https://talsen.team/_next/static/chunks/7122.8177bb56311d1f53.js

1"use strict";(self.webpackChunk_N_E=self.webpackChunk_N_E||[]).push([[7122],{7122:function(e){e.exports=JSON.parse('{"blogArticle":{"ki-automatisierung-gefahrstoffverwaltung":{"intro":{"p1":"Twice a year, the same ritual: Procurement sends dozens of emails to suppliers, waits for responses, sends reminders, opens PDF attachments, compares columns, manually types in values – and finally creates an updated hazardous substances list by hand.","p2":"Anyone familiar with this knows: it\'s not an exceptional project, it\'s mandatory. Under German hazardous substances regulations, employers are required to maintain a register of hazardous substances used in the workplace. This register must be kept up to date. In practice, companies set their own review intervals – for example semi-annually or annually – and update the data additionally when new information, revised safety data sheets, or new substances are added.","p3":"Yet in many companies, this process still runs almost entirely manually – with correspondingly high effort and error potential. For one of our clients, we are currently implementing a solution that brings together individual building blocks we have already developed for other clients into a seamless process for hazardous substances management. What we observe in this work – and why many companies face similar challenges – we share here.","highlightTitle":"The most important lever from our project experience:","highlightText":"Anyone who wants to use AI sensibly should not start with the question of which technology is currently possible – but with the question of which process regularly costs time even though its logic has long been clear."},"observations":{"heading":"What we observe","cardA":{"title":"The process is structured – but still manual","p1":"The logic behind hazardous substance updates is clear: query suppliers → evaluate responses → identify changes → update own list. Yet every step happens by hand. Because the data arrives in different formats – sometimes as Excel, sometimes as PDF, sometimes as an email without any attachment."},"cardB":{"title":"Human interfaces multiply the effort","p1":"Every handover between procurement, administration, and the specialist department costs time and carries the risk of transfer errors. Anyone who has ever incorrectly transferred a hazardous substance label or classification knows what is at stake."},"cardC":{"title":"Follow-up is time-consuming and unstructured","p1":"Not all suppliers respond to the first request. Reminders are written manually, sometimes forgotten, sometimes sent twice. Systematic tracking is often missing."},"cardD":{"title":"Comparing old and new lists is error-prone","p1":"Anyone who manually compares two Excel tables with 200 rows will easily miss changes, because the human eye is not optimized for this task."}},"solution":{"heading":"What we are implementing for our client","intro":"We are developing a solution that supports the process from the initial request through to the draft of an updated hazardous substances list. The goal is not to replace professional responsibility. The goal is to structure and automate recurring work steps and to better prepare the specialist review.","blocks":[{"number":"1","title":"Automated email sending and receiving","text":"Personalized requests to suppliers are automatically prepared and sent. Incoming responses are automatically recognized, assigned, and processed further."},{"number":"2","title":"Intelligent document processing","text":"Incoming responses and documents are read out, structured, and converted into a uniform format."},{"number":"3","title":"Automatic list comparison","text":"The new supplier list is compared with the previous version. Changes are automatically flagged and submitted for review."},{"number":"4","title":"Automated follow-up","text":"If suppliers do not respond, reminders can be sent after defined time intervals. The status of all requests remains traceable: Who has responded? Where is information missing? Where does follow-up need to happen again?"},{"number":"5","title":"Draft of the updated hazardous substances list","text":"From the processed supplier data, a recommended updated list is generated. This is submitted to the responsible specialist for review, assessment, and approval."}]},"patterns":{"heading":"What underpins this approach: proven patterns","intro":"From comparable automation projects, we know: what matters is not just the technology, but above all a clearly structured process.","items":[{"title":"Structure before intelligence","text":"AI components work better when the surrounding process is defined. That\'s why the starting point is questions like: Who sends the request? What information is needed? What data counts as complete? When is follow-up triggered? Where is specialist review required?"},{"title":"Human approval at the right points","text":"Full automation sounds tempting – but especially with regulated content, a targeted human review step makes sense. The AI handles recurring, data-intensive work steps. The final assessment remains with the responsible specialists."},{"title":"Uniform data formats as a prerequisite","text":"Anyone receiving ten formats from ten sources needs a normalization layer. Only when data is available in a comparable form can changes be reliably identified and processed further."},{"title":"Building in traceability","text":"Automated processes must be auditable. Who sent which request when? Which response was received? Which file was processed? Which change was identified? And who approved it?"}]},"checklist":{"heading":"TL;DR – The key takeaways at a glance","items":[{"pre":"Rule-based, recurring processes are ","bold":"ideal AI candidates","post":"."},{"pre":"","bold":"Structure before intelligence","post":": define the process first, then automate."},{"pre":"Maintain human approval at ","bold":"professionally sensitive points","post":"."},{"pre":"","bold":"Uniform data formats","post":" as a basic prerequisite for reliable comparison."},{"pre":"Automated processes must be ","bold":"auditable and traceable","post":"."},{"pre":"AI does not replace expertise – it ","bold":"relieves routine work","post":"."},{"pre":"","bold":"Systematic status tracking","post":" instead of manual follow-up."},{"pre":"Start with the process, not with the ","bold":"technology","post":"."}]},"cta":{"heading":"Let\'s talk","p1":"Do you have similar processes in your company that regularly cost time even though their logic has long been clear? Many of these workflows can be structured, automated, and made traceable with the right building blocks – without giving up professional responsibility.","p2":"We look forward to the exchange."}}}}')}}]);