Episode 36: ConductAI, The AI Operating System For Your Enterprise IT - JP Haas

Conduct AI reads SAP code to explain stalled orders

Conduct AI reads the custom code and configuration of enterprise systems to explain why they behave the way they do. Its founders came out of Palantir, where SAP was always a core system, and saw that IT teams did not understand logic accumulated over 20 years while long S/4HANA projects blocked business improvements. Because the intent behind exceptions is rarely documented, Conduct turns code into business capabilities and captures the missing reasons from users.

Takeaways

• Code holds the truth about SAP systems. Conduct was founded by three former Palantir leaders, and its tagline is that the truth is in your code. It reads custom code and configuration to show what makes a system unique without comparing customer data between instances.
• Long S/4HANA projects stall business improvements. IT teams Haas met were starting S/4HANA transformations that would keep them busy for years, and none had a good picture of their own system. A German defence company could not make more tanks for Ukraine because optimizing its production line had to wait.
• Business intent behind custom rules goes undocumented. A rule sending trucks over two meters high to gate B existed because of a low gate or a downstream constraint. The reason lived in the heads of the requester or the consultant, so Conduct asks users to build a knowledge layer.
• Small ABAP program gives transparency in 24 hours. Conduct installs one small ABAP program on the customer's SAP system and works out of the box. It has been used by Fortune 500 and Fortune 100 customers across every lifecycle stage, before, during and beyond transformations.
• Code reading finds what manual reviews miss. One customer's order type took six days instead of two because two edge conditions in custom code interacted. A performance issue that had been investigated for more than a week was diagnosed and fixed in an afternoon.
• Consulting must shift from days to value. Haas says consulting has to move from charging for manual work to creating as much value as possible in as little time as possible. Langguth argues tools push consultants back into offering options and trade-offs, and judgement remains essential.

How Conduct's fit-gap workflow runs

1. Generate a checklist of requirements from what the system does today, reading code instead of interviewing users.
2. Filter the critical requirements.
3. Move back to SAP standard where possible.
4. Complete the S/4 template.
5. Implement extensions, with autopilot suggesting a clean-core-conforming option alongside alternatives.