I saw an interesting job opening on LinkedIn from the IT giant NTT DATA Europe & Latam. They are looking for a SAP Digital Cloud Architect Advisor. And you know, reading the description, I caught myself thinking that they are not quite looking for a human. They are looking for an ideal algorithm that, for some reason, must have a human passport and ask for a salary.
First, who is NTT DATA? This is not a garage startup. This is a titan, part of the Japanese NTT corporation with a turnover of 30+ billion dollars, serving 75% of companies from the Fortune Global 100 list. When a client comes to them, it is usually a huge behemoth with a complex IT infrastructure and a desire to migrate their SAP to the cloud.
What pain are they trying to solve with this vacancy? Imagine a dialogue somewhere in their office:
— Karl, we have a new client, a large retailer from Germany. They want to move to S/4HANA in the cloud. We urgently need a migration plan, architecture, and cost estimate.
— Hans, I currently have three equally “urgent” clients. This retailer has custom developments with a 20-year history, a zoo of systems, and special cybersecurity requirements. I need to review all the documentation for RISE with SAP, compare it with Azure and GCP capabilities, conduct five workshops with them to extract all requirements, and then spend another week coordinating prices with our department. Give me at least three weeks!
— We don’t have three weeks. Competitors have already sent them their proposal.
Here it is, the pain. The process of preparing a technical and commercial proposal for a complex SAP cloud migration is long, expensive, and requires unique specialists who are always in short supply. Each client is unique, but the steps are essentially the same: analysis, design, estimation, document preparation. NTT DATA is looking for a person who will be this “bottleneck,” a translator from business language to technical language and back.
What if I told you that 80% of this work could be done not by a human earning €100k+ per year, but by a well-configured AI-based system?
Let’s call it, for argument’s sake, the “Architect-Assistant.” This is not just “ChatGPT, write me a migration plan.” This is a specialized system trained on specific data and processes.
How does it work?
Step 1: Knowledge Base Creation. We “feed” the system all available documentation: technical guides for RISE with SAP, price lists, descriptions of AWS, Azure, and GCP services, NTT DATA’s internal regulations on cybersecurity and compliance, and most importantly — dozens and hundreds of anonymized successful projects and commercial proposals from the past. All of this is indexed and transformed into a vector database, ready for instant search.
Step 2: Integration and Automation. The “Architect-Assistant” integrates with internal tools mentioned in the job description, such as the Quote Tool and Configurator. Instead of a human manually clicking in the interface, they can simply write a query: “Calculate the cost for client X with such-and-such parameters and three years of support.”
Step 3: Workflow.
Instead of a human architect, the sales manager interacts with the “Architect-Assistant.”
1. The manager uploads the initial client request and records of initial meetings into the system.
2. The system analyzes the information, compares it with thousands of cases in its database, and instantly generates a list of clarifying questions for the client. In essence, it creates the ideal agenda for that very “technical workshop.”
3. After receiving answers, the AI generates several architecture options in minutes (e.g., “optimal price on Azure,” “maximum performance on GCP,” and “balanced hybrid option”), a draft migration plan, and a preliminary cost estimate.
4. Ultimately, the system delivers a ready-made presentation for the client’s CIO and a detailed technical document for their IT department, already formatted according to NTT DATA’s corporate standards.
“But what about human interaction, experience, intuition?” you might ask. And I will answer: we are not removing the human; we are making them significantly more effective. That expensive and experienced architect they are looking for no longer spends weeks on routine data collection and drawing diagrams in Visio. They receive an almost ready-made solution from the AI and spend their valuable time verifying it, fine-tuning it, and most importantly — engaging in live communication with the client, where, armed with the full power of AI analytics, they act not just as a consultant but as a true visionary. One such expert can manage not three, but ten projects simultaneously.
How to overcome distrust and implement such a system?
Start with “shadow mode.” Let the system work in parallel with a new employee. At the end of the week, compare the results: speed, completeness, accuracy of proposals generated by the human and the AI. I am confident the results will surprise you. Then, the AI can be transitioned into the role of a “junior assistant” who prepares drafts for the live architect.
How to validate AI results? Very simply.
1. Comparison with reality. Take 10 completed projects. Give the AI the initial data for them and ask it to generate the architecture and plan. Compare what the AI proposed with what was actually implemented. Analyze discrepancies.
2. Feedback from the implementation department. After an AI-planned project moves into the implementation phase, the implementation team provides feedback: how accurate was the initial estimate, what risks were not accounted for. This information is invaluable fuel for retraining the model.
3. A/B testing. Launch two pre-sales preparation streams: one — the old way, with people; the second — with the help of an AI assistant. After a quarter, compare metrics: average proposal preparation time, percentage of won deals, profitability. The numbers don’t lie.
Ultimately, NTT DATA, by hiring one person for this role, solves the problem locally. By implementing an AI system, they solve it systematically, scaling their expertise and gaining a colossal competitive advantage. The only question is whether they are ready to hire one AI implementation engineer instead of five SAP architects. Judging by the job description, not yet. But the market will force them to.
Источник: https://www.linkedin.com/jobs/view/4392674216/
