Roche

Ortent Life Sciences Radar, research beta. Evidence to 6 October 2026, rule set v0.4.3. Scope: AI and computation in drug discovery (T1), and research-lab digitalisation and scientific data systems (T2). Roche here covers Roche and Genentech. Chugai, which is separately listed with its own R&D, is out of scope under clarification C42-12 and was not searched. Each ID in brackets links to its online source, and the source index at the end of this brief lists them.

Disclosure: Andrew Wyatt, Ortent's principal, was Chief Growth Officer at Sapio Sciences until April 2026. Software vendors, including Sapio and its competitors, appear in this brief only where public sources name them.

Status: research beta. Ortent has reviewed this brief; it has not been independently validated. It summarises activity identified in public sources. Search coverage varies by company and is set out under Coverage limits, and zero identified items does not mean no activity. It does not predict buying intent or measure operational maturity. For this radar, process and CMC development are grouped with manufacturing and excluded from the research-lab theme (T2); such work appears as context where it helps.

Summary

The reviewed sources identify six commitment changes involving Roche or Genentech in the two themes: two first disclosed in the year to 4 October 2025 and four in the year to 5 October 2026 [>ROG-E39, >E05, >E16, >E08, >E02, >E40]. Five are tagged to AI in discovery (T1) only [>ROG-E39, >E05, >E08, >E02, >E40]. The sixth, a collaboration with Medra, is tagged to both themes because a Genentech story says Medra's platform integrates with Genentech's laboratory information management system; its event date is uncertain [>ROG-E16]. Company decks and stories describe a Lab-in-the-Loop programme, AI use across antibody and small-molecule programmes and an AI agent for target assessment, with no stated start dates [>ROG-E19, >E20, >E25]. The 6 October recruitment snapshot contains 29 relevant R&D requisition postings across the two themes, none with ambiguous vacancy status [>ROG-R01 to >R29]. These observations identify programmes and functions for further investigation; they do not establish procurement opportunity, deployment completion or net hiring.

Announced commitments

Roche or Genentech published four of these six items itself [>ROG-E05, >E16, >E08, >E02]. The other two were announced by the partner, with the Head of Roche Corporate Business Development quoted in both, and we found no Roche confirmation of either [>ROG-E39, >E40]. Medra's post of 11 December 2025 is the earliest public mention of the Medra collaboration, but a partner list alone does not establish a commitment; the Genentech story of 9 January 2026 is the first company statement [>ROG-E16].

The Genentech story text blocks on the CSCoE and Medra were created on 7 January 2026 and edited later (21 May and 23 July 2026), so their wording at publication is not verified [>ROG-E08, >E16]. The CSCoE launch is dated only to "2025", which spans the window boundary, so its event date is uncertain [>ROG-E08]. The Medra collaboration began at an unstated date between September and 11 December 2025, which is also uncertain [>ROG-E16]. Both count only in the first-disclosed view.

The Medra row is tagged to both themes. T2 rests on the company statement that the platform integrates with Genentech's ML infrastructure and laboratory information management system, which places it in the company's own workflows (AR7a). T1 rests on AI-run experiments in the lab-in-the-loop [>ROG-E16].

Earlier and later steps on the same initiatives are not counted as changes. The Annual Report 2024 already disclosed an existing NVIDIA collaboration on ML models [>ROG-E01]. The Pharma Day 2026 deck restates the AI factory with two new data centres, and this is excluded as a restatement of the March launch [>ROG-E03]. Roche announced the opening of the Boston centre on 17 September 2026, with a 10-year lease on 95,000 sq ft; this is a milestone [>ROG-E07]. Dyno announced on 13 January 2025 that Roche had exercised an option to license a Dyno capsid, for a USD 7 million fee, and said this concludes the initial 2020 collaboration [>ROG-E38]. The 2024 Dyno deal is recorded as a new commitment under its own parent, not an expansion, because the release does not say it extends the 2020 agreement (C42-1) [>ROG-E39].

The Pharma Day 2026 deck of 28 September 2026 shows a timeline of AI investments in phase columns (2020-2022, 2023-2024, 2025-2026). Its entries include gRED Computational Sciences consolidation, an AWS Research Agent collaboration, an Autonomous Lab Accelerator, a Lab-in-the-Loop 2.0 initiative, a Foundational Datasets initiative, Prescient Design expansion to all molecules, an OpenAI pilot partnership and Herophilus automation software [>ROG-E10, >E15, >E17, >E21, >E26, >E27, >E28, >E29]. Undated entries in a slide list are recorded as newly observed capability, not as commitments (C42-18, C42-14). One reviewer counted eight of them as commitments; on that reading the T1 W1 first-disclosed count would be 11 and the T2 W1 count 4, all with uncertain event dates. An entry naming "MHS" with Anthropic in brackets states no scope and is excluded [>ROG-E18].

