Contently helps enterprise clients produce on-brand content at scale. Managing editors within Contently oversee client relationships and translate briefs into high-quality articles. The company wanted to build an AI content writer that could generate story drafts from standard briefs, saving time and cutting costs while maintaining editorial excellence. Building a system that could match the quality of professional writers with 10+ years of experience required specialized AI expertise that went beyond standard LLM integration, so they hired NineTwoThree to build the Contently AI Writer.
Agentic AI framework that mimics the mental process of seasoned writers to produce editorial-grade content.
Full-stack web application for AI content creation with SEO agents, style agents, and fact-checking agents.
Tone-matching, style control, and fact-checking integrated into the writing workflow.
Framework for testing AI vs. human drafts to continuously improve quality.
Contently needed an AI writing system that could reduce production time and cost while preserving the editorial standards its enterprise clients expected.
The AI had to match or exceed professional writers with 10+ years of experience. Industry-specific knowledge meant writers spent hours on outlines, and managing editors couldn't be expected to prompt like AI experts, so the system had to deliver from standard briefs.
Every article had to reflect each client's tone, brand identity, and editorial strategy while avoiding hallucinations. Some content was health-related, where errors carry high risk, which made simple out-of-the-box AI tools unreliable for Contently's standards.
Drafts had to pass SEO tools and meet high standards for searchability and credibility, at a fraction of the cost of traditional writing. Most AI solutions keep latency low by outputting lower-quality content, a tradeoff Contently couldn't accept.
We worked alongside Contently's managing editors, testing AI against human drafts to shape each part of the system.
We designed AI flows that mimic seasoned writers, starting with structured outlines and spiky points of view before expanding into prose. Editors get high-quality output from standard briefs, without extensive prompt engineering.
We integrated tone-matching, style control, fact-checking, and context processing to ensure the writing felt human and brand-safe. These subsystems work together to catch hallucinations and maintain editorial standards.
9-part
subsystem architecture for AI writing
We worked with managing editors to test AI drafts against human ones, identify areas for improvement, and train each subsystem accordingly. This continuous feedback loop helped the system learn from real editorial judgment.
We accepted longer generation times (about 5 minutes per article) to optimize depth and polish, emulating the psychology of human-like effort. The system prioritizes quality over speed, which produces better results.
Managing editors can now support more clients with the same effort, unlocking new revenue potential for Contently. The AI writer handles the heavy lifting of draft creation, freeing editors to focus on client relationships and strategy.
Managing editors produce 5x more content using Contently AI. The system generates first drafts in minutes, reducing the time and cost associated with manual writing while maintaining editorial quality.
AI-written drafts pass internal rubrics for SEO, tone, and accuracy, often rivaling or exceeding human counterparts. The integrated fact-checking and tone-matching subsystems ensure every piece meets enterprise editorial standards.
The modular subsystem design allows Contently to fine-tune the experience for each publication. As the platform scales, the architecture adapts without requiring a full rebuild.
.svg)
NineTwoThree AI Studio designed and developed a full stack web application for AI content creation for a SaaS company. The platform included features like SEO agents, style agents, and fact-checking agents. NineTwoThree provided excellent project management and communicated via email, Slack, Google Chat, and weekly calls.
Build a new platform, or add AI capabilities to existing software, we deliver solutions that work in real operational environments.
Meet with founders Andrew Amann & Pavel Kirillov
Custom AI implementation roadmap
No sales pressure
.webp)
