December 2025 Summaries

In 2025, Mintlify experienced a transformative year, evolving from a documentation platform into a key infrastructure layer for AI's understanding of technical knowledge, achieving over 8-figures in annual recurring revenue and onboarding more than 10,000 companies, including major names like Microsoft and PayPal. The company processed over one million documentation updates and facilitated more than 1 million AI queries monthly, highlighting a shift towards intelligent documentation that not only provides answers through AI integration but also updates itself autonomously via the Mintlify Agent. This evolution underscores a broader trend where static documentation is replaced by dynamic, interactive systems that cater to both human users and AI agents, with innovations like the llms.txt standard and self-updating documentation paving the way. The company also expanded its team, completed its first acquisition with Trieve, and maintained a robust pace of development with over 6,000 pull requests, signaling a commitment to enhancing developer experience and laying the groundwork for future advancements in knowledge management.

In November 2025, Mintlify experienced a significant security event when researchers discovered an XSS vulnerability in its static asset hosting, where cross-domain access allowed malicious actors to execute scripts on customer sites. The company swiftly responded by deploying a fix within 45 minutes of being notified, and engaged in a week-long collaboration with the researchers to identify and patch additional vulnerabilities, such as path traversal and server-side rendering code execution. The incident led to comprehensive security improvements, including stricter asset access controls, enhanced incident response protocols, and an updated bug bounty program, with Mintlify expressing gratitude to the researchers and affected customers for their cooperation and trust.

Efforts to improve the Mintlify assistant focused on rebuilding the feedback pipeline, transitioning conversation data to ClickHouse, and categorizing negative interactions. The analysis revealed that search quality was the assistant's main weakness, although most responses were strong. To address these issues, Mintlify updated its server to map feedback events to original conversation threads, enabling effective querying of negative feedback. The team categorized feedback into eight types using a large language model (LLM) and highlighted search as a key area for improvement. Despite a model upgrade, feedback consistency remained unchanged, leading to the expansion of the insights tab for better understanding customer concerns. UI improvements included better navigation through conversation threads and more stable response loading. Users are encouraged to provide feedback or join the Mintlify team to further enhance the assistant.

Mintlify introduces a self-updating documentation engine designed to maintain accurate and accessible documentation by monitoring codebases for user-facing changes and proposing updates in real-time. This tool addresses the challenge of keeping documentation in sync with rapidly shipped software by automatically detecting code changes that necessitate documentation updates, surfacing these updates to teams, and generating context-aware drafts for review. By integrating with selected repositories, Mintlify's agent ensures that documentation remains up-to-date without the need for manual tracking or inquiry, allowing teams to publish high-quality documentation quickly and efficiently. As modern development relies heavily on precise documentation for both human users and AI agents, this solution aims to bridge the gap between fast-paced software releases and documentation updates, maintaining a human element in the workflow while automating much of the process.

Mintlify, a platform providing documentation for numerous developer sites, faced challenges with slow cold starts affecting user experience due to frequent deployments that invalidated their Next.js ISR caching solution. To address this, they developed a custom caching solution using Cloudflare products, effectively bringing their cache hit rate from 76% to nearly 100%. This new system separates deployments from cache invalidation, utilizing a proxy layer to manage requests and cache updates, automatic version detection for revalidation, and a proactive prewarming strategy for documentation updates. The solution includes a revalidation coordinator using Cloudflare's Durable Objects to prevent race conditions and a dedicated revalidation worker for cache warming. The approach ensures that caches are consistently warm, regardless of whether changes stem from new code deployments or documentation updates, thereby improving page load speeds and reliability. The system's self-healing capability further enhances robustness, automatically retrying failed revalidations and managing locks. Mintlify invites interested individuals to explore career opportunities to work on such innovative engineering projects.

The mintlify agent now offers enhanced capabilities for managing documentation through integrations with Slack and a dedicated dashboard, allowing users to automatically generate pull requests, suggest updates, and maintain accurate documentation with minimal effort. By understanding existing documentation and code context, the agent can surface suggested changes and assist in creating new content or revising outdated information directly from prompts or pull requests. Users can interact with the agent via a dashboard panel or within Slack, where it can propose documentation updates and facilitate collaboration. As part of a broader trend toward self-updating knowledge systems, the agent captures content from various sources and proactively flags outdated information, ensuring documentation remains current and reliable for both human and AI consumers.

Dmytro Tovstokoryi, a part-time design engineer at Mintlify, undertook a project to address and improve various user interface (UI) quirks over a span of 10 days, inspired by Cole Bemis. Through a daily thread, Tovstokoryi shared his progress and invited feedback on UI bugs, which he promptly addressed in real-time, enhancing his learning experience. His improvements included fixing navigation layout shifts, addressing dynamic number updates to prevent layout jumps, implementing origin-aware animations for dropdowns, ensuring input field prefixes and suffixes activate the field when clicked, unifying hover states in dashboard navigation, creating a keyboard-accessible date range picker, and resolving button focus issues. While these fixes didn't address critical functionality, they contributed significantly to the polish and intuitive feel of the interface, demonstrating the importance of minor adjustments in achieving a seamless user experience.