Sign up to save your podcasts
Or
Share 286: Measure What Matters When Humans And AI Share The Queue; with Craig Stoss
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
286: Measure What Matters When Humans And AI Share The Queue; with Craig Stoss
Send us a text
AI has crossed a threshold in customer support: it no longer just routes and records—it actually does the work. We sat down with Craig Stoss, solutions lead at Codif, to unpack what changes when agents and co-pilots resolve real cases, trigger refunds, detect spam, and personalize replies across channels. The core shift is mindset: once software handles tasks, you need to manage it like a teammate—measure output, audit quality, plan capacity, and set honest expectations about what AI should own versus when a human steps in.
We map the strengths and limits on both sides. AI excels at repeatable workflows, multilingual responses, classification, and fast data gathering across commerce and CRM tools. Humans shine in ambiguity, ethical judgment, and emotional connection. That split demands better metrics: evaluate AI on loop frequency, hallucination rate, accuracy, and handoff health; evaluate humans on sentiment impact, recovery, and resolution quality. We also tackle the messy middle: what counts as “containment,” how AI prework changes case counts and AHT, and why your SLAs must adapt as simple issues disappear and human queues get harder.
Transparency matters for trust and compliance. Customers behave differently when they know they’re speaking with a bot, so make disclosure sensible and escalation simple. Inside your org, put AI into workforce management like a 24/7 agent with forecasted volume, coverage targets, and per-workflow goals—think 90 percent for FAQs, 80 percent for refund screening, and a lower bar for complex warranties. Budgeting evolves too: tokens and compute join salaries, compounding micro-saves into real capacity. The upside is big: more interesting human roles, fewer transfers, and a path to merge support and success so one empowered person, augmented by AI, owns the outcome end to end.
If you care about scaling support without losing humanity—smarter metrics, cleaner handoffs, and a realistic plan for AI and people to thrive together—this conversation is your playbook. Subscribe, share with a support leader who needs it, and leave a review with the one metric you’d change first.
View all episodes
By Charlotte Ward
5
22 ratings
Send us a text
AI has crossed a threshold in customer support: it no longer just routes and records—it actually does the work. We sat down with Craig Stoss, solutions lead at Codif, to unpack what changes when agents and co-pilots resolve real cases, trigger refunds, detect spam, and personalize replies across channels. The core shift is mindset: once software handles tasks, you need to manage it like a teammate—measure output, audit quality, plan capacity, and set honest expectations about what AI should own versus when a human steps in.
We map the strengths and limits on both sides. AI excels at repeatable workflows, multilingual responses, classification, and fast data gathering across commerce and CRM tools. Humans shine in ambiguity, ethical judgment, and emotional connection. That split demands better metrics: evaluate AI on loop frequency, hallucination rate, accuracy, and handoff health; evaluate humans on sentiment impact, recovery, and resolution quality. We also tackle the messy middle: what counts as “containment,” how AI prework changes case counts and AHT, and why your SLAs must adapt as simple issues disappear and human queues get harder.
Transparency matters for trust and compliance. Customers behave differently when they know they’re speaking with a bot, so make disclosure sensible and escalation simple. Inside your org, put AI into workforce management like a 24/7 agent with forecasted volume, coverage targets, and per-workflow goals—think 90 percent for FAQs, 80 percent for refund screening, and a lower bar for complex warranties. Budgeting evolves too: tokens and compute join salaries, compounding micro-saves into real capacity. The upside is big: more interesting human roles, fewer transfers, and a path to merge support and success so one empowered person, augmented by AI, owns the outcome end to end.
If you care about scaling support without losing humanity—smarter metrics, cleaner handoffs, and a realistic plan for AI and people to thrive together—this conversation is your playbook. Subscribe, share with a support leader who needs it, and leave a review with the one metric you’d change first.