Getting an AI agent to 100% pass rate used to take weeks of prompt tuning, test writing, and transcript review. Then your knowledge changed and you started over.
We fixed that. 🧵
Taylor Swift has a secret weapon. She calls it Taylurking. Months of quietly watching. Learning your story. Then showing up and already knowing everything. Everyone wants that moment. Nobody wants to do the work that earns it.
We're excited to welcome BJ's Wholesale Club to the Delight Spark stage on May 7.
They are one of the first retailers to go all-in on a truly personalized AI concierge. At Delight Spark, they'll share what that journey actually looks like.
This is the session retail teams won't want to miss.
Save your seat: https://t.co/WUD6UfeUNg
Voice AI gets talked about in keynotes and pitch decks. But the real test is a 115°F day in Las Vegas when every AC unit in the city decides to quit at once.
That's the reality for Dr. Cool Services. Peak summer means a flood of urgent calls from customers who need help now, not in five minutes, not after a callback. Miss that call and they're already dialing the next company.
Dr. Cool uses Delight ai's voice AI to pick up every call the moment it comes in. During their highest volume days, when a human team alone would be underwater, the AI handles the surge without dropping a single customer.
The best use cases for voice AI aren't hypothetical. They're 115 degrees, your phone is ringing off the hook, and nobody gets sent to voicemail.
https://t.co/20vR3Q7FqB
Companies create departments to fix problems they caused themselves... CX is no exception.
66% of CX leaders think their experience has improved. Only 17% of customers agree.
That's not a gap you close with better survey tools. That's the difference between treating symptoms and fixing the disease.
What if CX — as a department — was always the workaround? → https://t.co/94DsNg0yu9
Imagine we are in 2026 and you're still doing human handoffs 🤯
Introducing: Delight Spark. FINALLY! A user conference built for CX leaders who actually want to stay ahead of AI and create concierge experiences for their customers.
Delight Spark is coming to San Francisco on May 7, and the lineup is stacked:
- CX leaders from some of today's most AI-forward brands (@BJsWholesale , @flynorse )
- Big-time AI disruptors leading the charge (@AnthropicAI )
- Thought leaders like the King of CX himself (@KenHughesIE )
- And the gloriously unhinged https://t.co/NteXFimlNS team
Oh, and we made a music video about it. Watch it. Love it. Then do the responsible thing and grab your ticket. 🎟️
1.3 million jobs a year. A new one every 25 seconds.
hipages is Australia's #1 marketplace connecting homeowners with tradespeople. At that scale, support isn't back office — it's the bloodstream.
Their first instinct? Build it themselves. Custom AI agent, custom workflows, custom handoffs.
They stopped.
"AI is moving too fast to be treated like a traditional build. Trying to develop it ourselves would have slowed us down. Partnering lets us move faster, learn faster, and deliver value much sooner." — Jeremy Burton, CTO
Now tradies get instant answers on the same chat thread they use to talk to homeowners. One platform for the marketplace AND support. No app switching, no repeating yourself.
When something needs a human, the full conversation — every message, every signal — lands in Salesforce automatically. Agents never ask for context because they already have it.
"A single platform built to handle the whole journey will always outperform a stack you have to glue together yourself."
https://t.co/ER33kSESEI
Loyalty is an emotion before it's ever a metric.
@KenHughesIE is bringing 20 years of turning customers into superfans to Delight Spark 2026.
Brand advocates. Tribal fandom. The ultimate competitive advantage in the age of AI.
May 7 · SF · https://t.co/kUwtkghPiH
The difference between launching your AI agent in 6 weeks vs. 9+ months?
It's not the AI.
It's your content.
We asked enterprise CX teams four questions:
→ Is your content structured with clear headings and metadata?
→ Are policies documented consistently across all channels?
→ Do you have a single source of truth — or multiple conflicting versions?
→ How quickly can your content update when products or policies change?
Content readiness is the silent killer of AI deployment timelines. We built https://t.co/J24FBVG2cp to close that gap. But first, the honest audit has to happen.
Ask yourself those four questions 👇 . Watch the on-demand webinar: https://t.co/Y9ZbmXdUfa
CX entered the AI-first era, now your metrics need to.
Deflection rate is the vanity metric of AI support.
The real question isn't "did we avoid escalation?" It's "did the customer's problem actually get solved?"
We break down what to measure instead: https://t.co/UZD2Y3bZ8J
We had early access to @OpenAIDevs's new Realtime model, and it changes what’s possible for voice AI in production.
OpenAI's Realtime model sets a new bar for what voice AI can do. The low latency and transcription accuracy alone are a step change — and what we've seen in early access has been exceptional: handling interruptions, fast speech, and domain-specific terminology accuracy.
For https://t.co/KUHUzopIhf, this is the foundation we've been waiting for to deliver truly production-grade voice AI to the enterprise.
Congratulations to the team at @OpenAIDevs on the launch.
“There is no disruption in healthcare. There’s, at best, incremental transformation.” - @pelutran, CEO at @FerrumHealth
🎧 Listen to the full conversation → https://t.co/W0L98lam4b
95% of AI pilots stall.
Enterprise scale? Not a solo climb.
On the debut of MindMakers, @Get_Writer CSO Kevin Chung maps the ascent:
👨💻 Hands on keyboard → vendor + customer climb in sync
🤝 Hand on shoulder → vendor steadies as customer takes lead
🙌 Hands off → the summit: independence at scale
As Kevin puts it: “We’re part of the customer’s team—and that’s the only way it works.”
🎧 Full episode: https://t.co/8Jehclrmlh
Read the full case study to see how they implemented this in just months after struggling with other solutions for half a year: https://t.co/reRuHGxAJt
Here's how: @Lotte Homeshopping (one of Korea's largest retailers) built an AI agent named Moni using Sendbird's platform + @AnthropicAI's Claude 3.5 Sonnet to revolutionize their quality assurance process:
The results speak for themselves: 30-40% reduction in human agent workload, 24/7 partner support, dramatically faster time-to-market, and significantly improved partner satisfaction.