INMA has launched a ‘Siri-style’ AI answer engine, claiming to be the first industry association to do so.

Drawing exclusively on the association’s proprietary content – including blogs, reports, PDFs and multimedia content – the beta-level tool provides members (only) with answers derived from its archive, and “directly relevant to media professionals”.

Member engagement manager Jennie Rambo told GXpress the answer engine was a profound change in how you got answers from the INMA network.

Designed for simplicity – focussing on asking questions and receiving precise answers without unnecessary distractions – it not only sets out to deliver concise answers, but also offers follow-up suggestions, additional information and links to source materials for deeper exploration.

INMA says at launch, the multilingual AI draws on a decade of 7,460 ‘best practices’ from INMA’s global media awards, 5,712 blog posts, 2,337 presentations from conferences, study tours and master classes, 72 original reports, plus video and audio files from webinars, conferences and master classes.

Executive director and chief executive Earl Wilkinson says with almost 90 per cent of INMA content walled off from public AI and search engines, its surfacing and contextualisation something the news industry has never seen before.

“AI solves a problem of INMA content discovery and insights by cutting through media formats such as HTML, PowerPoints, PDFs, video, and audio,” he said. “The new answer engine makes media professionals smarter faster and provides INMA members a leg up on ideas, insights, and best practices.”

The answer engine is being launched in beta mode, but has been tested by the association’s board members, initiative leads and senior staff. Board members encouraged the launch, adding they found the answers invaluable. “It’s not perfect, yet we know INMA members will help improve ‘Ask INMA’ in the weeks and months ahead,” Wilkinson said.

Built on the OpenAI LLM and developed in partnership with Poland-based Techlabee.ai, the engine’s searches take longer than public answer engines because it does not rely on pre-trained knowledge. Expect a typical inquiry to take 30-60 seconds to answer, as the engine needs to ‘read’ and process more information on the spot, being in “deep research mode” all the time.

“While this takes more time, the answers are more accurate and relevant,” Wilkinson said. “In short, the wait is usually worth it.”