Three moves this week point the same direction. The era of one general model for everything is closing. The money and the talent are flowing into vertical AI now. The clearest signal was a biopharma AI platform deal that folded a specialist team into a much larger lab. Google also shipped a new world model. A CHRO survey ranked AI hiring above every other priority. And OpenAI planted its first lab outside the United States. For founders and HR leaders, the message is simple. Generic AI is now table stakes. The real edge sits in the specifics.
Cohere Buys Its Way Into a Biopharma AI Platform
Cohere announced on May 19 that it is acquiring Reliant AI, a research startup with teams in Montreal and Berlin. (Source: Cohere) It will then fold the team into a new product called North for Pharma. In addition, Reliant brings biomedical datasets, a research workbench, and live customers like GSK, Kyowa Kirin, and Medicus Pharma. (Source: BNN Bloomberg)
Why a Biopharma AI Platform Beats a General Model Here
Drug research runs on messy, regulated, technical data. As a result, a general chatbot stumbles on it. So Cohere is buying the data, the customers, and the people at once. Meanwhile, Reliant’s co-founders both join as senior leaders. That is the pattern to watch. In vertical AI, the moat is no longer the model. It is the domain data, and the team that knows how to use it.
For founders, this changes the math on building a niche AI tool. Say your product is a thin wrapper over someone else’s model. A foundation-model company can then absorb your whole category in one deal. The defensible plays are proprietary data, regulated workflows, and customer trust. North for Pharma is also agentic. It acts on tasks rather than just answering questions. That is the same shift now reaching AI agents in HR workflows.
What to do: If you sell vertical software, audit what only you have. Then lock down your data rights and your customer relationships. Those are what survive when the model layer turns into a commodity.
Google Pushes Agentic AI With a New World Model
Google opened I/O on May 19 with Omni, a multimodal world model. It generates video grounded in real-world physics. (Source: Tech Startups) Google also showed Gemini 3.5 Flash for faster agentic coding, plus a new agent called Gemini Spark. Spark is built to complete tasks across Search, Android, YouTube, and connected apps.
So what? Spark chases the same agentic territory as the biopharma AI platform up top. The difference is focus. For HR teams, an assistant that acts across your apps could one day handle scheduling, onboarding nudges, and document chases on its own. The question for your vendors is simple. Which of our tools expose the APIs an agent would actually need to do that work?
CHROs Now Rank AI Hiring Above Everything Else
Checkr’s 2026 CHRO Insights Report found that 37% of CHROs name AI-driven hiring as their top competitive advantage this year. (Source: Checkr) For instance, that sits ahead of employee experience at 28% and HR-tech modernization at 17%. Still, it is a real reordering of priorities at the top.
Here is the catch. Ranking AI hiring first is easy. Running it well is hard. Biased screening logic and undocumented models create legal exposure. So if you are scaling headcount, treat AI hiring as a process you can audit, not a black box. Our breakdown of how AI is reshaping recruitment covers where the gains and the traps sit.
OpenAI Plants Its First Lab Outside the US
OpenAI signed an MoU with Singapore’s digital development ministry on May 20. It is committing more than S$300 million, about US$234 million, to build its first applied AI lab outside the United States. (Source: CNBC) In particular, the team is set to grow past 200 roles, focused on public services, finance, and healthcare.
For founders hiring across Asia, this matters. A frontier lab setting up locally pulls senior AI talent into the region. It also raises the going rate for that talent. So if you are building a team in APAC, expect more competition and a stronger local pool. Companies already hiring AI engineers across the region will feel both the pull and the lift.
Quick Hits
- The European Commission published draft guidelines for classifying high-risk AI systems on May 19. The consultation runs until June 23, ahead of the AI Act becoming fully applicable on August 2, 2026. (Source: European Commission)
- Andrej Karpathy, an OpenAI founding member and former Tesla AI lead, joined Anthropic’s pre-training team on May 19. It is a notable win in the lab talent war. (Source: TechCrunch)
- A new arXiv benchmark, GroupMemBench, tests how well AI agents recall information across multi-party group chats. Most memory tests miss that gap. (Source: arXiv)
If this week has you rethinking how vertical your own stack should be, start with the foundation. Asanify’s global HRMS keeps payroll, compliance, and people data in one place. That is the clean, owned data that makes any AI layer you add later actually useful.
Biopharma AI Platform: Quick FAQ
What is a biopharma AI platform? It is an AI system built for drug research and healthcare teams, not a general-purpose chatbot. These platforms train on biomedical data and tune for tasks like literature review and regulatory analysis. As a result, they handle the field’s specialized language and rules.
Why are large AI companies acquiring vertical specialists? Because the foundation model itself is becoming a commodity. The lasting advantage now comes from proprietary data, regulated workflows, and existing customers. Buying a specialist team brings all three at once, and faster than building them.
What does agentic AI mean for HR teams in 2026? Agentic AI takes action rather than just answering questions. So it can complete multi-step tasks like onboarding steps or scheduling on its own. For HR teams, that means picking tools with open APIs and clear audit trails. An agent is only as safe as the oversight around it.
Not to be considered as tax, legal, financial or HR advice. Regulations change over time so please consult a lawyer, accountant or Labour Law expert for specific guidance.
