Talk to Leadership

Why · The Research

When AI eats SaaS, who owns the frontline?

The research that validates the Frontline OS thesis isn’t RedeApp’s. It’s the convergence of three independent views from venture capital, the major consultancies, and the hyperscalers themselves — on the SaaS-to-AI shift, the consultancy supercycle, and the structural deskless gap. We didn’t make the argument first; we built the platform that answers it.

See the vs-incumbents argument Read the thesis →

Satya Nadella’s widely-cited remark that “business applications as we know them will collapse” into the agent layer captured a view that was already forming across venture capital, analyst houses, and the major consultancies. The argument has three threads: (1) the SaaS-to-AI shift is real and asymmetric; (2) consultancies face the largest services opportunity in a generation; (3) the deskless workforce — structurally underserved by traditional SaaS for two decades — is where the asymmetric upside lives. The convergence point is well-supported in the public record. Below, the sources that anchor each thread.

The SaaS-to-AI shift

Thread 1 · The view from venture capital and the hyperscalers.

The 'SaaS-is-being-unbundled' argument moved from contrarian to consensus inside 18 months. The hyperscalers’ pivots and the venture publications agree on the shape; they differ only on the timing.

  • Satya Nadella · Microsoft

    “Business applications as we know them will collapse” into the agent layer. Microsoft’s Copilot-first repositioning is the institutional admission. The seat-based UI-driven SaaS model is no longer the unit of value.

  • a16z · AI agents are eating software

    Sarah Wang and others have written extensively on the natural endpoint of LLM-native applications: the disappearance of dashboards in favor of conversational and autonomous interfaces. The point-tool UI layer is compressing fast.

  • Foundation Capital · Service-as-Software

    Joanne Chen and Jaya Gupta frame a multi-trillion-dollar TAM expansion as software starts billing for outcomes rather than seats. The 'pay for a license' model is structurally challenged by the 'pay for the work' model.

  • The incumbent pivots

    Salesforce’s Agentforce. Microsoft Copilot. ServiceNow Now Assist. Workday’s agent system of record. Four different framings of the same admission: the seat-based UI-driven SaaS layer no longer captures most of the value being created.

The consultancy supercycle

Thread 2 · The view from the major consultancies.

The largest services opportunity in a generation isn’t a narrative claim; it’s showing up in P&Ls. Six major firms have publicly disclosed the shift in client demand from buy-and-configure-SaaS to design-and-operate AI-augmented processes.

  • Accenture

    Multiple billions in cumulative GenAI bookings since FY24 with double-digit sequential growth in AI-related services revenue. The cleanest financial signal that the consultancy supercycle is structural, not narrative.

  • Capgemini Research Institute

    ‘Harnessing the value of generative AI’ and ‘Turbocharging software with Gen AI’ document a sharp shift in client demand. Resonance and AI-first delivery messaging operationalize the thesis.

  • Deloitte & BCG

    Deloitte’s annual Tech Trends and Engineering AI practice. BCG’s Build for the Future index. Both show client-side reorientation toward AI-augmented operating models, not toward more SaaS configuration.

  • McKinsey QuantumBlack & IBM Consulting

    McKinsey QuantumBlack’s expansion and pricing power in agentic engagements. IBM Consulting’s positioning around watsonx and Consulting Advantage. TCS AI-WisdomNext, Infosys Topaz — same playbook, different firms.

The deskless gap

Thread 3 · Structurally underserved for two decades.

The 'deskless gap' isn’t new. What’s new is that AI makes closing the gap economically tractable for the first time. The 2.7 billion deskless workers globally remain the largest underserved labor market in the world.

  • Emergence Capital

    Coined ‘Deskless Workforce’ as a category and quantified the imbalance: roughly 80% of the global workforce is deskless, yet enterprise SaaS investment has historically targeted the 20% at a desk.

  • Gallup & Microsoft Work Trend Index

    Frontline workers consistently report worse access to information, training, communication, and tooling than their desk-bound counterparts. Pew Research reports 96% smartphone ownership among frontline-age cohorts, 92% among Gen X, with 20% smartphone-dependent for internet access.

  • McKinsey & Gartner

    McKinsey’s frontline operations work and Gartner’s research on ‘deskless digital workplace’ both flag the same structural under-investment. The pattern is documented; the platforms to address it have lagged the documentation.

  • IBM Cost of a Data Breach

    Average cost of a data breach in 2024: $4.88M per incident. Frontline organizations running shadow IT (WhatsApp, GroupMe, personal devices for work) carry an outsized share of that exposure — one of the structural arguments for governance-grade frontline platforms.

Input · Why the data model is the moat

Identity that holds when everything else moves.

In knowledge work, the AI transition is largely a reshuffle: existing desk workers gain agents that ride on top of mature SaaS estates (Microsoft 365, Salesforce, Workday, ServiceNow). The data, identity, and integration layers already exist; they get re-fronted by AI.

In frontline work — healthcare floors, manufacturing plants, logistics fleets, hotels and restaurants, retail stores, senior living, construction sites, energy fields, food production — those layers do not exist coherently. Communication runs on a fragmented mix of consumer messaging, paper, tribal knowledge, and a long tail of point apps frontline workers rarely log into. There is no system of record for what happened on the floor. There is no canonical identity for an employee who has a badge number, a clock-in number, and a regional HRIS ID that don’t reconcile.

The result: applying AI to frontline operations without first building the underlying operating system produces the failure mode that has plagued frontline digital transformation for two decades — a polished demo that cannot scale because the foundation is missing.

This is the fact pattern that determines how the next cycle plays out. Whoever ends up holding the layers that appreciate — systems of record, identity fabrics, integration planes, distribution surfaces — owns the deskless AI era.

Input · Why the data model is the moat

Identity that holds when everything else moves.

The shorthand ‘AI is killing SaaS’ obscures more than it reveals. A more precise read of the public record is that AI compresses the value of three SaaS layers and increases the value of two.

LayerDirectionWhy
Point-tool UIsCompressingForms, dashboards, ticket UIs absorbed into conversational interfaces. Seat licenses lose justification.
Workflow logic / business rulesCompressingLLMs encode and execute rule-based workflows once context is structured. Vendors monetizing only workflow engines exposed.
Vertical content / templatesCompressingGeneric vertical templates (HR letters, safety scripts, training outlines) generated, not purchased.
Systems of record + proprietary behavioral dataAppreciatingAgents are only as useful as the data they ground in. Whoever owns the operational data owns the agent’s competence.
Identity, access & integration fabricAppreciatingAgents acting on behalf of users need a permission model the enterprise trusts plus an integration plane (API + MCP) into the systems they touch.

This is the most important framing in the briefing. The question is not ‘how much SaaS gets killed.’ The question is who ends up holding the layers that appreciate.

Sources: public statements and publications from Microsoft, Salesforce, Workday, Accenture, Capgemini Research Institute, Deloitte, BCG, McKinsey, IBM, a16z, Bessemer Venture Partners, Foundation Capital, Emergence Capital, Gartner, Forrester, Gallup, Microsoft Work Trend Index, IBM Cost of a Data Breach Report, and Pew Research smartphone-adoption data. Full citations available on request. The RedeApp Analyst Briefing v1 (May 2026) consolidates these into one consistent narrative arc; available under NDA via the leadership form.

Next step

Solve digital abandonment. Deploy the Frontline OS.

The research above explains why the category is shifting. The vs-incumbents page explains why HCM extensions, communications platforms, and EXP wrappers can’t structurally pivot into the gap — even when they announce 2026 frontline agents.

Talk to leadership