By event date, we counted two T1 initiatives with a qualifying commitment change in W0 (5 October 2024 to 4 October 2025), the Boston centre and the second Dyno collaboration [>ROG-E05, >E39], and two in W1 (5 October 2025 to 5 October 2026), the NVIDIA AI factory and Earendil [>ROG-E02, >E40]. By first disclosure, the T1 count is two in W0 and four in W1, adding the CSCoE and Medra, whose event dates are uncertain [>ROG-E08, >E16]. T2 is zero in W0 in both views; in W1 it is zero by event date and one by first disclosure (Medra, which appears in both themes) [>ROG-E16]. We identified no qualifying negative change. Milestones disclosed number two in T1 in W0 [>ROG-E11, >E38] and six in T1 in W1 [>ROG-E07, >E12, >E13, >E22, >E23, >E32], with none in T2. Existing capabilities first disclosed number five in T1 and one in T2 in W0 [>ROG-E01, >E09, >E14, >E19, >E35, >E33], and ten in T1 and three in T2 in W1 [>ROG-E10, >E20, >E21, >E25, >E26, >E27, >E28, >E36, >E37, >E51, >E17, >E29, >E34]. These have no stated start date and are not counted as changes.

No comparison between years is made for any cell the coverage table marks as not comparable. It marks none, because each source family had the same search status in both years, but several families were only partly searched in both (see Coverage limits). On that basis the searches found two T1 changes by event date in each year, and more first disclosures in W1 than in W0 in both themes. This is a difference in what the searches found, not a measure of company activity. The adjudication notes add that the W1 investor material is richer, because the Pharma Day 2026 deck carries a dated AI timeline.

Under the opposite reading of C42-4 (accelerators, borderline), the Autonomous Lab Accelerator would be an accelerator launch with collaborations shown by logos, and a new commitment with an uncertain event date: the T2 W1 first-disclosed count would be 2, not 1, and on one reviewer's tagging of the programme to both themes the T1 W1 first-disclosed count would be 5, not 4. The adjudicator did not apply C42-4, because the programme is internal rather than a cohort or challenge, and the slide states no start, funding, resource or launch action [>ROG-E17]. The opposite reading of C42-2 changes no Roche count. Separately, the adjudicator kept the second Dyno collaboration in T1 as AI design of a component of the therapeutic before candidate selection, and left open whether capsid design is discovery under AR6, since the capsid is a delivery vector rather than the active molecule. If the Dyno deal were not counted in T1, T1 W0 would be 1 in both views [>ROG-E39]. If Earendil were also read as an asset deal without stated AI work for the company, T1 W1 would be 1 by event date and 3 by first disclosure [>ROG-E40].

Identified recruitment

On 6 October 2026 the two reviewers screened 167 and 90 requisitions listed on careers.roche.com and careers.gene.com; merged on requisition number, 184 are classified. We observed 29 relevant requisition postings in the R&D theme class, none with ambiguous vacancy status [>ROG-R01 to >R29]. Of these, 22 are T1, 5 are T2 and 2 are tagged to both themes. Roles tagged both appear in each split, so the T1 split is 24 and the T2 split is 7 (C42-19). A further 1 is a research fellowship, 8 are internships or graduate placements, 97 are enterprise roles, 7 are plant, quality or manufacturing roles and 42 are out of scope. Four postings show no openings field, none of them in the R&D theme class. These are requisition postings listed on one day, not hires, headcount or a trend.

