How Do I Explain Level 1 vs Level 2 Automation to My CFO?

When discussing AI-driven automation with your CFO, clarity is paramount—especially in customer experience (CX) operations where every dollar spent must translate into measurable ROI. Distinguishing between different levels of automation, particularly Level 1 and Level 2, is critical to align investments with company goals and compliance concerns.

In this post, we’ll break down the distinctions between Level 1 and Level 2 automation, grounding the explanation with industry examples like Zowie’s per-conversation pricing, Intercom’s Fin approach, and Zendesk’s Advanced AI capabilities. We’ll also dive into key concepts such as knowledge vs https://bizzmarkblog.com/what-does-deterministic-execution-mean-in-customer-service-ai/ process automation, how deterministic workflows reduce hallucination risks, and why enterprise observability and audit trails are essential for compliance. Finally, we’ll frame it all within a solid ROI narrative based on cost per ticket metrics.

Understanding the Three Levels of Automation in CX

Automation in customer service typically unfolds across three levels:

Knowledge automation: AI helps agents find information faster or surfaces relevant answers directly to customers.
Process automation: AI automates business processes such as ticket routing, simple issue resolution, or workflow execution.
Orchestration automation: Complex end-to-end workflows that coordinate systems, teams, and AI actions to resolve multi-step issues.

This post focuses on differentiating Level 1 (mostly knowledge automation with some limited process automation) vs Level 2 (richer process automation)—a distinction that’s critical for CFOs who want to understand where AI investments add real value.

Level 1 Automation: AI That Assists - Knowledge Automation

Level 1 automation primarily helps agents or customers find information quickly. It leverages:

Large Language Models (LLMs) for natural language understanding
Retrieval-Augmented Generation (RAG) to pull in relevant knowledge base articles or FAQs

Here, AI acts as an assistant—it suggests answers, auto-suggests responses, or powers smart knowledge search—but the actual resolution still requires human oversight or manual follow-up.

Example: Intercom Fin’s AI Assist

Intercom’s Fin demonstrates Level 1 automation well: it uses AI to streamline agent workflows by suggesting knowledge articles or next best actions, speeding up resolutions. However, it doesn’t fully automate complex ticket handling. This means costs scale largely with agent seats, as AI assists but doesn’t replace full human involvement.

Why CFOs appreciate this level: Faster agent handling times improve cost per ticket, but gains are incremental and dependent on labor costs.

Level 2 Automation: AI That Resolves - Process Automation

Level 2 moves beyond assistance to actual automated resolution of certain issue types. This level includes:

Automated workflows that perform deterministic actions (e.g., password resets, order status updates)
Instant conversation handling without agent intervention
Integrated system updates following business rules

Deterministic workflow execution is key here—by strictly defining logical, rule-based actions, Level 2 automation minimizes AI hallucination risks, a concern that often unsettles risk-averse CFOs.

Example: Zowie’s Per-Conversation Pricing Model

Zowie employs Level 2 automation with a pricing model based on conversations rather than seats. AI bots handle entire customer interactions autonomously when possible, freeing up agents for complex cases. This model aligns costs directly with automation success and scalability rather than ada chatbot vs forethought headcount, offering clearer ROI visibility.

Why CFOs love this level: Clear cost-per-conversation metrics make it easier to predict margin improvements and reduce labor spend.

Zendesk’s Advanced AI: Bridging Levels with Orchestration Automation

Zendesk’s AI offerings include advanced orchestration capabilities—moving towards Level 3 automation by coordinating multi-step workflows involving AI and humans. However, due to regulatory needs, many enterprises limit full automation, reinforcing the importance of audit trails and compliance.

Why Audit Trails and Enterprise Observability Matter

Especially in regulated industries, it’s not enough for AI to just work; everything must be traceable. Audit logs that track AI decisions, retrain cycles, and workflow histories are vital for compliance and troubleshooting.

Enterprise observability: Real-time monitoring with dashboards showing AI performance metrics and exceptions.
Audit trails: Immutable logs showing who or what performed which action, critical for internal and external audits.

Always ask vendors to show these logs **before** demos. It’s a personal quirk born from too many “automation that broke at 2am” stories where insufficient traceability delayed incident response.

Crafting the ROI Narrative: Cost-Per-Ticket Magic

The CFO’s favourite metric? Cost per ticket. Every automation initiative should map directly to improvement here:

Automation Level Typical Cost Impact Cost Metric Key Risk Level 1 (Knowledge) Moderate reduction via faster agent handling Agent time per ticket (seconds saved) Overreliance on assistants, incomplete resolution Level 2 (Process) Significant reduction via automation of common issues Cost per conversation (Zowie model) Automation errors, hallucinations Level 3 (Orchestration) Highest potential but complex to implement End-to-end case resolution cost Workflow failures, compliance gaps

Your ROI story should clearly articulate how automation reduces labor costs, dependency on seats, and error rates by automating predictable interactions and freeing skilled agents for high-value services.

Common Vendor Pitfalls and How to Avoid Them

Beware of FAQ bots masquerading as AI agents: These are often Level 0 or partial Level 1 tools that don’t truly automate processes.
Don’t accept hand-wavy “autonomous” claims without audit logs and demos: Insist on detailed traceability, especially for regulated contexts.
Pricing should scale with automation, not seats: Vendors like Zowie demonstrate that per-conversation pricing aligns incentives better and drives ROI.

Final Thoughts: Speak CFO, Not Just Tech

Explaining Level 1 vs Level 2 automation to your CFO boils down to clearly framing:

What kind of problems AI is solving (assist vs resolve)
How automation maps to tangible cost savings, especially in cost per ticket
Risk controls like deterministic workflows and auditability that protect your business
Pricing approaches that reward efficiency over headcount

By using real-world examples from pioneers like Zowie, Intercom (Fin), and Zendesk (Advanced AI), you provide credibility and concrete comparisons. Combining this with a focus on knowledge vs process automation and enterprise compliance requirements ensures your CFO hears a compelling, risk-aware ROI story that supports smarter AI investments.