Inngest Profiles Featured's Orchestration Layer

Highlights

• The workflow platform published a customer story on how Featured's HARO and Connectively pipeline ships 2,500 articles a month on durable orchestration the team never has to operate by hand.
• Online, Sept 14, 2026
• The numbers, per the story
• 2,500: Articles published a month across three brands.
• 42: Average expert answers collected per question before a run starts.
• 100M+: Emails HARO sends a year to more than 60,000 journalists.
• 100,000+: Members on Connectively, Featured's expert-response platform.
• 2.5M+: Pitches fielded across HARO and Connectively combined.
• 7: People running all three brands today, down from an eight-person orchestration team in 2020.
• In 2020, keeping Featured's pipeline running took eight people: a content strategist, two community-outreach specialists, a technical SEO manager posting articles by hand, staff writers, and a digital PR manager holding it together. Founder Brett Farmiloe called that group "a team of 8 or so people that were the orchestration layer to make sure things ran reliably." Today, the story reports, that orchestration layer is software, and a team of seven runs HARO, Connectively, and Featured together.
• <p>Inngest helps our trains (workflows) run; without losing sleep or manually shoveling coal into an engine.</p>
• The piece traces how a run moves from an open question to a published article. Rather than a durable timer that sleeps until a deadline, Connectively runs a cron sweep every 15 minutes that re-reads which questions have actually passed their current deadline — so a publisher pushing a deadline back needs no special handling. Once a question closes, one run works through 60 to 150 durable steps, evaluating experts on experience, expertise, authority, and trust, then admitting answers one at a time so a diverse set of perspectives makes it into the article instead of near-duplicates.
• What runs where, per the story
• Orchestration decides what happens and when. The database decides what's allowed to happen at all.
• The story also covers two deliberate limits on what the orchestration layer is allowed to own. A Postgres uniqueness constraint, not workflow logic, guarantees an expert is only notified once per published article. And most functions don't retry: a run that can't produce an article the team would stand behind marks itself failed and stops, rather than risk auto-publishing something under a publisher's masthead.