AI Agents
Featured — Visa and Inflow bring payments to AI agents
Why it matters: Agentic AI can now spend on your behalf with enterprise guardrails. That unlocks end-to-end automation (procurement, cloud credits, data subscriptions) without a human in the loop for every transaction — with budgets, policies and audit trails in place.
- Inflow is launching agent-native commerce infrastructure on top of Visa Intelligent Commerce to let AI agents request, approve and settle payments under policy controls.
- Visa execs call AI agents a “new buyer category”; 71% of businesses say they’re willing to optimize products for agents, per Visa’s research.
- Initial focus is B2B/infra spend (compute, inference, data, storage) vs. consumer shopping; granular spend limits and whitelisted vendors reduce risk.
- Practical take: If you’re piloting AI employees, start by mapping what they need to buy (APIs, credits) and set transaction caps, vendor allowlists, and approval thresholds.
Sources: Forbes • Inflow announcement
|
Microsoft discloses RCE flaws in agent frameworks (Semantic Kernel)
Two CVEs (CVE-2026-26030, CVE-2026-25592) show how prompt injection can pivot into code execution and file writes via default plugins.
Action for builders: upgrade Semantic Kernel to 1.39.4+, audit tool/plug-in allowlists, and log every tool call with inputs/outputs.
Microsoft Security Blog
|
Teradata launches governed platform for enterprise AI agents
The new Autonomous Knowledge Platform routes every agent tool call through a managed control plane (auth, schema validation, RBAC/ABAC, audit).
So what: A path to move agents from pilots to production without bolting on bespoke guardrails for each workflow.
InfoWorld
|
|
AI News
OpenAI rolls out three realtime voice models for agents and apps
OpenAI’s GPT‑Realtime‑2 (reasoning + tool calls), GPT‑Realtime‑Translate (70+ input languages → 13 outputs), and GPT‑Realtime‑Whisper (live transcription) target call centers, education, events, and creator tools.
- Practical impact: builds voice agents that can interrupt, maintain longer context, and trigger business tools mid-call.
- Pricing is token- or minute-based; budget for concurrent sessions and guardrails to prevent abuse.
TechCrunch
|
CoreWeave beats revenue, lifts capex floor; backlog hits $99.4B
The neocloud provider reported $2.08B Q1 revenue, raised 2026 capex floor to $31B amid component cost pressure, and expanded deals with Meta ($21B), Jane Street ($6B), and Anthropic.
- So what: AI infra remains supply‑constrained. Expect continued price and lead‑time pressure on GPUs and high‑end memory, impacting AI project TCO.
- SMB tip: lock multi‑cloud options and pre‑reserve inference where possible to avoid spot‑market spikes.
Channel NewsAsia (Reuters)
|
Cloudflare cuts ~20% of staff as it retools for “agentic AI‑first” work
Cloudflare said internal AI tool usage is up 6x in three months and that the company is rebuilding teams and workflows around AI agents; it expects $140M–$150M in restructuring charges for 2026.
- Takeaway for leaders: Don’t just “add a copilot.” Redesign roles, approvals, KPIs, and risk controls for AI‑centric workflows — or you won’t see productivity gains.
LiveMint
|
|
🚀 Want AI working for YOUR business? Most companies are experimenting with AI chatbots. We deploy AI workforces — AI Employees that follow up on leads, resolve support tickets, publish content, chase invoices, and screen 200 job applicants overnight so your hiring manager starts Monday with the top 10. Each role has a cost profile and human oversight, managed through one platform. This newsletter? Written by an AI Employee, approved by a human — so our team stays focused on what only humans can do. AIToken Labs helps businesses design their AI Workforce Operating Model — starting with the 2-3 roles that deliver ROI in the first 60 days. Book a free 40-minute AI Workforce Blueprint Session. → https://schedule.aitokenlabs.com/blueprint/40min
|
|
|
