Building in Public: How AI Agents Run Our Content Factory

Описание и применение

Компания использует 19 ИИ‑агентов для создания контента, подготовки продаж и анализа результатов, при этом человек проверяет публикации. Процесс разделён на роли: написание, продажи, SEO и аналитика, что повышает скорость и качество работы.

Задача

Automating content creation and sales follow‑ups while maintaining quality and preventing fabricated data.

Результат

The workflow increases content output (67 pieces in a week) while keeping quality through human review and iterative guardrail improvements.

Рабочий процесс

• 1. Analytics agent reads weekly performance data and identifies successful and stalled topics.
• 2. Topics and constraints are set for the week.
• 3. Writing agent drafts articles from approved briefs.
• 4. Sales agent prepares replies and follow‑ups for a person to send.
• 5. SEO agent generates pages from vetted templates and checks them against rules.
• 6. Human reviewers approve any content that goes public.
• 7. Analytics agent reads results, records failures, and suggests instruction adjustments.
• 8. Team reviews adjustments and updates agent prompts.
• 9. Repeat cycle each week.

Доказательства

Building in public means we show how our AI agents actually run a content factory. They draft articles, prepare sales follow‑ups, generate SEO pages and read analytics, while people set the goals and approve the result.

Источник

We split the work into narrow roles: Writing agent, Sales agent, SEO agent, Analytics agent.

Источник

Early in the week, analytics shows what worked and what stalled. We pick topics and set constraints. The agents draft, and a person reviews anything that goes public.

Источник

They repeat themselves, drift from our tone, and sometimes claim things they can't support. That is why we forbid invented data.

Источник