The T1 postings describe AI and computational roles in discovery: - in gRED's AI for Drug Discovery group (Prescient Design): ML infrastructure on AWS [>ROG-R01], model-serving and inference infrastructure for in-silico drug discovery workflows [>ROG-R19], reasoning language models and agents for biomolecular design [>ROG-R02, >R03, >R26], agentic small-molecule design and synthesis planning linked to automated synthesis platforms [>ROG-R09, >R10, >R14], antibody and protein ML including developability [>ROG-R20, >R21, >R24], and generative modelling research [>ROG-R23] - in AI for Biology and Translation: foundation models and agents for target and drug discovery [>ROG-R07, >R12, >R16, >R22], and ML design of nucleic acid sequences in lab-in-the-loop cycles [>ROG-R28] - elsewhere: ML with quantum chemistry for small-molecule design [>ROG-R08], computational chemistry and computer-aided drug design with ML [>ROG-R13, >R27], siRNA design in pRED working with AI experts on AI design solutions [>ROG-R11], and computational neuroscience in the CS CoE using LLMs and AI agents to nominate targets [>ROG-R17]

The T2 postings describe lab systems and scientific data roles: - in the CS CoE Data and Digital Catalyst: a senior lab automation software engineer in Basel or Penzberg, to build orchestration systems, instrument drivers and "a digital representation of the lab", and to integrate instruments with LIMS and ELNs [>ROG-R25]; a software engineer for research pathology data pipelines and integrations between LIMS, ELN, workflow orchestration systems and AWS components [>ROG-R04]; and two principal software engineers, in Basel and Welwyn, for the research data ecosystem, one of them also building molecule design tooling [>ROG-R06, >R29] - a Lab Automation Digital Solution Lead at Genentech in South San Francisco, described as lab solutions delivery lead for RDT, leading product delivery "in the Lab digital asset and AI based automation space" [>ROG-R18]

The two postings tagged both are ML engineers in the CS CoE AI Enablement group in Welwyn. Each builds AI/ML tools for drug discovery scientists and maintains "data systems for collecting, structuring, and storing diverse scientific data" [>ROG-R05, >R15].

A postdoctoral fellowship covers imaging-based compound screens using lab automation [>ROG-R30], and a pRED student internship programmes a robotic pipetting workflow [>ROG-R37]. Enterprise postings include a Domain Architect for Pharma R&D in a federated RDT model, with a dotted line to the Global Head of Digital Technology for Pharma R&D [>ROG-R41], and an orchestration engineer in the Accelerated Compute Engineering team, with "AI Factory" in the title, for workload scheduling on Roche HPC platforms [>ROG-R115].

Posting dates come from a field in the page data that may record a refresh, so they are not reliable.

No LinkedIn company insights were collected for Roche.

Technology mentions in recruitment

Mentions are classified from the exact posting wording. They do not establish installation, an incumbent contract or company-wide use. The table records 69 Roche mentions: 2 explicit use, 13 integration or environment, 8 required experience, 5 preferred experience and 41 alternatives. - Named as existing systems: AWS, where the AI for Drug Discovery DevOps engineer is to "Design, implement, and maintain scalable and reliable ML infrastructure on AWS" [>ROG-R01], and PyTorch, where the foundation models engineer is to "Own the production Python/PyTorch codebases" [>ROG-R26]. - Integration responsibilities: LIMS and ELN, as categories with no product named [>ROG-R04, >R25]; AWS components [>ROG-R04, >R29]; Terraform and Kubernetes as infrastructure-as-code [>ROG-R22]; automated and high-throughput synthesis platforms [>ROG-R10]; GPU-backed services and in-silico drug discovery workflows [>ROG-R19]; cloud-based production environments [>ROG-R14]; laboratory automation [>ROG-R30]; and robotic pipetting [>ROG-R37]. - Required experience: Kubernetes [>ROG-R01, >R04, >R05, >R19], Helm [>ROG-R19] and digital twin simulation [>ROG-R25]. LIMS and ELN appear as required experience only in the enterprise Domain Architect posting [>ROG-R41]. - Preferred experience: AWS [>ROG-R05, >R06, >R19, >R22] and LangGraph [>ROG-R22]. - Lists of examples: PyTorch, JAX or TensorFlow [>ROG-R07, >R16, >R21]; RDKit or OpenEye [>ROG-R09, >R10]; Schrodinger, MOE, OpenEye and Rosetta [>ROG-R13, >R27]; Psi4 [>ROG-R08]; SageMaker [>ROG-R01]; Triton and vLLM [>ROG-R19]; LangChain and LangGraph [>ROG-R04, >R09, >R26]; and commercial liquid-handling or laboratory automation platforms [>ROG-R25].

No category (ELN, LIMS, cloud) is recorded as explicit use for Roche, and no ELN or LIMS product is named in the R&D postings. Mentions were extracted only for requisitions a reviewer tagged to a theme, so most enterprise postings are not covered.

