1.0

SAP and AI

SAP AI Agents in Pharma: What Joule Changes, and the Team It Needs

SAP AI agents are software assistants built into SAP that can carry out work, not just answer questions: compare supplier bids, chase a blocked batch release, prepare a financial close. SAP launched them at scale in 2026. Novartis is already piloting one, and Takeda has announced plans to use SAP's agents in regulated manufacturing.

By Suleman AsifNovember 8, 20269 min read

A project team of five sits along a white table, one woman pointing a pen at a laptop screen while her colleagues look on.
SAP's agents are ready to try. Most pharma teams do not yet have the people to run them safely under GxP.

1.1

What SAP launched

Joule agents in 2026, in plain terms

The hard part for a drug maker is not switching them on. It is finding people who can build, connect and validate agents in a GxP setting, a skill set that barely shows up in job ads yet.

SAP's AI assistant is called Joule. Through 2026, SAP turned it from a chat window into a layer that runs business work.

  • May 2026, SAP Sapphire. SAP announced its "Autonomous Suite": more than 50 Joule Assistants across finance, supply chain, procurement, HR and customer experience. Each assistant coordinates a set of specialized agents, drawn from more than 200, that carry out individual tasks. SAP also launched Industry AI, seven solutions with industry-specific process logic and regulatory requirements built in.
  • Building your own. Joule Studio 2.0, launched at the same event, lets developers build their own agents (CIO reported they can use Python, Claude Code or Cursor) and run them on SAP-managed infrastructure. CIO reported it was available free through the end of 2026.
  • October 2026, SAP Connect. CIO reported that general availability of the assistants, agents, Joule Studio and Joule Work was beginning that month, offered as SAP-managed products through SAP's SAP for Me portal.
  • Late October 2026, SAP TechEd. SAPinsider reported that the keynote counted 300 SAP agents.

Access depends on how you buy SAP. CIO reported in May that customers on RISE with SAP get three Joule assistants activated in their first year, while GROW with SAP customers get the full portfolio of agents when they sign up. Individual AI features have their own terms. SAP's product page for AI in its Batch Release Hub for Life Sciences, for example, says it is included with Joule Base and requires a Joule Base or AI Units subscription.

Sources: SAP News, SAP Unveils the Autonomous Enterprise (May 2026); CIO, SAP's biggest AI bet yet (May 12, 2026); CIO, SAP advances its autonomous AI agenda (October 6, 2026); SAPinsider, SAP Joule Studio: Building and Deploying Agents (October 30, 2026); SAP, Batch Release Hub for Life Sciences, batch release processing.

1.2

Pharma is already testing

Novartis, Takeda and batch release

Two large drug makers have gone on record.

Novartis is piloting the Joule Sourcing Assistant. In SAP's October 6, 2026 announcement, Christoph Buerki, Head of Procurement at Novartis, said: "We've co-developed and recently launched a pilot of SAP's Joule Sourcing Assistant. Our vision is for AI agents to handle the manual heavy lifting, enabling sourcing professionals to focus on strategy, supplier engagement and value creation." Sourcing is a sensible first step: the agent prepares the analysis, and a person still makes the decision.

Takeda plans to use Industry AI in regulated manufacturing. According to SAP's November 3, 2026 announcement, Takeda will use it in GxP environments, with agents that monitor production KPIs, flag deviations and start corrective workflows. That is a step closer to the quality system, where the rules are strictest.

Batch release is the obvious target. Releasing a batch means pulling quality, production and supply data from many systems before a qualified person signs off. In 2022, SAP said its work with leading pharma companies showed that a working group of quality control professionals takes around 40 hours to reach the release decision for a single batch. SAP's batch release software already uses Joule to surface the batches that need investigation because a release check is blocked.

Sources: SAP press release, SAP Puts the Autonomous Enterprise to Work (October 6, 2026); SAP News, Gartner Emerging Market Quadrant for integrated supply chain AI agents (November 3, 2026); SAP News, Turbocharging batch release processes in the life sciences industry (July 2022).

