Merck Animal Health's selection of Salesforce Agentforce Life Sciences as its "OneCRM," announced 6 May 2026, is the most serious enterprise commitment to agentic AI the protein sector has yet seen.
The disclosed scope spans veterinarians, producers and channel partners across all geographies, with explicit references to production monitoring devices, automatic feeders, prescription history and subscription management. MAH's VP of Animal Health IT, Juanjo Francesch, framed the deployment as consolidating biopharma, monitoring and identification systems onto a single platform.
What the press release does not say matters more.
It does not name a go-live date, a market sequence, integration partners, the legacy systems being retired, or the specific action classes agents will be allowed to execute. The interesting question for institutional investors is not whether MAH bought the right platform. It is what livestock customer engagement actually looks like five years out.
Start with the Floor
The constraints come first because they shape everything else.
Federal VCPR rules in the US, FDA Guidance on medically important antimicrobials, and the Veterinary Feed Directive make one thing non-negotiable: prescription decisions stay with a licensed veterinarian holding a current relationship with the herd or flock. Regulation (EU) 2019/6 goes further, banning prophylactic group antimicrobial use and reserving certain molecules for human medicine. The UK VMD, Brazilian MAPA and Australian APVMA mirror much of this architecture.
No agentic workflow touching refills, dose changes, off-label use or food-animal stewardship will close the loop autonomously. Any deployment claiming otherwise is either lying or breaking the law.
That regulatory floor is the friend of good veterinarians, not the enemy of useful AI.
The 2031 Operating Picture
Five years out, livestock engagement for a sophisticated US protein producer running cattle, swine or poultry takes a recognizable shape:
Unified account graph. Herd ID data, SenseHub monitoring streams, vaccine and parasiticide history, ordering cadence, and technical services rep notes sit in one substrate. Today that data lives in five systems on a good day and twelve on a bad one.
Disease-pressure outreach. When BRD pressure spikes in a feedlot region, or HPAI surveillance flags a layer cluster, agents push MLR-cleared regional alerts to the producers and veterinarians most exposed, paired with a draft technical visit and a product availability check.
Inventory-aware program planning. Integrators planning multi-cycle vaccine programs get availability projections across the program horizon, with allocation risk surfaced before the cycle starts. Stockouts that today force last-minute substitutions become exception cases.
Refill packet preparation. Agents assemble jurisdiction-aware refill packets with withdrawal-period checks, VFD logic where relevant, and antimicrobial stewardship logging, then route to the veterinarian for authorization. The vet still decides. The vet decides faster, with cleaner context.
Closed-loop technical services. A SenseHub anomaly in a high-value dairy customer's herd triggers a contact action, an inventory check on the likely indicated products, and a calendar prompt for the regional technical services veterinarian, all before anyone files a ticket.
Leverage, not Displacement
The best production animal veterinarians already operate at the limit of their administrative capacity.
They run between operations, draft protocols, field refill calls, review monitoring dashboards, and try to think strategically about herd health. Every minute spent assembling context is a minute not spent exercising clinical judgment.
Agentic systems take the context-assembly tax off their plate.
The same logic applies to manufacturer technical services reps, who today spend roughly half their windshield time on activities an agent layer can pre-stage. It applies to producer operations leads juggling vaccine programs across six houses or twelve pens. It applies to integrator procurement functions trying to anticipate allocation risk before a cycle starts.
None of this requires the AI to be smarter than the professionals. It requires it to be faster at retrieval, more consistent at routing, and disciplined enough to hand the decision back at the right moment.
Structural Advantage and its Limits
MAH's competitive position rests on a portfolio that is genuinely difficult to replicate. Biopharma, identification through Allflex, monitoring through SenseHub Dairy, Beef and Feedlot, and digital tools all sit under one corporate roof. Stitched onto a single account graph, those assets generate workflow visibility that Zoetis, Boehringer Ingelheim, Elanco, Ceva and Dechra cannot match without comparable device estates. The Salesforce platform itself is buyable tomorrow by any of them. The data exhaust from two million-plus monitored cows is not.
Three things temper this advantage.
The first is veterinary skepticism of manufacturer-led producer engagement, which has historical roots in earlier loyalty programs that strained clinical gatekeeping. The second is distributor leverage. Covetrus and MWI are consolidating, Patterson Veterinary is now under Patient Square Capital, and all three hold the order flow and operate software ambitions of their own. The third is the practice management software estate, where IDEXX, ezyVet, Provet Cloud and Covetrus AVImark each control APIs that determine how deeply any manufacturer agent can actually integrate.
Five years is enough time for those negotiations to mature. It is also enough time for them to stall.
Bottom Line
The protein sector should expect, by 2031, a livestock customer engagement layer materially less fragmented than today, with manufacturer agents handling intake, retrieval, routing and orchestration across most non-clinical workflows.
The veterinarian's clinical authority will be more efficiently deployed, not displaced. The technical services rep will visit the right operation at the right time. The procurement lead will see allocation risk three weeks earlier.
The differentiating capabilities sit downstream of the platform decision: depth of integration with distributors and practice software, MLR governance that scales without breaking, and discipline about which actions agents are allowed to execute autonomously. Platform parity will arrive within twenty-four months across the major animal health players. Operating parity will not.