Q: What have you learnt about using AI for workflow efficiencies?
A: AI is definitely growing up and out into existing systems that drive or support workflows, to help to speed up humans delivering quality work of all kinds across publishing companies — be that work in content generation, quality control, data conflation and presentation thereof. Or in the case of at least one customer, laser-focused sales targeting.
On the whole, AI can already do a great job of automating repetitive tasks such as content tagging, article formatting, first-pass style and fact-checking — all things that remove manual labour and time from editorial teams, leaving more time for all-important content generation and value creation.
In some cases, AI is also being used to create entire sub-products in its own right with automatic, dynamic translations of pre-existing content, opening up entirely new geographic markets by simply plugging directly into existing workflows without really adding to cost at all, let alone anyone’s time.
Of course, AI can also generate content but in an industry increasingly reliant upon quality and true value, we see the co-pilot concept being adopted by publishers with far greater velocity. Great results can come in terms of accelerated proof-reading, searching for related items for prior editorial ‘stance’ on particular topics and optimising headline generation for specific channels. As well as adding / managing metadata around taxonomies, articles and content.
So, the main learning we’re seeing is AI being used in multiple different ways to support and assist rather than replace or remove processes. We think this has the potential to change but it’s not there yet. And even when it is, we see even higher quality of even more content generation as the output. Efficiency for more, not less.
Q: In which use-case has AI proved most effective?
A: We mainly see a growth of demand for CMS editing tools enriched with AI for publishers looking to enrich and support an efficient workflow. The concept of generating more content, of better quality, optimised for multiple channels, in less time and without adding extra resources is of course super attractive — who doesn’t want this in an audience-centric industry?
The issue we stumble across time and time again is that the demand for change is often tracking at a different pace to the maturity of the technology itself. Things are moving at such speed that by the time requirements have been gathered, work estimated and development scheduled in an already overburdened roadmap, AI can do even more to help.
We see lots of AI-derived promises in development, which — by the time they are ready for use — are already surpassed by AI’s capabilities. This makes it really difficult for publishers to invest wisely in their tech, particularly when there is demand across multiple departments for AI tools and support: AI-driven audience insight, AI-curated and automated marketing, AI-led lead gen... and on it goes.
To counter this huge demand — at least in the short-term while publishers and suppliers like us alike figure out where AI is most useful — we have developed specialised things for the content teams such as a browser extension that plugs into a CMS to ingest content, brand, writing styles, editing styles, headline lengths, SEO and other preferences before offering suggestions in a CMS agnostic method. It’s supporting the CMS, its functions and workflows, without yet being fully embedded. This enables our customer to start to adopt AI practices that will soon be commonplace, without enormous investment in CMS-wide change or even replatforming.
This tool effectively allows editorial staff to work across systems in tandem with their acceleration tool rather than it either waiting in a development queue — or in the cases of smaller publishers, entirely impossible to shoehorn into a rigid off-the-shelf CMS without breaking something! Possibly the bank.
This use-case has been a good example of where publishers are embracing AI wisely and experimentally before fundamentally altering their tech stack. Or building an entire roadmap based on AI that hasn’t taken the sheer rate of acceleration in this area into account.
Of course, the goal will be a completely embedded set of tools within the CMS. And we are experimenting in this area ourselves. But we love to see publishers trying on the practices first before sacrificing all other product development in favour of AI tools.
Three best practice top tips
- Accept that your current systems can take a while to grow and look for ways to boost the humans using accessible AI tools right now. Don’t wait for the roadmap. AI is moving way too fast for even the most agile of programming teams.
- Great AI tools are being paired with good integral search to add relevance to content sets. Look for AI solutions that will use your content as well as provide insights or suggestions to bolster or check the current piece.
- The “Emperor’s New Clothes” phenomenon is huge at the moment when it comes to AI. And AI changes every few months. Don’t lock to current AI patterns or switch to AI enabled publishing and workflow that lack in other more practical areas. AI should not be the reason to replace a good system that does many things well.
Stewart and the other contributors to our AI Special will take part in an ‘AI Special – Q&A’ webinar on Tuesday, 28 January. Click here for more information and to register.
With a combined experience of over 80 years in the publishing industry, Full Fat Things works with B2B and B2C publishing businesses to develop sustainable digital products with deep integrations with infrastructure and workflows. We create fully customised outcomes using open-source software to enable ultimate flexibility now and in the future.
Tel: 020 7099 3875
Email: info@fullfatthings.com
Website: www.fullfatthings.com
This article was included in the AI Special, published by InPublishing in December 2024. Click here to see the other articles in this special feature.