1.3

The GxP question

What regulators allow AI agents to do

In most companies an AI agent that makes a mistake causes rework. In pharma, an agent that touches GxP data or a release decision can cause an inspection finding. The rules for AI are still being written, and that uncertainty is itself a reason to involve the right people early.

  • United States. The FDA's guidance on using AI to support regulatory decisions for drugs and biologics, published in January 2025, was still marked as a draft on the FDA website when we checked on November 8, 2026. It asks companies to define the context in which a model is used and to assess its risk and credibility.
  • United States and Europe together. In January 2026, the FDA and the European Medicines Agency published ten guiding principles for good AI practice in medicine development, including traceable data and documented processing.
  • Europe. A new EU GMP annex on AI, Annex 22, went out for consultation in 2025. The draft says generative AI, large language models and models with probabilistic outputs "should not be used in critical GMP applications". As of November 8, 2026, the EU's official GMP guidelines page still lists no Annex 22, so the text is not final. It matters to any US company that makes or ships medicines for the EU.

The practical result: agents that prepare, summarize and route work are easier to justify than agents that make GMP decisions on their own. Wherever an agent touches validated systems, it needs a defined intended use, testing, human oversight and a validation record. Our guide to GxP compliance hiring for IT and validation staff explains how validation teams work around these systems.

Sources: FDA, Considerations for the Use of Artificial Intelligence to Support Regulatory Decision-Making for Drug and Biological Products (draft, January 2025); FDA and EMA, Guiding Principles of Good AI Practice in Drug Development (January 2026); EMA, report of the multi-stakeholder workshop on Annex 22 (June 30, 2026); European Commission, EudraLex Volume 4 (checked November 8, 2026).

1.4

The skills gap

Why most SAP teams are not ready

SAP customers themselves say people are the bottleneck. In research by the Americas' SAP Users' Group (ASUG), fielded in May 2026 with 175 members, lack of internal skills and resources was the most cited barrier to advancing data and AI strategy, named by 34% of respondents. Among organizations with only an emerging data strategy, the figure rose to 43%.

Adoption is also at an early stage. In separate research by Precisely and ASUG, published in November 2026 and based on 169 SAP stakeholders, only 3% reported fully automated, AI-driven processes, 46% used AI for a mix of automated and human-assisted tasks, and 51% described themselves as early explorers of AI-assisted development.

Job ads tell the same story. In our pharma IT hiring index, a count of US Indeed postings on November 8, 2026, "Joule" appeared in zero pharma job ads and "SAP BTP" in four, while about 1,000 pharma ads mentioned SAP. Companies are not yet advertising for these skills by name, which usually means they will look for them inside existing SAP teams, consultancies or a handful of early specialists when projects start.

Sources: ASUG, How SAP Customers Are Closing the Data and AI Execution Gap (research fielded May 2026, 175 ASUG members); Precisely, SAP automation trends for 2027 (November 2026, 169 SAP stakeholders).

1.5

The team

Who a pharma SAP AI project needs

An SAP agent project in a regulated company is not one hire. It pulls on several skill sets that rarely sit in the same person.

Roles behind an SAP AI agent project in pharma, and why each one is hard to fill
RoleWhat they doWhy it is hard to fill in pharma
SAP BTP and Joule Studio developerBuilds and extends agents and the apps around them on SAP's Business Technology PlatformThe tools are new in 2026, so few people have production experience
Integration architectConnects SAP agents to non-SAP systems such as LIMS, MES and quality systems, and to other vendors' agentsPharma companies run many validated systems side by side, and each connection needs control
SAP process lead (procurement, quality, supply chain)Decides what an agent may do in a process, and where a person must approveNeeds deep process knowledge and the authority to say no
CSV and AI validation leadDefines intended use, risk, testing and records so the agent can be used in GxP workAI validation guidance is still draft, so judgment matters more than templates
Security and access specialistControls what data and actions an agent can reach, and keeps an audit trailAgents act with system permissions, so a mistake has real reach
Data engineerSupplies clean, traceable data that agents can rely onAgents are only as good as the data under them

The table shows the shape of the team, not a hiring method. In practice, the hardest profiles are the ones that join two worlds: an SAP developer who has worked under validation rules, or a validation lead who understands how an AI agent behaves. Our guides to SAP staffing for pharma and data engineering staffing cover two of these roles in more depth.

