In February 2025 I was sat in Dublin Airport with two colleagues.
We were en route to SAP America's Newtown Square HQ in Philadelphia where I was due to deliver a talk before heading off to Las Vegas to watch Warrington Wolves play at the Allegiant stadium.
My talk that year was a ramble about the Industrial Revolution in the North West of England and what we could learn from it to take into SAP transformation projects. Off the wall, but it seemed to hit the spot for the 1,000+ people in the room and on livestream.
My 4-hour layover in Dublin could have been spent in two ways. One involved Guinness, the other involved working. As I was about to go on a long weekend in Vegas with Tony Cronin, Avtar Davatwal and Julian Rawlinson, work seemed the safest option.
Back in 2018, Julian Rawlinson and I spent the whole of December figuring out how best to guide customers through their S/4HANA journey. We had two clients in the thick of the challenge and we realised that it was close to impossible for SAP customers to build an effective business case. We'd just completed a research project at the UKISUG - UK & Ireland SAP User Group Connect event and the results had an interesting correlation:
Go figure.
How can you get buy-in to something that you can't even articulate?
Due to a random set of circumstances, we were also playing with Graph Databases at the time, understanding how they worked and using Graph Query Language to solve a problem in an unrelated area of the business. Stupidly (or sensibly in hindsight) I said to Julian:
"Why don't we build our own Graph Database that links up all the functionality in S/4HANA with business pain points?"
Or words to that effect.
18 months later, we'd launched FusionGraph, our own solution that we've used on every S/4HANA Phase Zero for the last 6 years. With FusionGraph, we can ingest a customer's SAP usage and visualise the new functionality that can provide a foundation for their business case. FusionGraph is what led us to be so successful with our S/4HANA business case work, proper competitive advantage when we come up against the Big 4 or an SI offering to do a "free Phase Zero".
I'm conscious we're now in the depths of a Christopher Nolan style time loop, so I'd better start extracting us back to the surface.
In FusionGraph, we'd accidentally created an AI foundation before ChatGPT thrust AI into our day-to-day lives. Graph Knowledge models were a very new thing and we'd built one.
So, during that layover in Dublin Airport in Feb 2025, I decided to connect the GraphDB we'd built for FusionGraph with ChatGPT. Essentially a Vector DB built on top of 6 years' worth of hands-on consultant and curation. By the time the flight boarding call came over the tannoy in a soft Irish accent, it was delivering astonishing responses.
Why? Because the data taxonomy we'd built was perfect for an LLM to interact with.
Through a ChatGPT interface, it was like talking to an SAP Consultant with 10 years' experience in every SAP module. Not SAP experience, but deep S/4HANA knowledge.
By May 2025 we'd started to use it within our Phase Zero engagements. Our consultants were using it to accelerate their understanding of niche areas of S/4HANA, validating their assumptions, finding blind spots. We started to throw customer pain points at it and it provided answers.
Like, really, really good answers.
By July it had an internal name. An identity. A personality.
Then, in September I said something else to the team...
"We need to turn this into a product that customers can use themselves to reduce their dependency on expensive SAP consultants."
So that's what we did.
When I look back over the past 15 months, although my email footer said "CEO and Founder" I'd been wearing CTO clothes. I stopped writing on here, because I was scratching my intellectual itch building something exciting. I went super deep on AI, and I learned things that are pretty alien in the mainstream SAP world.
A world where for decades, SAP experts have simply been doing what they did on the last project while the world evolves around them. A world where "thought leaders" harp on about SAP failures in their own high-volume echo-chamber, but don't offer solutions.
A world where "experts" say that moving to S/4HANA is too hard or impossible to justify, choosing to focus on how many points they'll lose by, rather than grabbing the game by the scruff of the neck and playing a part in turning the underdog into the victor.
At my age, in a big complex industry, I wanted to do something that would move the needle.
Last week I posted my Deshoring Manifesto.
"Deshoring is the deliberate replacement of costly consulting and systems integration services (whether onshore or offshore) with Agentic AI: intelligent, always-on, commercially independent AI agents that work directly for and with your team."
The opportunity with Deshoring is folding time, compressing the effort required to do the things that take the longest on SAP projects, and those require the most expensive skills.
That's the Deshoring sweet spot, the saving opportunity is more than 50% of your SAP consultancy spend.
In Feb 2026, we started quietly providing two Agentic AI solutions that enable Deshoring to a handful of pilot clients who have delivered massive reductions in cost and timescales to deliver SAP projects. We now have our first paying customers and an onboarding pipeline out to July.
S4SensAI is an agentic SAP Architect that knows more about S/4HANA than any human consultant, answers in seconds and cost 95% less than an SI consultant.
S4SensAI is built on our own S/4HANA knowledge base that powers FusionGraph. One of the frequently asked questions when we demo is "is it like Joule for Consultants?", kind of, but Joule for Consultants wears S4SensAI pyjamas at bedtime.
ABAPBanZAI is an agentic SAP Custom Code analyzer that can document an entire Z Codebase and reverse engineer functional specs into business language in a few days. It rebuilds two decades of RICEFs and identifies which business process each custom object supports rather than relying on names like "ZACTS03 - Analysis report version 3".
Feed S4SensAI the output from ABAPBanZAI and we can run fully autonomous fit-to-standard analysis on custom ABAP code to identify where custom functionality can be delivered (in full or in part) using standard S/4HANA functionality.
Code remediation moves all of your old ABAP code to new ABAP code, whether you need it or not. We can run clean-core fit-to-standard on your custom code and then spec (and write) the BTP replacement for the stuff you can't standardise, in days.
S4SensAI can define SAP solutions, solve SAP problems, write functional and technical specs, write interface specs, write ABAP Code (including RAP ABAP), write BTP code, create test scripts (and test data), it can build business cases in business language with support KPIs and ROI, and it creates project plans (and RAID logs and RACIs). It runs on Claude as an LLM and also integrates with Teams, SharePoint, Jira, Confulence, Miro, LeanIX and Microsoft Planner.
Now I'm back in CEO mode.
So, if you're in the market to reduce your SAP costs by 50%, get in touch. Ask the team to do you a 30-minute demo and show you just how powerful S4SensAI and ABAPBanZAI are.
Heading to Sapphire in 2026?
if you're at Sapphire Orlando or Madrid and you see me and my yellow shoes, stop me and I'll show you in person. Book time with me here.
Staying home instead?
If so, you can watch my keynote from this year's NextGen Enterprise Architect conference where I discuss Agentic AI for SAP delivery.