Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteYes, according to SAP’s chief AI strategy officer, Sean Kask, who argued at the Wave by Vento conference in Turin that established software companies must rebuild their products around AI or risk failing. “Every software company has to become an AI company, or they’ll perish,” he said, in remarks reported by The Next Web on October 8, 2026. Kask’s claim is a strategic argument about how software should be designed and what data it should run on. It is not a measured forecast that any particular company will fail.
What Kask means by “becoming an AI company”
Kask’s point is that adding AI as one more feature to an existing product is not enough. He made the argument in a session moderated by CNBC’s Carolin Roth, titled “The Model Europe Built,” and the reporting paraphrases his position as companies needing to “rebuild their products around AI, not add it as one more feature.” That paraphrase is the reporter’s wording, so the direct quotation is the sentence above.
As an Amazon Associate I earn from qualifying purchases.
The example the report gives is an interface where a user asks a question in plain language and the system produces the tables they need on screen. The distinction is about how people interact with software and what sits underneath it, not about rebranding or attaching a chatbot to a menu. That reading follows the examples in the article. SAP has not published a formal definition of an “AI-first” product in the material reported.
Two ways to build AI into a software product
The table below is an editorial framing of the contrast Kask draws, not a vendor classification.
#1 Best Overall
| Question | Bolt-on AI feature | AI-first redesign (as Kask describes it) |
|---|---|---|
| Where the user starts | An existing screen, with an assistant panel added | A plain-language request that drives the whole task |
| What the output is | Usually text or a suggestion beside the existing output | Generated tables and views the user works with directly |
| Dependence on data | Often limited to what the existing screen exposes | Depends on structured enterprise data and how systems connect |
| Main test | Whether users open the feature | Whether the generated result is accurate enough to act on |
Why SAP is focused on tables
Kask separated two kinds of model. Language models learn by predicting text. Tabular models work on structured rows and columns and are used for numerical prediction and classification. He argued that tables carry a disproportionate share of business value and that language models alone do not serve them well.
SAP’s internal tabular models and SAP-RPT-1
According to the report, SAP had built tabular models internally for a couple of years, and Kask said SAP uses SAP-RPT-1 in production. The article does not describe the production scope, customers, or performance measures for that model.
The Prior Labs deal
The report says SAP’s rationale for acquiring Prior Labs was its tabular-model work and overlapping research with SAP. The figures below are as reported by The Next Web and attributed to Kask or the article; the report does not include the benchmark results, methodology, or transaction documents.
- Completion: reported as July 2026.
- Investment: reported as more than €1 billion over four years, to develop Prior Labs into a frontier AI lab.
- Benchmark: Kask reportedly said the Prior Labs model led the TabArena benchmark.
- Applications: reported as cancer diagnosis and bank transactions.
- Structure: Kask reportedly said SAP will keep Prior Labs as a separate lab with room for its own research, and keep its model open weight for researchers and startups.
Kask also compared the approach with older workflows. In his account, one foundation model could do in days work that previously required teams to train and tune many task-specific models over weeks or months, and it beat methods such as XGBoost on accuracy. No test conditions or scope for that comparison are given, so treat it as his claim rather than a measured result.
Rank #3
Why SAP does not build its own large language model
According to the article, SAP does not build its own large language model, citing cost and convergence. Kask said customers can use models from Google, OpenAI, Anthropic, and Mistral. SAP uses more than 100 models internally and tests each use case against them to choose the best fit. The report does not state an evaluation protocol or a provider list beyond the event date.
Where Kask says Europe should compete
Kask argued that Europe should avoid competing mainly in what he called a “red ocean,” the crowded race to build ever-larger language models. He said the more useful investment is in assets that make models useful, such as enterprise data and systems. As an example, he said SAP’s knowledge graph links 500,000 tables and 7 million fields, which he said helps make agents accurate enough for workplace use. The report also notes that Mistral had launched Large 4, described as a one-trillion-parameter open-weight model.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.The figures Kask cited, and how to read them
The following statements were made or reported at the event and attributed to Kask in 2026. The report does not include the underlying methodology, so they should be cited as statements made by SAP’s AI strategy lead, not as established sector-wide statistics.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →- “About 80% of business data is unstructured.”
- “The 20% held in tables generates 80% of a company’s value.”
- SAP’s knowledge graph links “500,000 tables and 7 million fields.”
- “Every dollar of SAP software sold generates $6 to $10 for its ecosystem of partners.”
- SAP uses “more than 100 models internally.”
How to test the claim for your own software business
Kask’s argument is a useful checklist even for companies that are not SAP. The steps below are an analytical framework drawn from the strategies described in the report.
- Identify the task that would change. Pick one workflow where users currently move between screens to assemble a result. A redesign only matters if it removes steps for a specific user.
- Check whether your data is structured enough. Kask’s case rests on tables and linked enterprise systems. If the relevant data sits in documents or unconnected systems, a conversational front end will be only as accurate as the data underneath.
- Evaluate models on your task. SAP’s stated method is to test each use case against several models and pick the best fit. Vendor benchmarks, including TabArena results, are not a substitute for a test on your own data.
- Decide what you can control. Compare how much you depend on a single provider, whether a model can be run openly, and how it connects to existing systems.
- Measure user benefit. Track whether the redesigned workflow reduces time to a usable result or error rates, rather than whether people try the new feature.
Where the partner economics fit
For software companies, the partner side of Kask’s pitch matters. The article says SAP licenses startup technology into its own products and resells partner offerings through its store. The partner-economics figure above is a claim made at the event; any program terms, eligibility, or commercial conditions should be checked directly with SAP.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
