Kore.ai launches "Artemis" — an AI‑native platform for multiagent enterprise systems

What changed: Kore.ai debuted the Kore.ai Agent Platform, Artemis edition, a production-focused multiagent platform that includes a compiled Agent Blueprint Language (ABL), an AI architect called Arch to generate and refine agent blueprints, and a dual‑brain runtime designed to keep deterministic controls separate from model behavior.

Why it matters: Enterprises that need repeatable, auditable agent deployments can use Artemis to move pilot work into production faster because it standardizes agent definitions, enforces governance before deployment, and integrates with Microsoft Azure and Microsoft Agent 365 for enterprise identity and telemetry.

Try/watch: If you’re evaluating multiagent systems, request a short POC that exercises ABL validation and the platform’s audit/export of agent actions—those traces are the parts you’ll need to show security and compliance teams.

Veeam announces DataAI Command Platform to govern data access for autonomous agents

What changed: Veeam unveiled the DataAI Command Platform (announced at VeeamON), a unified data+AI trust layer that maps data, identities, and agent access across live and backup stores (a “DataAI Command Graph”) and bundles security, governance, compliance, privacy, and targeted recovery features aimed at agentic workloads.

Why it matters: For operators worried about agents reaching sensitive systems at machine speed, the platform’s focus on enforcing policies at the data source and correlating agent actions with backup state promises faster incident response and more precise recovery—practical benefits for regulated industries and large distributed estates.

Try/watch: Prioritize a discovery pilot that measures how many high‑risk data objects the graph can identify and whether the platform can enforce blocking or redaction at the source before agents have broad access.

Salesforce surfaces Agentforce Coworker inside search bars for in‑app agent actions

What changed: Salesforce’s Agentforce ecosystem got a new surface called Agentforce Coworker — a beta feature that embeds an AI teammate into searchable interfaces so agents can retrieve CRM context and take actions (for Agentforce customers), and Salesforce updated Agentforce admin and certification materials to align with Spring ’26 changes on May 22, 2026.

Why it matters: Buyers and admins can treat Agentforce Coworker as a low‑friction entry point: it reduces the need to switch tools by letting agents fetch and act on record context from the same search box users already use, but it also raises governance questions because agents will be operating inside transactional systems.

Try/watch: Before wide rollout, test the feature in a sandbox and validate action approval flows, role‑based constraints, and the audit trail for agent‑initiated CRM changes.

DeepMind’s Co‑Scientist and peer systems push agentic workflows into scientific research (coverage roundup)

What changed: Coverage on May 22 highlighted DeepMind’s Co‑Scientist work and similar multiagent research systems that combine iterative hypothesis generation, debate, and experiment planning to accelerate research workflows; press/briefings noted early biomedical use cases such as candidate identification and multi‑omic analyses.

Why it matters: Founders and R&D leads should see these systems not as turnkey lab replacements but as workflow accelerants—Co‑Scientist–style agents can compress early discovery cycles, but they require careful validation, data provenance controls, and human sign‑off in regulated science workflows.

Try/watch: If you work in applied research or life sciences, run a scoped pilot focused on hypothesis generation with strict provenance and human review gates; monitor reproducibility and regulatory traceability as your success criteria.