The Governance Gap That Created a Market

Gartner predicts that by 2028, the average global Fortune 500 enterprise will have over 150,000 AI agents in use — up from fewer than 15 in 2025. That is a roughly 10,000x increase in three years, and the governance infrastructure to manage it is nowhere close to keeping up.

Gartner defines the resulting problem as agent sprawl: an uncontrolled accumulation of AI agents built by different teams without centralized governance or consistent oversight.

That 87% gap — the share of large enterprises operating without adequate controls — represents real exposure: security blindspots, compliance risk, redundant costs, and ungoverned systems making decisions on behalf of the business.

It is precisely this gap that xpander.ai is building into.

What xpander Is Actually Selling

xpander today announced a $7.5 million seed round led by Pico Venture Partners, with participation from Emerge Ventures, Samsung Next, and SeedIL. Alongside the funding, the company is making its enterprise AI agent platform generally available — positioning it as a vendor-neutral control plane for building, running, and governing agents across different models, frameworks, and infrastructure environments.

The company was founded by David Twizer (CEO), Ran Sheinberg (CPO), and Moriel Pahima (CTO), all former AWS principal engineers.

At the core of the platform is its Universal Harness — a model-, framework-, and cloud-agnostic runtime that executes AI agents as portable workloads within a customer's own environment while securely rendering agent interfaces on demand. In practice, this means:

  • Agents built with LangChain, Strands, or Agno can run inside xpander without rebuilding
  • Deployment targets include AWS, Google Cloud, Azure, private VPCs, and fully air-gapped on-premises environments
  • Developers get a language-agnostic REST API, a Python SDK, and Model Context Protocol support for MCP clients like Claude Desktop and Cursor

xpander also introduced Omni, its own enterprise AI agent, which achieved a 90.9% score on the GAIA benchmark.

The Lock-In Argument

Twizer's pitch draws directly on his AWS background. His argument is that multi-cloud became a default enterprise strategy not because any single cloud failed, but because organizations recognized the structural risk of deep single-vendor dependency — and that AI is following the same arc.

"The third issue is the most critical part: it's being locked into one vendor. Everything that you do is actually owned by the company that you chose to work with — their tools, their roadmap, their political view of how agents should react to everything that you do." — David Twizer, CEO, xpander.ai

The corollary, though, is worth noting: vendor neutrality does not eliminate lock-in, it relocates it. Enterprises can swap models and frameworks underneath xpander's harness, but xpander's own control plane — handling execution, identity, memory, governance, and auditability — becomes the new dependency. The company's public documentation does not yet detail how portable that operational state is if a customer terminates its enterprise license.

A Market That Has Already Moved

xpander is entering a competitive field where the battle has shifted well above the model layer.

  • LangChain's LangSmith Deployment already offers infrastructure for deploying and governing production agents, with full in-Kubernetes deployments for data-residency-sensitive environments and hybrid options.
  • CrewAI enterprise includes SSO, role-based access controls, workload identity, and policies, deployable in CrewAI's cloud, customer VPCs, or customer-owned infrastructure.
  • Temporal approaches the problem from durable execution — crash recovery, retries, human approvals, and persistent state across failures — rather than end-to-end agent management.
  • OpenAI's Frontier and Google's Gemini Enterprise Agent Platform are expanding from the model layer upward, adding managed runtimes, agent identity, centralized governance, and policy enforcement at scale.

In other words, the claim of running agents outside a single provider's cloud is no longer differentiating on its own. Several established players already offer that flexibility.

xpander's bigger bet is that enterprises will want a framework-independent control plane sitting above an increasingly heterogeneous collection of models, agent frameworks, and employee-facing AI interfaces — and that no single framework vendor is well-positioned to credibly serve that role. The platform offers features including visual agent development, multi-step workflow orchestration, tool and API integration, user and memory management, and real-time adaptation.

What This Means for Enterprise Teams

What enterprises actually need is an internal development platform for agents that handles the full lifecycle from build through production operations — because the gap between "we built an agent" and "we operate agents at scale" is a platform engineering problem, and treating it as anything less means every team reinvents deployment, governance, and observability from scratch.

For startup founders and technical leaders making platform decisions now, xpander's launch underscores a principle that is becoming industry-wide consensus: with 81% of enterprises planning to tackle more complex agent use cases this year, the harder question is operational — and the common trap is a team getting a prototype working in two weeks, then spending six months assembling deployment pipelines, monitoring, governance, and rollback infrastructure, an "assembly tax" that kills many pilots before they deliver value.

xpander's customer base already includes global enterprises across retail, manufacturing, financial services, technology, and government. The $7.5M seed will be used to accelerate market penetration — though for a company competing against well-capitalized framework vendors and hyperscalers expanding upward, capital efficiency and tight enterprise deployment cycles will matter as much as the architecture itself.