Grammarly Adds Full Document Rewriting via Custom LLM

Grammarly has officially launched a full document rewriting capability powered by its own custom-built large language model, marking the company's most ambitious AI feature to date. The new tool goes far beyond traditional grammar and spelling corrections, enabling users to restructure, rephrase, and completely transform entire documents with a single prompt.

This move positions the San Francisco-based company — already used by over 30 million daily active users — as a direct competitor to generative AI writing platforms like Jasper, Copy.ai, and even ChatGPT's document editing features.

Key Facts at a Glance

• Full document rewriting allows users to transform tone, structure, and style across entire files — not just sentence-by-sentence edits
• The feature is powered by Grammarly's proprietary LLM, not a third-party model like GPT-4 or Claude
• Available initially to Grammarly Business and Grammarly Enterprise subscribers, with plans for broader rollout
• The custom model was trained on billions of writing samples with a focus on professional communication
• Grammarly claims the tool reduces document editing time by up to 70% compared to manual rewriting
• The feature supports over 25 document types, including emails, reports, proposals, and marketing copy

Grammarly Moves Beyond Grammar Into Generative AI Territory

Grammarly's evolution from a spell-checker to a full-fledged AI writing platform has been years in the making. The company first introduced its GrammarlyGO generative AI assistant in April 2023, offering sentence-level suggestions and short-form content generation. This latest update represents a quantum leap in ambition.

The new full document rewriting feature allows users to upload or paste an entire document and issue natural language instructions. Users can request changes like 'make this more formal,' 'simplify for a non-technical audience,' or 'restructure this report to lead with recommendations.'

Unlike previous versions of GrammarlyGO that operated at the paragraph level, the new system analyzes the full context of a document before generating rewrites. This holistic approach ensures consistency in tone, terminology, and argumentation throughout the entire piece.

Inside Grammarly's Custom LLM Architecture

The decision to build a proprietary large language model rather than relying on OpenAI's GPT-4 or Anthropic's Claude sets Grammarly apart from most AI writing tools on the market. According to the company, the custom model was specifically architected for writing assistance tasks rather than general-purpose conversation.

Grammarly's engineering team reportedly trained the model on a curated dataset of billions of professional writing samples, with particular emphasis on business communication, academic writing, and technical documentation. The company states that this specialized training gives its model a significant edge in understanding writing conventions, style guides, and professional tone.

Key technical differentiators of the custom LLM include:

• Context window optimization designed for long-form documents up to 50,000 words
• Style-aware embeddings that understand the difference between formal business writing and casual communication
• Structural reasoning capabilities that can reorganize document sections logically
• Brand voice consistency features for enterprise users who need to maintain corporate style guides
• Low-latency inference that generates full document rewrites in under 15 seconds for standard-length documents

The model reportedly runs on a hybrid cloud infrastructure, with enterprise customers having the option to deploy on private cloud instances for enhanced data privacy — a critical selling point for regulated industries like healthcare and finance.

How the Feature Stacks Up Against Competitors

The AI writing assistant market has exploded over the past 2 years, with dozens of tools vying for user attention. Grammarly's new rewriting feature enters a crowded but lucrative space, currently estimated at $5.3 billion and projected to reach $12 billion by 2027.

Compared to ChatGPT, which can rewrite documents through its conversational interface, Grammarly's tool offers tighter integration with existing workflows. The feature works natively within Google Docs, Microsoft Word, Slack, and the company's browser extension — meaning users don't need to copy and paste text into a separate interface.

Jasper and Copy.ai have focused primarily on marketing copy generation, leaving a gap in the professional and enterprise communication space that Grammarly is now aggressively filling. Meanwhile, Microsoft Copilot offers document rewriting within the Microsoft 365 ecosystem but requires an additional $30 per user per month subscription.

Grammarly's pricing for the rewriting feature is bundled into existing Business plans starting at $15 per member per month, potentially undercutting Microsoft's offering for organizations that don't need the full Copilot suite.

Enterprise Adoption and Data Privacy Safeguards

Enterprise customers represent Grammarly's fastest-growing segment, with the company reporting over 70,000 teams using its platform. The full document rewriting feature addresses a critical enterprise need: maintaining consistent communication quality across large organizations.

The company has implemented several privacy safeguards specifically designed for the enterprise market. User documents are not used to train the underlying LLM. All data is encrypted in transit and at rest. Enterprise administrators can set granular permissions controlling which teams and individuals can access the rewriting feature.

Grammarly also announced SOC 2 Type II certification for the new feature and confirmed compliance with GDPR, CCPA, and HIPAA requirements. These certifications are essential for adoption in sectors like banking, healthcare, and government where document privacy is non-negotiable.

Early enterprise adopters include several Fortune 500 companies, though Grammarly has not disclosed specific names. The company reports that pilot programs showed a 40% increase in document turnaround speed and a measurable improvement in writing quality scores across participating organizations.

What This Means for Writers, Businesses, and the AI Industry

The practical implications of Grammarly's move extend across multiple stakeholder groups. For individual writers, the feature democratizes access to professional-quality editing that previously required hiring a human editor or spending hours on manual revisions.

For businesses, the tool promises significant productivity gains. A marketing team can draft a rough report and transform it into a polished client-facing document in seconds. An HR department can rewrite internal policies to be more inclusive or accessible without engaging external consultants.

For the broader AI industry, Grammarly's decision to build a custom LLM signals a growing trend among vertical software companies. Rather than wrapping third-party APIs, companies are investing in domain-specific models tailored to their core use cases. This approach offers several advantages:

• Greater control over model behavior and output quality
• Reduced dependency on external API pricing changes
• Better data privacy guarantees for sensitive enterprise content
• Faster iteration cycles without waiting for upstream model updates
• Competitive differentiation that's harder for rivals to replicate

This trend mirrors what companies like Bloomberg (BloombergGPT), Salesforce (Einstein GPT), and Adobe (Firefly) have already pursued — building proprietary models tuned for specific professional domains.

Looking Ahead: Grammarly's AI Roadmap

Grammarly's full document rewriting feature is likely just the beginning of a broader AI transformation for the company. Industry analysts expect the platform to expand into several adjacent capabilities over the next 12 to 18 months.

Anticipated developments include real-time collaborative rewriting where multiple team members can guide AI rewrites simultaneously, multi-language document transformation that goes beyond translation to include cultural adaptation, and voice-to-document features that convert meeting recordings into polished written summaries.

The company's valuation, last reported at $13 billion following a 2021 funding round, could see a significant boost if the new AI features drive enterprise adoption as projected. Grammarly has been widely rumored to be exploring an IPO, and a successful AI product launch strengthens that narrative considerably.

For now, the full document rewriting feature is rolling out in phases. Business and Enterprise subscribers gain access immediately, while Grammarly Premium users can expect availability in Q3 2025. Free-tier users will receive a limited version with a cap of 3 full document rewrites per month.

The writing assistant market is entering a new phase where simple grammar checking is table stakes. Grammarly's bet on a custom LLM and full document transformation signals that the future of AI writing tools lies not in catching errors — but in fundamentally reshaping how professionals create and communicate through text.