1.6

Timing

Why the SAP AI and S/4HANA clocks are linked

SAP sells its new assistants and agents as SAP-managed cloud products, and its activation offers are tied to RISE with SAP and GROW with SAP, the same programs many pharma companies are using to move off ECC. ECC loses mainstream maintenance on December 31, 2027. That puts AI agent plans and S/4HANA migrations on the same timeline and in competition for the same people.

Companies that plan the two together can design processes with agents in mind from the start, instead of migrating first and reworking later. Companies that plan them separately often find the same handful of experienced SAP people booked on the migration when the AI project needs them.

1.7

The hard part

Why these hires are difficult to judge

SAP offers courses and certifications for Joule and its Business AI tools, so AI keywords are easy to add to an SAP CV. That makes the screening harder, not easier. Knowing what Joule can do is different from having connected an agent to a validated quality system and defended it to an auditor.

Telling those apart takes a detailed conversation with someone who understands both SAP and GxP, and time that most SAP program teams do not have during a migration. That is the specialist work Geeqers does before you see a single name.

We recruit SAP, cloud, cybersecurity and AI specialists for US pharma and biotech companies, with a shortlist in 7 business days. Our fee is only payable after your hire starts, agreed in writing up front, and every placement includes a 90-day free replacement. Searches can be run confidentially. If an SAP AI project is on your roadmap, send us a hiring brief and we will tell you what the market looks like for the roles you need.

1.8

FAQ

Frequently asked questions

What are SAP AI agents?

SAP AI agents are software components built into SAP's applications that carry out specific tasks, such as analyzing supplier bids or reconciling accounts. In SAP's 2026 model, Joule Assistants coordinate several agents to run a larger process, and people interact with them through Joule. SAP announced more than 50 assistants and more than 200 agents in May 2026.

Are pharma companies using SAP Joule agents yet?

Yes, in early stages. Novartis said in October 2026 that it had co-developed and launched a pilot of SAP's Joule Sourcing Assistant. SAP said in November 2026 that Takeda will use its Industry AI in regulated manufacturing, with agents that monitor production KPIs and flag deviations in GxP environments.

Can SAP AI agents be used in GxP processes?

Only with care. The FDA's AI guidance for drugs was still a draft as of November 8, 2026, and the EU's draft Annex 22 says generative AI and large language models should not be used in critical GMP applications. Agents that prepare and route work for human approval are easier to justify than agents that decide on their own, and any use in GxP work needs validation.

What skills does an SAP Joule project need?

Typically SAP BTP and Joule Studio development, integration with non-SAP systems, process ownership in areas such as procurement or quality, AI validation, security and access control, and clean data. In pharma, the scarcest people are those who combine SAP technical skills with experience of GxP validation.

How do companies get access to Joule agents?

SAP offers its assistants and agents as SAP-managed cloud products, and access depends on your contract. CIO reported that RISE with SAP customers get three Joule assistants activated in their first year and GROW with SAP customers get the full agent portfolio. Check the terms for your own systems with SAP or your SAP partner.

Image: "A diverse group of professionals collaborating on a laptop in a modern office workspace" by Andrea Piacquadio via Pexels, used under the Pexels licence. Dates, figures and quotes were checked against the sources linked above on November 8, 2026.

Keep reading

Get started

Tell us about the role. You’ll see our fee in writing before we start.

You’ll have a shortlist within 7 business days, and you pay nothing until your new hire starts.

Hire Talent

Looking for a role yourself? Find Jobs

A recruiter hands a hiring manager a candidate profile on a clipboard across the table.