Deployment evidence

• A Genentech story of 9 January 2026 says: "Today, all of our antibody programs and ~90% of eligible small molecule programs integrate AI to streamline discovery" [>ROG-E20]. The text block was edited on 21 May 2026, so the date of this wording is uncertain.
• The March 2026 release says Roche "now operates" the hybrid-cloud AI factory, and that in R&D the NVIDIA BioNeMo platform enhances Lab-in-the-Loop [>ROG-E02].
• The Pharma Day 2026 deck lists two impact milestones in its 2025-2026 column: an antibody designed with AI support entering the development portfolio (first patient in), and a lead-initiation target entering with AI agent support. Neither is dated more precisely [>ROG-E22, >E23].
• The deck describes TargetNexus, an AI agent for target assessment, as "on track to be part of 80% of research portfolio decisions by Q4". This is a stated aim, not a result [>ROG-E25].
• The same deck states that "40% of Q4 '25 through Q2 '26 pipeline decisions had a tracked AI / computational contribution". The measured period crosses the window boundary and no start date is given for this use, so it is recorded as existing capability, not a change. A Reuters report of the investor day led the top-up search to this slide [>ROG-E51].
• Recursion reported delivery of a first Neuromap under its collaboration with Roche and Genentech on 23 October 2024 [>ROG-E11], and of a second, microglia map on 29 October 2025 [>ROG-E12]. On 5 August 2026 it reported that Genentech had advanced the collaboration's first neuroscience target into a joint early discovery programme [>ROG-E13]. These are partner posts; the advance date is not stated.
• An AWS case study describes the gRED Research Agent, built using Anthropic Claude Sonnet 3.5 in Amazon Bedrock Agents, which is "expected to help Genentech automate over 43,000 hours of manual effort" in biomarker validation. The case study shows no publication date, so its first public date is not established [>ROG-E15].
• A Genentech story of 4 December 2024 says the computational sciences team "led the adoption of unified electronic lab notebooks across all of Roche". No date is given for the adoption [>ROG-E33].
• Roche inaugurated a new building for the Institute of Human Biology on 23 March 2026, for up to 250 researchers, as part of a CHF 1.4 billion site investment. The release links the institute to combining human organoid models with AI [>ROG-E32].
• The IR Digitalization Day deck of 25 November 2024 describes gRED's Lab in a Loop strategy to transform drug discovery and development with data and AI [>ROG-E19].
• Apart from the share of programmes in ROG-E20 and the share of pipeline decisions in ROG-E51, no adoption or benefit metric is recorded in the event rows for these items. The 43,000 hours and the 80% figure are expectations, not results [>ROG-E15, >E25].

Context

Roche announced on 29 January 2025 that Wafaa Mamilli would join as Chief Digital Technology Officer, taking over worldwide responsibility for the Roche Group Informatics function from the CFO; the release ties the split of the two roles to the enterprise-wide use of AI [>ROG-E42]. No discovery remit is stated, so this is context (C42-7). Roche announced on 30 June 2025 that the Head of pRED would retire [>ROG-E43], and on 17 February 2026 appointed Mark Dawson as Head of pRED from 1 May 2026 [>ROG-E44]. Neither release states an AI or lab-systems remit.

Company stories name heads of AI and computational units without appointment dates: John Marioni as head of gRED Computational Sciences [>ROG-E09], Rich Bonneau as Vice President of Machine Learning, Drug Discovery [>ROG-E35], Tommaso Biancalani as head of BRAID, the AI division for biological research [>ROG-E36], Anwesha Dey as Executive Director of AI for Oncology and Cancer Biology [>ROG-E37], and Andy Lash as Head of Engineering, Lab Automation [>ROG-E34].

Under rule AR7a, the Caris and A-alpha Bio collaborations listed under data generation on the Pharma Day 2026 deck are data access, not T2 [>ROG-E30, >E31]. Roche is a founding member of the Shared AI License (SAIL) Foundation (Q1 2026 results, 23 April 2026); no discovery scope is stated [>ROG-E41]. The agreement to acquire PathAI (7 May 2026) is AI for digital pathology in diagnostics, so it is context [>ROG-E45]. Two media reports found in the top-up search are also context. Lab Manager reported on 9 March 2026 that Roche has implemented a Tecsys cloud-based inventory platform for lab consumables and reagents across more than 1,000 laboratories, linked to its SAP system; the article does not place it in R&D lab or scientific data systems, and no implementation date is given [>ROG-E49]. Swissinfo (Keystone-SDA) reported on 9 July 2026 that ABB and Roche had announced a partnership to develop robotic solutions for clinical laboratories. This is diagnostics product work, not the company's own R&D labs [>ROG-E50]. A HY 2025 results slide lists "Future capabilities (e.g. lab-in-the-loop, peptides, organoids)" among R&D savings reallocated, with no amount attributable to lab-in-the-loop [>ROG-E24].

