To understand how we can use AI in enterprise migrations it’s worth looking again at why application system migrations have become amongst the most complex, expensive and precarious data intensive projects for any company to embark upon once it has been persuaded to take the plunge.
It amazes me how many times a large-scale systems migration project doesn’t appear to have a detailed plan. There’s a general lack of understanding about the challenges ahead. The cynic in me wonders if an oversimplified business case has been swallowed whole by the C-suite who have been wooed by attractive high-level proposals, shiny new features and a list of intangible benefits.
In the ERP/ CRM arena vendors are often culpable. As they look to cut costs, steer customers to new platforms and sunset old code, how much focus is truly devoted to the individual needs of their own customers? Large-scale-embedded business applications have evolved to the extent that they run almost every business activity and feed the performance measures relied upon by most departments. Vendors know this and are frightened to set agendas.
The applications vendor, the partner Systems Integrator (SI) or services team all too often, I believe, make migration sound much too easy. Of course they do. The doctor does the same for us when speaking to us about open heart surgery. It’s important not to frighten the patient! To achieve success, however, it’s essential that the right people involved really know the details.
While “vanilla” packaged software should be well documented by the vendor, a typical enterprise implementation is almost always poorly documented. When business processes change nothing gets updated. Knowledge about systems leave when people leave an organization. Over time, internal teams write thousands of custom reports, custom fields and bespoke data tables. These are deployed with minimum or zero documentation.
The lack of documentation creates severe bottlenecks.
- Invisible Code: You cannot map what you don’t know exists. Data teams regularly discover hundreds of custom tables during a migration that no living employee knew were active.
- Wasted Discovery Time: Business analysts often spend the first 3 to 6 months of a project conducting interviews and running manual SQL queries just to figure out how the current system; e.g. JD Edwards operates.
- Re-inventing the Wheel: Without documentation, teams waste budget rebuilding complex JD Edwards custom workflows for example in SAP S/4HANA, only to find out later that the business process is obsolete or could have been handled by standard SAP features.
Our incoming enquiries here at Silwood usually arrive at times when there are similar problems. Someone starts shopping for the elusive complex enterprise ERP application silver bullet.
To illustrate some of the key areas where Silwood adds value let’s consider a JD Edwards to SAP systems migration. Our flagship product - Safyr - accelerates migration projects by automating metadata discovery to reveal how data is structured in both systems.
Instead of guessing or manually digging through thousands of cryptic database tables, data teams use Safyr to reverse-engineer the specific, customized structures of both the source (JD Edwards) and the target (SAP) environments. Ultimately this means you can:
- Eliminate guesswork in cryptic data structures by helping to decipher table names: JD Edwards uses non-intuitive table codes (e.g. F0101for Address Book, F0911 for Account Ledger). Safyr provides you with the business friendly, logical names for these physical table names.
- Extract Customisations: Safyr pulls metadata directly from active Data Dictionaries of specific implementations, meaning it captures all tables including custom tables and fields. It never provides a vanilla template of the application data model.
- Define Scope: Migration teams can filter out irrelevant tables and create curated “Subject Areas” (e.g. just the tables relating to JDE Accounts Receivable).
- Identify Relationships: Safyr uncovers implicit parent-child relationships and data dependencies within JD Edwards. This tells the ETL (Extract, Transform, Load) team the exact sequence in which tables must be extracted to prevent broken links.
- Export to Data Tools: Rather than keeping metadata trapped in silos, you can use Safyr to export data models directly into industry-standard modelling and cataloging platforms. This helps architects map data visually.
- Reduce reliance on 3rd party experts: Traditionally, companies have had to hire expensive specialists in JD Edwards or SAP etc., just to explain where specific data sits. Safyr puts that data intelligence directly into the hands of your internal business analysts.
Migration teams without enough experience in ERP/CRM applications are often faced with trying to look inside an impenetrable black box before they can get started. The initial steps required in planning, scoping, specification and design simply can’t get off the ground. Delays ensue closely followed by finger pointing. The C-suite struggles to understand why everything has slowed down or come to a complete stop.
Safyr has proven to add value, delivering up to 90% productivity gain for migration teams, analysts and architects. As well as migrations, it can also be used to support many other data intensive projects.
However, with the advent of AI we’ve been exploring how we can help our customers go further.
AI has shifted user expectations in the modern enterprise. Conversational, natural language querying is moving from an optional luxury to a standard user interface requirement.
Instead of memorizing how to use a screen designed for an analyst, or how to navigate multi layered menus, users increasingly expect to interact with their systems by simply stating their intentions or asking questions in plain English. In response:
We have added new capabilities to enhance the value of Safyr by making your AI ‘customer metadata’ aware. This means that customers can ask AI questions about their ERP metadata as implemented.
Essentially, we believe that this gives our customers the ability to ask a broader range of questions of the metadata in their own systems that previously would have required more time and resource.
This goes beyond the natural limitations of the standard Safyr interface.
We would like to offer you an opportunity to look at the additional capabilities and value we are opening up by enabling Safyr content with AI.
Think what questions you would you like to ask your own ERP/CRM systems?
Examples might be:
“What are the important Business Views in my JD Edwards system?”
“We want to build a data mart that uses data extracted from our S/4HANAsystem. Here is the requirements document. Please give me a list per requirement of which Tables and Fields we need to access in S/4HANA to locate the necessary data in Excel format.”
“What are the core tables and views required to build a complete profile of a Vendor?”
“Where are PII/GDPR fields such as employee home addresses or banking details stored across my ERP?”
The list goes on!
Let us know what you would like to ask using the quick form link below:
If you want to learn more from a demonstration of Safyr with AI, or to ask us a question related to your ERP (from our list of supported ERP applications) please complete the Contact Form.