The News
Zalos raised $3.6 million in seed funding to build “Computer Agents” that automate finance workflows by operating existing enterprise systems the same way humans do.
Analysis
Computer Agents Emerge as the Next Layer of Enterprise Automation
Enterprise AI is entering a new phase where systems move beyond generating insights to executing workflows directly within existing applications. Zalos’ approach of training agents on screen recordings to operate ERPs, spreadsheets, and financial tools reflects the rise of what many are calling “Computer Agents.”
Unlike traditional automation, which depends on APIs and structured integrations, these agents interact with systems through user interfaces, effectively mimicking human behavior. This allows organizations to automate workflows across fragmented systems without requiring deep integration work.
This aligns with a broader shift observed in the application development market. Research from Paul Nashawaty highlights how AI is moving into operational layers of the enterprise, where automation must interact with complex, real-world systems rather than clean, API-driven environments.
For finance teams, this is particularly relevant. Many organizations operate on deeply entrenched ERP systems that are costly and risky to replace, creating a barrier to modernization.
Finance Systems Highlight the Limits of API-First Architectures
Finance operations expose a structural limitation in enterprise software: systems were not designed to interoperate seamlessly. ERPs, CRMs, spreadsheets, and banking systems often lack complete or reliable APIs, forcing teams to manually reconcile data across systems.
Zalos’ positioning (automation that “sits on top” of the existing stack) reflects a growing recognition that replacing core systems is not always practical. Instead, enterprises are exploring layered approaches that augment existing infrastructure.
This represents a shift away from traditional integration strategies. Rather than building new pipelines between systems, Computer Agents treat applications as interfaces to be operated, bypassing integration gaps entirely.
This approach may be particularly appealing in finance, where system replacement carries high operational and career risk. ERP implementations can take over a year and often deliver incremental improvements rather than transformational gains.
Market Challenges and Insights
Finance operations are among the most sensitive areas for enterprise automation. Unlike marketing or customer support, where partial automation can still deliver value, finance workflows require near-perfect accuracy, auditability, and compliance.
This creates a high bar for AI adoption. Systems must:
- Maintain detailed audit trails
- Operate within strict controls and approvals
- Deliver consistent, repeatable outcomes
Zalos’ focus on auditable logs and enterprise security standards reflects these requirements. Every action performed by an agent must be traceable, particularly for processes such as billing, reconciliations, and financial reporting.
Another key challenge is reliability. While general-purpose AI agents have gained attention, domain-specific use cases like finance require specialized models and evaluation frameworks to achieve production-level accuracy.
Our research indicates that enterprises are increasingly prioritizing governance and reliability over raw AI capability as they move from experimentation to operational deployment.
How Developers Are Rethinking Automation in Enterprise Systems
For developers and platform teams, the rise of Computer Agents introduces a new architectural model for automation. Instead of relying exclusively on APIs, automation can now occur at the interface layer, where agents interact with applications as users would.
This has several implications:
- Automation can be deployed faster without waiting for system integrations
- Legacy systems can be extended rather than replaced
- Workflows can span multiple tools without centralized orchestration
However, this model also introduces new challenges. Developers must ensure that agents operate reliably across changing interfaces, handle edge cases, and maintain consistency across workflows.
Additionally, observability becomes critical. Systems must track agent behavior, validate outputs, and provide transparency into automated decisions, particularly in regulated environments like finance.
Looking Ahead
The emergence of Computer Agents signals a broader evolution in enterprise automation. Rather than replacing legacy systems or relying solely on integrations, organizations are exploring ways to overlay intelligence on top of existing infrastructure.
Zalos’ approach highlights how this model can be applied to finance operations, one of the most complex and risk-sensitive areas of the enterprise. If successful, it could offer a path to modernization that avoids the disruption of large-scale system replacements.
For the application development market, this matters because it expands where and how AI can be applied. Automation is no longer limited to systems with clean APIs or modern architectures. Instead, AI can operate directly within the messy, fragmented environments that define much of enterprise software today.
As Computer Agents mature, they may become a foundational layer in enterprise architecture, bridging gaps between systems, accelerating workflows, and redefining how work gets done across the organization.
Deepfake Detection Economics Shift Toward Always-On Voice Security
Logs-First Observability Challenges the Three-Pillar Model
Security Data Overload Threatens Visibility in AI-Driven Enterprises
Lean Infrastructure Powers Massive Scale in AI-Era Web Platforms
AI Discovery Engines Reshape Content Platforms as Real-Time Systems
