Held in June at our headquarters in Affi (VR), the second edition of Digital Content Factory Revolution, the event launched in June 2023 and designed to share annual news and reflections on the revolution that AI is bringing to the processes of content production and distribution in the fashion industry.
Actually, the first integration of an AI algorithm in Chalco’s engine dates back to 2018. This was the first auto-tagging service that automatically generated tags and metadata from product images. The service has evolved over the years and is now a vital resource for the brand that has adopted and perfected it over time.
Six years have passed but it feels like twenty. The progress we have seen the major AI engines make, especially in the last year, is incredible so let’s go back to the main highlights of the event.
An intense day, but one that we feel like telling you about, albeit in extreme summary. Intense because the topic is complex but also because it is not easy to maintain the right balance between overview and pragmatism in AI. This was the challenge we set ourselves at the event, we hope we succeeded and we will try again in two articles: this one, dedicated to insights of a more general nature and method, and a second one (read more here) in which we recount some practical applications presented at the event of AI services already available in the Chalco BrandLife platform.
AI: our scope of application
First general note: the perimeter of application. AI potentially touches every area of business, we focus on the contribution AI can bring to content production and distribution processes. But let us better understand what fits into this perimeter.
If by content we mean any asset, information or data related to a product master structure (read more here), we immediately understand that the scope of AI in this case is particularly delicate since it directly insists on what the brand has most valuable, the product. Even more precisely on the assets, information and data through which the product will digitally present itself in the contexts of sales (B2B, e-commerce, marketplace, …) and promotion (digital marketing, social media, …) omni-channel.
AI is not synonymous with magic
Having realised that we are dealing with a critical issue for business success, we come to the second general note: AI is not synonymous with magic. AI does not come down the chimney and lie lightly on top of existing processes by simplifying or replacing them without adequate preparation and integration. Planning is essential and with it clarity of the perimeter and process within which the service will fit.
AI as a process resource
Questioning process, we enter into the third general note: AI as a process resource. If AI is not a magic wand, then what is it? In our view, the one according to which content production can be likened to a factory process (more on that here), AI intervenes in that process as a resource that, like people or tools of various kinds, optimises or replaces a workflow or series of workflows managed manually today. Often AI receives input from other process steps and returns output that in turn will be processed in subsequent steps.
Training and quality control: human stuff
In the wake of the conviction that AI is not magic, we have repeatedly emphasised the importance of human input at two fundamental moments in the use of an AI engine. The first is the initial implementation phase of the service, a phase which often requires human intervention, but which is necessary for the training of the algorithms through the correction and thus progressive reduction of errors and the gradual routing towards autonomy of the service.
The second, fundamental moment of human intervention in the results of AI services is quality control. Particularly in the context of content generation or manipulation, and even more so in the fashion context, processing results are almost never perfect. We refer to inaccuracies of a technical nature, think of post-production for instance, but also to inaccuracies linked to the sensitivity and peculiarity of the brand and its creative direction. The human component is destined to diminish over time, but a photograph to date obliges us to emphasise that it is still necessary, and often its underestimation becomes the cause of the failure or interruption of many AI-related projects.
Agnostic approach to AI services: Chalco AI Gateway
The fifth and final general note has to do with the topic of integrations and more specifically with the approach we decided to follow by designing Chalco AI Gateway, the centralised integration engine for all AI services in the Chalco BrandLife suite.
AI is evolving at astronomical speeds, new services are constantly being created and more and more specialised, and it follows that often the best result can be obtained by combining different services. It is precisely for this reason that Chalco AI Gateway was created with an agnostic approach to the individual AI service and is rather result-oriented. Thanks also to the consulting support provided, when needed, by our Hyphen-Formula team, Chalco AI Gateway now transparently integrates and modulates different AI services according to the specific workflow and expected output.
