Tencent expands WorkBuddy and prepares Hy4 for enterprise agentic work

What changed: Tencent announced new enterprise capabilities for WorkBuddy, its AI office productivity agent, and said its upcoming Hy4 and Hy4 Flash models will add stronger performance on long-horizon productivity tasks, reasoning efficiency, and visual understanding. WorkBuddy can connect with messaging applications and enterprise systems, while Tencent Cloud TokenHub lets companies access multiple models through one API.

Why it matters: Companies evaluating workplace agents can test whether an agent fits into the tools employees already use instead of forcing a new interface. The model choice and single-API approach may also make it easier to balance quality, cost, and regional availability across workloads.

Try/watch: Ask vendors for a live demonstration of a complete multistep workflow—including permissions, failure handling, and audit logs—not just a model benchmark.

Cresta opens self-service tools for building customer-service agents

What changed: Cresta launched end-to-end self-service for developers building customer-facing AI agents. Developers can work through their own code repositories and continuous integration and delivery pipelines, then use Cresta infrastructure for controlled releases, access management, usage visibility, cost tracking, and compliance. The company says some enterprise teams have moved from requirements to production in as little as one week.

Why it matters: This moves agent deployment closer to normal software engineering rather than one-off chatbot configuration. Buyers should compare not only an agent’s answer quality but also how easily their teams can test, roll back, monitor, and control production changes.

Try/watch: Run a small support workflow through the full build-to-production process and measure escalation rate, resolution time, operating cost, and rollback time.

Persona brings supervised agents to identity operations

What changed: Persona introduced Persona AI, a suite comprising Persona Assistant, Persona Agents, and Persona Extend. Its Case Review Agents gather evidence, check it against an organization’s policies, and recommend a decision with visible reasoning; Persona says the products were tested with more than 30 early-access customers and already support hundreds of thousands of identity decisions in production.

Why it matters: Identity work is a practical test of agentic automation because speed must be balanced against fraud, privacy, and regulatory risk. The design keeps consequential decisions under human control while automating evidence collection and routine analysis, offering a useful pattern for other high-stakes workflows.

Try/watch: Define which decisions an agent may recommend, which require human approval, and what evidence must be retained for every case.

SAP adds business context and governance for enterprise agents

What changed: SAP announced new tools across Joule Studio, Joule Work, and SAP Business AI Platform, including no-code and pro-code agent development, access to more than 70 models, multistep workflow planning, and stronger governance. SAP also introduced data-fabric and ontology capabilities intended to give agents trusted information about business entities, relationships, processes, and policies.

Why it matters: Enterprise agents often fail because they lack reliable context—not because the underlying model cannot generate text. Teams using SAP environments can focus less on generic prompting and more on connecting agents to governed business data, defining permitted actions, and testing whether outputs reflect current policies.

Try/watch: Before expanding an agent pilot, document the business concepts and rules it must understand, then test answers against governed source data and deliberately ambiguous cases.