On March 23, monday.com launched Agentalent.ai — a platform where businesses post job roles, review qualified candidates, and hire. The candidates aren't human. They're AI agents. The platform, built in collaboration with AWS and Anthropic, treats agent deployment like recruitment: authentication checks, qualification screening, even a two-week regret period with full refunds if the agent doesn't perform.
This isn't a metaphor anymore. The language of work — hiring, onboarding, performance review — is being applied to software. And it signals something bigger than a product launch: the emergence of a genuine labour market for AI.
The marketplace model goes mainstream
monday.com isn't alone. Hexaware Technologies launched Agentverse on March 17 with over 600 ready-to-deploy AI agents spanning customer service, financial reconciliation, regulatory compliance, demand forecasting, and IT operations. Moltify.ai has entered beta with a pay-per-task model — no subscriptions, just hire an agent for a specific job.
The numbers behind this shift are striking. Gartner projects that 40% of enterprise applications will feature task-specific AI agents by the end of 2026, up from less than 5% in 2025 — an eightfold increase in a single year. Global AI spending has reached $2.52 trillion, a 44% year-over-year jump, with agentic capabilities driving a significant share of the software and services segments.
"Every company will soon have a blended workforce of humans and AI agents," said Roy Mann, co-CEO of monday.com. "Agentalent.ai helps companies define roles, evaluate capability, and onboard AI agents alongside human teams using processes they already understand."
The gap between adoption and scale
Here's what makes this interesting rather than just another AI hype cycle: the data shows a massive gap between experimentation and production. McKinsey found that while 79% of organisations are using AI agents to some degree, only 10% of enterprise functions actually use them at scale. Twenty-three percent of organisations are actively scaling an agentic system in at least one function. The other 68% report that half or fewer of their employees interact with agents in daily work.
That gap is exactly what marketplaces aim to close. The traditional approach — build your own agent from scratch, fine-tune models, wire up integrations — is expensive and slow. A marketplace says: skip the build, pick an agent that's already qualified for the job, plug it in.
Hexaware claims organisations deploying Agentverse can target 40-60% productivity gains in knowledge workflows, 60-80% faster response times across digital channels, and 20-50% cost reductions through automation. Those are aspirational figures from a vendor, worth noting. But even at a fraction of those numbers, the economics are compelling for a 20-person operation paying $80,000-$120,000 per role.
The risks nobody's managing yet
There's a catch, and it's serious. The rush to deploy agents is outpacing the ability to govern them.
Bessemer Venture Partners' research identifies agent security as the defining cybersecurity challenge of 2026. McKinsey's internal AI platform "Lilli" was compromised in under two hours during a red-team exercise — demonstrating how quickly agentic threats outpace human response capabilities. A Dark Reading poll found 48% of cybersecurity professionals now identify agentic AI as the single most dangerous attack vector.
Shadow AI is already a problem. Approximately 22% of employees at monitored companies are running AI agents without IT approval — systems that read email, access CRMs, modify records, and send messages inside enterprise networks without security oversight. Shadow AI breaches cost an average of $4.63 million per incident, $670,000 more than standard breaches.
And marketplace supply chains introduce their own attack surface. The popular open-source agent framework OpenClaw has 512 known vulnerabilities, 135,000 exposed instances, and 341 confirmed malicious skills in its marketplace — with 17% of skills exhibiting suspicious behaviour. Supply chain attacks hide malicious code inside plugins that perform their advertised function while silently exfiltrating data.
What this means for your business
If you're running a 10-50 person company, this shift creates both opportunity and obligation.
The opportunity is real: you can now access capabilities that were previously locked behind six-figure enterprise contracts or months of custom development. Need an agent to handle first-line customer enquiries, reconcile invoices, or draft marketing briefs? Marketplaces will offer that as a service you can trial in days, not months.
The obligation is governance. Before deploying any agent — marketplace or otherwise — you need to answer three questions:
-
What data can this agent access? If an agent touches customer records, financial data, or employee information, you need the same access controls you'd apply to a human contractor. More, actually, because agents don't take lunch breaks and can process data at machine speed.
-
Who's accountable when it goes wrong? monday.com's two-week refund policy is a start, but accountability for a hallucinated customer response or a data breach runs deeper than a subscription refund. Your risk framework needs to account for autonomous decisions.
-
Can you audit what it's doing? If you can't review an agent's actions after the fact — every API call, every record modification, every outbound message — you don't have governance. You have hope.
What to watch
The consolidation of agent marketplaces will accelerate through 2026. Expect Salesforce, ServiceNow, and HubSpot to launch their own — or acquire existing ones. The interoperability standard to watch is Model Context Protocol (MCP), with Gartner noting that 30% of enterprise app vendors are expected to launch MCP servers to support cross-platform agent collaboration.
For Australian businesses specifically, the intersection with the upcoming Privacy Act amendments creates a unique compliance dimension. If an AI agent makes a decision about a person — screening a job application, assessing a loan, triaging a support ticket — that decision likely falls under the new transparency requirements coming in December 2026.
The era of "hire an AI agent" is here. The question isn't whether you'll have AI workers alongside human ones. It's whether you'll deploy them with the same rigour you'd apply to any new hire — or whether you'll let them in through the side door, unmanaged, and hope for the best.
Sources
- monday.com Launches Agentalent.ai — monday.com Investor Relations
- monday.com Welcomes AI Agents to Its Platform — monday.com Investor Relations
- Hexaware Launches Agentverse — PR Newswire
- Gartner: 40% of Enterprise Apps Will Feature AI Agents by End 2026 — Virtual Assistant VA
- 10% of Enterprise Functions Use AI Agents, McKinsey Finds — Yahoo Finance / McKinsey
- AI Agent Adoption 2026: What the Data Shows — Joget / Gartner, IDC
- Securing AI Agents: The Defining Cybersecurity Challenge of 2026 — Bessemer Venture Partners
- A Guide to Agentic AI Risks in 2026 — Security Boulevard
- AI Agent Security in 2026: The Risks Nobody's Managing Yet — Alacritous