Empathy by algorithm: how humans and AI co-author exceptional support stories
Customer support has always been about more than solving problems. At its best, it’s about recognizing a person’s emotions in moments that feel high-stakes: when a package is missing, an account is locked, or a product isn’t working as expected.
When customers reach out, they’re often not just asking for help. They’re asking to be seen and understood.
At Boldr, we’ve long believed that empathetic, human-first experiences are what transform a service interaction into a relationship. That belief is what makes our new partnership with Forethought so exciting. Together, we’re combining Boldr’s global support expertise with Forethought’s world-class AI agents to help teams scale the kind of care that people remember.
And more importantly, the kind of care that builds loyalty.
When things go wrong, and why that’s not always bad
There’s a common idea in customer experience called the Service Recovery Paradox. It’s the concept that when a customer experiences an issue, but that issue is resolved swiftly and with genuine care, they often become more loyal than if nothing had gone wrong in the first place.
In other words, the way you handle a rough moment can have a bigger impact than a dozen frictionless ones.
Of course, the tricky part is recognizing when a moment is going off track, especially when you’re handling hundreds (or thousands) of interactions a day. That’s where emotion-aware AI can play a powerful role.
Imagine a customer reaches out about a delayed shipment. Their first message is calm, but by the second reply, the tone has shifted. They’re frustrated, and they’ve started using words like “ridiculous” or “unacceptable.” If you’re relying on manual triage, that message may sit in a general queue. But with AI-powered sentiment analysis, that shift is flagged instantly. The conversation gets prioritized, routed to a more experienced agent, and supported with relevant context and response suggestions.
The result isn’t just a resolution. It’s a rescue.
And that rescue moment is what turns frustration into fandom.
What real teams are doing, and why it works
Support organizations across industries are already using this blend of human insight and AI-driven signal detection to create standout experiences, even when something goes wrong.
Grammarly needed something that could handle complexity and speed as their old chatbot struggled with context and follow-ups. Within just a week and a half, Forethought was live and the results were immediate: 87% of email and chat tickets now deflect automatically, CSAT holds steady at 4.2, and when an issue requires a human, the AI hands it off seamlessly with all the context intact.
Fast deployment, quick time to value, and a support experience that actually works for both customers and agents.
ActiveCampaign faced a growing volume of routine questions, overwhelming their team. Their old chatbot offered little relief, relying on rigid scripts that left customers frustrated. By switching to Forethought’s AI, known internally as “Ace,” the company now deflects more than 60% of chat tickets before they ever reach an agent. This resulted in spending less time on repetitive issues, customers got faster resolutions, and efficiency gains are equivalent to saving five full workdays every week.
Salesforce has gone one step further by implementing an AI-based escalation predictor. By analyzing support history, conversation tone, and response patterns, the tool flags cases likely to escalate before the customer ever asks to speak to a manager. Managers can then proactively intervene or offer additional support to agents, preventing the escalation altogether. According to their team, this has resulted in more than a 50% reduction in escalated cases.
In each of these cases, AI didn’t take over the job of a support agent. It simply made the human part of support more impactful.
A shift from automation to augmentation
There’s a lot of noise in the market right now about AI in customer experience. But not all AI is created equally, and not all teams want to hand over the full support experience to a bot.
That’s why we champion agentic AI, not just automation.
Most AI today is reactive, It waits for you to ask a question, then replies. Useful but limited. This agentic AI can observe, reason, and take action based on emotional and operational context without the need for human intervention. They surface insights, flag risks, and deliver real-time feedback and resolutions in the tools your agents already use.
For teams that already lead with empathy, agentic AI acts as a co-pilot. It’s the system that taps you on the shoulder and says, “Hey, I think this customer needs a little more attention.”
That’s what agentic AI is about: action, impact, and results you can measure.
A playbook for rolling out emotion-aware AI (without losing your human touch)
If you’re considering implementing these tools, here’s how we recommend getting started.
1. Audit your support conversations
Look for common sources of friction, where customers get upset, and the kind of language that signals frustration. These patterns will help you train your AI models and define sentiment thresholds that matter. With Forethought, this analysis happens automatically, so your team can focus on taking action instead of figuring out customer sentiment by digging through data.
2. Start small and focused
Begin with a pilot: one queue, one channel, one segment of customers. Use sentiment detection to trigger alerts or route cases differently, and measure whether those interventions result in higher FCR or CSAT. Keep in mind that a flat CSAT can also be a positive signal, it means the AI is performing on par with human agents and maintaining the customer experience.
3. Use AI to guide, not gatekeep
Your AI should be capable of resolving customer issues end-to-end, following workflows and escalating to a human agent when needed or when requested by the customer. Still, your teams remain the ultimate owners of the relationship. AI can suggest next steps, surface articles, or flag risk, but trust is built when humans step in at the right moments.
4. Tie insights back to coaching
Use tools like Agent QA to evaluate how agents handle emotionally charged moments. Were they able to de-escalate? Did they offer clear solutions and acknowledge the customer’s feelings? This is where AI-driven insights can make your QA process more holistic.
5. Build feedback loops
As you roll out emotion-aware workflows, check in with your team. Are the alerts helpful? Are they surfacing at the right time? Are suggested replies improving handle time or leading to more natural interactions? Your team’s feedback is just as important as the model’s output and when shared with product teams, it can create a powerful feedback loop that enhances CX and product innovation.
Building a better kind of support
The future of support isn’t robotic, impersonal, or cold. It’s fast, empathetic, and grounded in real relationships: powered by AI, but delivered by humans.
With Forethought’s agentic AI platform and Boldr’s people-first approach to managed CX outsourcing and EOR, we’re helping companies unlock this vision. Not just for the occasional VIP customer, but for every customer moment.
By blending technology that understands emotion with teams that lead with it, we’re co-authoring a new kind of support story. One where service recovery isn’t just possible, it’s predictable, and where every “uh-oh” has the potential to become an “oh wow.”
