pricing with no credits or overages, a usable free tier, and premium models included. |
Best-Fit Roles: Strongest for Marketers and Researchers who write long-form. A narrow fit for Salespeople on high-stakes documents. Little here for Finance, Operations, or HR.
Conclusion
Lex is the best AI editor we’ve used for the writing itself, and close to useless for anything else. That focus is the whole product. It won’t run your campaigns or plug into your stack, but for the person whose deliverable is prose, it turns the AI into a demanding second reader instead of a ghostwriter. That’s the right division of labor. Lex can find the weak argument. Fixing it is still your job.
This review is part of a larger series in which we evaluate AI tools through the lens of the B2B teams that actually use them.
Last updated: 8/12/2026.