SAP has announced an agreement to acquire TechWolf, an AI platform that links tasks, skills and changes in the labour market. Completion is expected in the fourth quarter of 2026, subject in particular to regulatory approvals.[1] The deal raises a practical question: how can training and internal mobility be better prepared without reducing an employee’s abilities to the information recorded in software?
A graph to bring work and skills together
TechWolf draws on the HR and business systems already used by the organisation to build a context graph. This graph links the tasks associated with roles, skills, labour market information and company strategy.[1]
The aim is to match the activities actually carried out with the skills that can be associated with them. This approach seeks to complement inventories based solely on job titles.
In an IT team, for example, the same job title can cover user support, cloud administration or application integration. Skills may also be used beyond the usual boundaries between departments.
SAP is preparing closer integration with its HR ecosystem
SAP plans to integrate TechWolf into SuccessFactors and to use this information in scenarios linked to Joule, notably for workforce planning and talent management. New products are expected once the deal has closed.[1]
This integration could bring skills mapping closer to concrete decisions on training, mobility or how teams evolve. Its usefulness will, however, depend on the quality of the data and on how the results are interpreted.
Four uses to test against the quality of the information
This editorial grid presents possible benefits and the checks they call for. It does not describe gains already measured at a customer.
| Use | Intended operational effect | Control question |
|---|---|---|
| Training | Identify the know-how to develop for an activity. | Is the need confirmed by the person and their team? |
| Internal mobility | Identify skills that can be transferred between assignments. | Are aspirations and real experience taken into account? |
| Recruitment | Distinguish internal resources from the needs to be filled. | Is a skill missing from the system really missing? |
| Evolution of jobs | Analyse the tasks likely to change with AI. | Are the constraints and responsibilities of the work represented? |
An inferred skill remains a suggestion to be checked
Digital traces are incomplete. The activities visible in software can obscure informal mutual help, incident resolution or rarely used skills. They are therefore not an exhaustive measure of an employee’s value.
TechWolf states that, by default, employees can accept, reject or add to the skills suggested by the AI. The organisation can also opt for validation by the manager. The tool distinguishes between skills declared by the employee, assessed by the manager and inferred by the AI.[2] This information should not be merged into a single score.
For SMEs, start with a specific training need
An SME could map the tasks and skills involved in migrating an application to the cloud, then compare the results with employees’ experience to build a training pathway. This is a working scenario, not an SAP offering specifically announced for SMEs.
A limited pilot would make it possible to measure the relevance of the suggestions, the corrections made and the time needed to validate them. These results could then be compared, against the integration cost, with an approach based on interviews and a shared framework.
Before launch, the organisation should specify the data sources, access rights and the decisions the tool is meant to inform. A training recommendation must not become an individual assessment by default.
Data quality determines decision quality
The Data Management skill starts with identifying the sources, how up to date they are and what they actually make it possible to observe. Missing data does not mean that a skill is absent.
Teams must also harmonise vocabulary, preserve the provenance of information and distinguish people’s own statements from inferences produced by AI.
Finally, governance requires coordinating HR teams and users, defining how validation works and dealing with errors. The value of a skills graph depends on results that are reliable, explainable and useful to the people concerned.
Preparing jobs better without boxing people in
Mapping work at task level can help prepare training and mobility. It must, however, leave room for experience, aspirations and skills that are not visible in systems. The graph can inform a discussion about work; it should not replace it.
Further reading
Four CCCLX reads on data, data governance and integrating AI into organisations.
Talk to your database: Oracle improves its NL2SQL AI
Databricks extends Delta Sharing to AI and aims to break down data silos
Merck and Google Cloud: 1 billion for agentic AI
Two months, 1.1 billion dollars: River AI is already exciting investors
Sources
- 01
SAP News Center (6 October 2026). “SAP to Acquire TechWolf, Giving Enterprises Evidence-Based View of Work in the Age of AI”. Announcement of the agreement, the context graph and the planned integrations with SuccessFactors and Joule.
- 02
TechWolf (accessed 6 October 2026). “Who validates the AI’s suggestions, the employee or their manager?”. Explanations of validation by the employee, governance options and the different skill assessments.
Europe accuses Anthropic of discrimination after Fable 5 is blocked
A US decision that sparks an international crisis The sudden suspension of access to Anthropic’s Fable 5 and Mythos 5 models outside the United States now goes well beyond technology. What initially looked like a national security measure is taking on a major...
Mobi Fold and Spotlight 2: Logitech aims to simplify everyday hybrid working
The office is everywhere, but the tools do not always keep up Hybrid working has profoundly changed professional habits. Video-conference meetings, frequent travel, days split between home, the office and coworking spaces: employees now move between multiple...
+80%: agentic AI triggers a record surge in enterprise SSD prices
Artificial intelligence is putting new strain on infrastructure Artificial intelligence is often associated with graphics processors and massive computing centres. Yet another essential component of digital infrastructure is coming under unprecedented pressure:...
Databricks extends Delta Sharing to AI and aims to break down data silos
Data is no longer enough: intelligence must now be shared too Artificial intelligence is transforming the way companies use their data. Yet one major obstacle remains: the resources needed for AI projects are often scattered across different platforms, teams and...
RTX Spark: Nvidia accelerates the shift to AI-optimised PCs
Nvidia wants to redefine the PC for the age of artificial intelligence Unveiled at Computex 2026, the RTX Spark chip marks a new stage in the evolution of personal computers. With this Arm-based platform, Nvidia is not simply seeking to offer a more powerful...
WiFi 8 is arriving sooner than expected: what it means for businesses
A new generation of networking is already on the horizon While WiFi 7 is only just beginning to spread among businesses and consumers, the industry is already preparing the next step. TP-Link has just unveiled the first WiFi 8-compatible router, confirming that the...
Webedia-Elephant turns to Gemini Enterprise to industrialise generative AI
IT outages become a board-level issue Long regarded as mere technical incidents, service outages are now taking on a whole new dimension. According to a study by Oxford Economics for Splunk, Global 2000 companies suffer an average of 60 downtime incidents a year, at...
$600 billion lost: IT outages take on a new dimension
IT outages are becoming a boardroom issue Long regarded as mere technical incidents, service interruptions are now taking on a whole new dimension. According to a study conducted by Oxford Economics for Splunk, Global 2000 companies suffer an average of 60 downtime...
Lightwell: IBM and Red Hat tackle the open source patching headache
AI finds flaws faster than companies can fix them Artificial intelligence is profoundly transforming cybersecurity. While it makes it possible to detect vulnerabilities more quickly, it also creates a new challenge: the ability to fix a growing number of flaws within...
AWS strengthens OpenSearch to connect AI to enterprise data
AI agents are forcing the cloud to evolve The rise of agentic AI is profoundly transforming cloud infrastructure. Unlike traditional conversational assistants, intelligent agents must access data, interact with several systems and carry out tasks autonomously. To meet...
