eBay VP Dan Leiva Exits AI Support Role, Launches CXAmplify As Company Leans Harder Into Automation
eBay's VP of Customer Service & Marketing Technology, Dan Leiva, has stepped down after nearly nine years to start a private consulting firm, CXAmplify. He says he'll help companies optimize customer experience, implement AI correctly, and drive growth.
Leiva led a 900+ person organization across Product, Design, Engineering, and Tech Ops, running the stack for Customer Service (digital and live), Payment Operations, CRM, and Marketing experiences. His exit comes as eBay pushes further into AI-driven support while many sellers are asking for more human help, not less.
Why this matters for customer support leaders
- AI is moving from pilot projects to core operations. The bar for quality, escalation logic, and measurement just went up.
- Cost-saving pressure is real. If you don't define where AI ends and a human begins, the customer will feel the gap.
- Leadership turnover can break CX continuity. Document your playbooks and guardrails so strategy survives org changes.
eBay's AI push: what happened under Leiva
From 2017 onward, Leiva's team modernized infrastructure and scaled AI across digital support and live operations. In 2023, eBay reportedly spent about $12M overhauling automated support systems to route more users into AI chat, which some sellers say worsened call routing for premium accounts.
After early-2024 layoffs, CEO Jamie Iannone and then CFO Steve Priest told investors they were deploying more AI into global customer experience operations. Their example: AI reads long seller emails and drafts the initial response before an agent steps in.
"And that customer agent sometimes had to read like a 12 paragraph e-mail where somebody explained it, we would hire someone to read that. Now we have AI read that email, write the initial response. So it's fundamentally changing our pace of how we work and the pace of innovation."
Strategy vs. sentiment
A recent Senior Director, CX Engineering job ad signals a "clear directive" to leverage AI for hyper-personalized and proactive interactions, with an emphasis on cost and GMV impact. You can view current postings on the eBay Careers site.
At the same time, many sellers say service has lost the human touch. Some were unexpectedly removed from the Concierge program (even those told they had "lifetime" status) as eBay tightens access to the top 10% by GMV. The message sellers hear: if you're not a top earner, expect lower-tier support or AI self-service.
Concierge then and now
Back in 2017, eBay leaders promised to "fix customer service" and move premium Concierge-style support closer to the default experience. That broader rollout never materialized.
Leadership also drew criticism for large bonuses and severance connected to support performance, including during periods of layoffs and controversy, along with an alleged connection to the 2019 cyberstalking scandal. Fast forward to now: Concierge is contracting while AI expands.
New partnerships, shifting org chart
As one of Leiva's final moves, eBay announced a partnership with Artium.AI and OpenAI to build a next-gen AI customer service platform, plus more AI-focused investments planned for 2026. This follows another senior departure: Chief Privacy Officer and VP AI & Data Responsibility, Anna Zeiter, also left after 11.5 years.
It's unclear whether Leiva's responsibilities will roll up under existing leaders (possibly Dawn Block, now VP Global Customer Experience) or be backfilled. eBay has not responded to requests for comment.
If you run support, here's the practical playbook
- Define the AI boundary: List the top 20 intents you'll automate, the confidence thresholds to proceed, and the exact triggers for instant human handoff.
- Design escalations first: Customers forgive a bot if the escape hatch is fast, accurate, and respectful. Make live assist obvious and immediate for high-risk flows.
- Instrument quality, not just cost: Track first contact resolution, recontact rates, time-to-human, gross seller and buyer satisfaction, plus revenue/save impact by segment.
- Guardrails over guesswork: Use policy-aware prompts, PII controls, and abuse filters. For enforcement or payments, require human-in-the-loop by default.
- Train the humans: AI changes agent work from typing to judgment. Coach on verification, exception handling, and empathy. Update incentive plans accordingly.
- Ship small, learn fast: Start with contained use cases (summarization, intent triage, guided flows). Expand only when quality clears your bar for 4+ weeks.
- Close the loop with customers: Ask how AI interactions feel. Publish what's automated, what's not, and what you improved based on feedback.
What to watch next
- Who inherits AI support ownership and whether KPIs skew toward cost or experience.
- How eBay balances "AI everywhere" with meaningful access to trained agents.
- Whether Concierge criteria stabilize or continue to shift based on GMV concentration.
- Real outcomes from the Artium/OpenAI platform: time-to-resolution, accuracy, and seller retention, not just volume handled.
Bottom line
AI can remove friction. It can also create it. The difference is your boundary lines, your escalation speed, and the courage to prioritize experience when it conflicts with cost.
If you're building your own AI support stack, you might find curated training by role useful: AI for Customer Support and the AI Learning Path for User Support Specialists.
Your membership also unlocks:
