The new self-service: from automation to operating leverage

TL;DR

• Enterprise leaders define self-service in three ways: by outcome, by access, and by handoff.
• The strongest programs honor preference. People choose an agent or a human, and the context travels with them either way.
• Buyers want both paths: 67% of B2B buyers prefer a rep-free buying experience, and 69% prefer to validate AI-generated insights with a sales rep.
• Operating leverage comes from four practices: one shared memory, handoffs that carry context, shared definitions, and outcome-based measures.

DevRev and AWS recently hosted an executive dinner in Palo Alto with The Ortus Club on a single theme: the new self-service. Around the table sat engineering, AI platform, and business systems leaders from across cloud, data, security, payments, software, and consumer technology. The moderated conversation opened with one question: what does self-service actually mean inside your enterprise?

Every leader answered differently, and the differences turned out to be the most useful part of the night.

Three definitions of self-service

Self-service as outcome. Self-service succeeds when the person gets what they came for. The path matters less than the result. An agent that resolves a request end to end counts. So does a human who resolves it in one touch because the context was already there.

Self-service as access. Self-service means anyone in the organization can query the data and act on it without waiting in someone else's queue. A sales leader pulls pipeline coverage directly. A support manager sees account health without filing a request with analytics.

Self-service as handoff. Agents carry the work end to end, and a human steps in when the moment calls for judgment, approval, or empathy.

Each definition is right, and each one describes a different layer of the same system. Outcome is the goal. Access is the foundation. Handoff is the design.

Preference is the operating principle

The table landed on a shared principle: self-service should honor the person's preference. Some people want the agent. Some want a human. Most want both at different moments, and the context should carry across every switch.

Buyer research points the same way. In a Gartner survey published in March 2026, 67% of B2B buyers said they prefer a rep-free buying experience (Gartner). A May 2026 Gartner survey found 69% prefer to validate AI-generated insights with a sales rep (Gartner).

Those numbers describe the same person. They research on their own, and then they want an expert to pressure-test what they learned. Self-service that serves them well makes both paths feel like one conversation.

Where self-service gets hard

The leaders named two challenges that come up before any technology decision.

Expertise lives in silos. Knowledge is spread across teams, systems, and people's heads. Consolidating it, or handing it from one team to another, is slow. Several leaders described platform programs where agreeing on what "customer" or "account" means across departments took longer than connecting the systems.

Independent builders drift apart. When multiple subject matter experts build agents and workflows on their own, the results diverge. One leader described continuous alignment on tools, decisions, and scope as the requirement that keeps a live, customer-facing platform coherent.

What operating leverage looks like

Automation removes a task. Operating leverage changes how much the whole organization can do. Four practices came up again and again.

1. One shared memory of the customer and the employee. Every conversation, with an agent or a person, lands in the same place: questions asked, actions taken, commitments made. Every agent and every human works from the same picture.

2. Handoffs that carry the conversation. When a request moves from agent to human, the human arrives briefed. The customer picks up where they left off and never repeats themselves.

3. Shared definitions before shared agents. Teams agree on what an account, a resolved request, or an at-risk renewal means. Agents built on shared definitions stay aligned as more teams build.

4. Measures based on outcomes. Deflection counts conversations that ended. The stronger measure asks whether the person got what they needed, knew what would happen next, and got it in one pass.

Thank you to the leaders who joined us for an open, generous conversation, and to The Ortus Club and AWS for co-hosting. To see how conversation and account history travel with every handoff, read DevRev's partnership with OpenAI on employee and customer self-service.