Site and manufacturing context: a USD 50 billion US investment programme restates the Massachusetts R&D centre [>ROG-E06]. A Genentech story says new South San Francisco research facilities "are being designed" to integrate AI technologies [>ROG-E46]. The new Holly Springs manufacturing facility includes an AI-powered digital twin, which is manufacturing and outside T2 [>ROG-E47].

We identified no qualifying pause or cancellation in the two themes in the searched sources. The conclusion of the initial Dyno collaboration is recorded as a milestone, because the release states its basis (an option exercise) [>ROG-E38]. A top-up reviewer ran one negative-event web search per theme on 6 October 2026, each covering both years, and found no qualifying item. This is the result of a small number of searches, not evidence that none occurred.

Coverage limits

• The shared web-search budget ran out after 6 searches by one reviewer and 9 by the other. A top-up reviewer then ran 10 further searches on 6 October 2026: one negative-event search per theme; the T1 event-type searches for leadership, consortia, regulatory programmes and open releases; one combined acquisition and venture search covering both themes; two T2 searches for lab automation and lab data agreements; and one university collaboration search. Each search covered both years in one query, with no date filter. The top-up added one existing-capability row and two context rows, and no change count moved [>ROG-E51, >E49, >E50].
• Complete in both years: the Roche and Genentech press release archives, the annual reports, the venture arm pages, the capacity, partnership and publication searches, the negative-event searches for both themes, the T1 searches for leadership, acquisitions, venture investments, consortia, regulatory programmes and open releases, and the T2 searches for acquisitions and venture investments. Partial in both years: results presentations and transcripts, partner newswires, consortium announcements and academic collaborations, which were only partly searched, and the T2 searches for leadership, consortia, regulatory programmes and open releases, which were not run.
• Neither reviewer retrieved results-call transcripts or webcast audio. Diagnostics Day decks were not read.
• Business Wire blocked access and GlobeNewswire timed out. One reviewer did not search PR Newswire; the other reached the Earendil release there [>ROG-E40]. The Medra release of 11 December 2025 on Business Wire was not fetched, so Medra's own post was used [>ROG-E16].
• Recursion's investor site, Fierce Biotech and MobiHealthNews blocked access, and Barchart returned an empty page. Recursion news posts and other copies were used instead [>ROG-E11, >E12, >E13]. The IHI site's robots.txt returned 403, so no IHI pages were used. SEC EDGAR returned a request-rate block and was not used. GEN and the NVIDIA blog disallow the reviewer's agent in robots.txt and were not fetched.
• One reviewer notes that items in the 2023-2024 column of the Pharma Day 2026 timeline were probably public earlier; this was not checked, including in the top-up search [>ROG-E10, >E15, >E27, >E29, >E31].
• First-public dates are uncertain for the Annual Report 2024 items, which rest on the results publication date [>ROG-E01, >E14], and for Genentech story blocks edited after publication [>ROG-E08, >E16, >E20]. The AWS case study has no publication date [>ROG-E15].
• A Genentech logo on the Apheris AI Structural Biology network page is not verified, and the row is excluded [>ROG-E48].
• Chugai was not searched (C42-12). Under the opposite reading, with Chugai in scope, these counts would not change, but every Chugai cell would be unsearched.
• Patents are not among the searched signal types. Publication counts are raw Europe PMC query counts and are not events.
• Recruitment: the Genentech careers sitemap lists no job URLs, so one reviewer screened Genentech category pages for data science, research informatics and research roles. Only the 184 requisitions screened by either reviewer are in the table.
• Two independent reviewers searched Roche, and the final rows are their source-verified union, plus three rows from the top-up search [>ROG-E49, >E50, >E51]. Of 48 matched event pairings, 25 were found by both. The union does not show that both did not miss the same events. The second Dyno collaboration and Earendil were found by one reviewer only [>ROG-E39, >E40].

Two questions to investigate (Ortent's interpretation)

1. Lab automation and lab data roles sit in several units. The CS CoE Data and Digital Catalyst is hiring a lab automation software engineer to build orchestration systems and instrument drivers and integrate them with LIMS and ELNs, and engineers for research data pipelines that include LIMS and ELN integrations [>ROG-R25, >R04, >R06, >R29]. A Genentech Lab Automation Digital Solution Lead is to be lab solutions delivery lead for RDT [>ROG-R18]. A Genentech podcast names a Head of Engineering, Lab Automation [>ROG-E34], and Medra's platform integrates with Genentech's laboratory information management system [>ROG-E16]. The Pharma Day 2026 deck describes an Autonomous Lab Accelerator with autonomy levels from fully manual to lab-wide autonomy [>ROG-E17]. What is the remit of each unit for lab automation and lab data systems? Who decides on new lab automation platforms and on integrations with LIMS and ELN, and how are the interfaces between the CS CoE, RDT and the research sites governed?
2. Compute and AI capability for discovery are described in several places. The AI factory release quotes Wafaa Mamilli as Chief Digital and Technology Officer [>ROG-E02], a role that took over group informatics [>ROG-E42]. An Accelerated Compute Engineering posting with "AI Factory" in its title covers workload scheduling on Roche HPC platforms [>ROG-R115], while the AI for Drug Discovery group hires its own engineer for ML infrastructure on AWS [>ROG-R01]. The deck lists an OpenAI pilot partnership and an AWS Research Agent collaboration [>ROG-E28, >E15], and the external AI-discovery deals with Dyno and Earendil quote the Head of Roche Corporate Business Development [>ROG-E39, >E40]. Who holds decision rights over compute and over external AI model or platform agreements for discovery: the research organisations, the CS CoE, RDT or business development? How is discovery use of the AI factory funded and allocated?

Sources for the main claims

• ROG-E39 Dyno Therapeutics, second collaboration with Roche: https://www.dynotx.com/news/dyno-therapeutics-forms-new-strategic-partnership-with-roche-to-advance-aav-gene-therapy-vectors-for-neurological-diseases
• ROG-E05 Roche, launch of the Roche Genentech Innovation Center Boston: https://www.roche.com/media/releases/med-cor-2025-03-07
• ROG-E16 Medra, Series A post (earliest mention): https://www.medra.ai/series-a
• ROG-E16, E08, E20 Genentech story "AI Fuels Genentech's R&D Ecosystem": https://www.gene.com/stories/ai-fuels-genentech-r-and-d-ecosystem
• ROG-E02 Roche, NVIDIA AI factory: https://www.roche.com/media/releases/med-cor-2026-03-16
• ROG-E40 Earendil Labs, collaboration with Genentech (PR Newswire): https://www.prnewswire.com/news-releases/earendil-labs-announces-research-collaboration-with-genentech-to-discover-and-develop-therapeutic-bispecific-antibodies-in-oncology-302888622.html
• ROG-E01, E14 Roche Annual Report 2024: https://assets.roche.com/f/176343/x/09457b2a19/ar24e.pdf
• ROG-E07 Roche, opening of the Boston centre: https://www.roche.com/media/releases/med-cor-2026-09-17b
• ROG-E38 Dyno Therapeutics, option exercise by Roche: https://www.dynotx.com/news/dyno-therapeutics-announces-exercise-of-option-by-roche-for-next-generation-aav-vector-for-neurological-gene-therapy
• ROG-E10, E17, E21 to E23, E25 to E29, E51 Roche Pharma Day 2026: https://assets.roche.com/f/176343/x/51deda2587/20260928_final_pharma-day-2026-publish-1.pdf
• ROG-E49 Lab Manager, Tecsys laboratory inventory platform: https://www.labmanager.com/roche-deploys-cloud-platform-to-automate-laboratory-inventory-across-global-network-35026
• ROG-E50 swissinfo.ch (Keystone-SDA), ABB and Roche laboratory robotics: https://www.swissinfo.ch/eng/swiss-ai/abb-and-roche-join-forces-to-revolutionise-laboratories-with-robotics/91720158
• Other event rows and all requisitions: URLs and archived copies in the Roche tables. No SRC source files are cited for Roche.

Source index

Each ID in brackets links to the online source for that row. Job posting links come from a snapshot of the company job site on 6 October 2026; postings that have closed since are marked and no longer linked. Where a company site blocked access, the link goes to the copy that was read.