Proactive AI Support Is a Feature, Not a Replacement for Reactive Channels
CX Today reported on 28 August 2026 on comments from Mridul Ghosh, Field Digital Officer in Concentrix's Office of Technology, and Andy Lee, Executive Chairman of Crescendo, warning CX leaders against letting AI-driven proactive outreach crowd out reactive support. Both argue that predictive, behavioural-data-driven contact — flagging a delayed flight, a late order, or suspicious account activity — only works when it responds to a signal the customer actually gave. Push it further, using inferred intent from location, past activity, or unrelated interactions, and the same automation that cuts avoidable contacts starts to feel like surveillance. Their core point for buyers: reactive channels, staffed by humans who can exercise judgment on billing disputes, complaints, and sensitive issues, remain essential and shouldn't be trimmed just because a vendor's proactive module looks impressive in a demo.
What's the actual risk with proactive AI support?
Over-triggered automation turns a helpful nudge into an intrusive one, and customers start feeling monitored rather than served.
The mechanism is straightforward: AI systems infer intent from data points a customer never explicitly offered — browsing history, location, past purchases. When that inference is wrong, or simply unwelcome, the result is irrelevant recommendations or excessive notifications at scale, because the system is automating a guess, not a request. Ghosh frames the dividing line as intent: acting on what a customer signals versus what they never chose to share.
When should support teams still wait for the customer to reach out?
Billing problems, product failures, complaints, and sensitive personal issues still call for reactive, human-led service rather than predictive contact.
These are the categories CX Today singles out as poor fits for proactive automation — situations where a customer wants control over when and how the conversation starts, not a brand pre-empting them. If your roadmap treats proactive AI as a wholesale replacement for reactive queues rather than a layer on top of them, you're solving for operational cost at the expense of the interactions that most need empathy.
How do you stop "proactive" from becoming surveillance?
Tie every automated outreach to an explicit, recent customer action, and cap frequency rather than letting the model decide.
Andy Lee's framing is useful here: the test isn't how much automation you're running, it's whether the customer experience is measurably better because of it. That's a harder question to answer than "did we send the notification," and it's the one vendors rarely volunteer metrics for.
What should buyers ask their AI support vendor before switching?
Ask how the vendor defines "proactive success," who measured it, and whether reactive headcount was cut to fund the rollout.
This site has already flagged that vendor confidence in CX AI claims is running ahead of verifiable evidence — see our pieces on Avaya's AI strategy warning and on why polish stops counting as proof once every vendor sounds excellent. The same scepticism applies here: a vendor citing fewer inbound contacts as proof of proactive success is measuring deflection, not satisfaction. Ask for complaint-rate and repeat-contact data specifically from customers who received proactive outreach, not aggregate figures.
Frequently asked questions
Does this mean proactive AI support isn't worth buying?
No — both sources still see clear value in it for well-signalled, low-ambiguity cases like delay alerts. The warning is against over-extension, not adoption.
Which vendors were named in the coverage?
Concentrix and Crescendo, through comments from Mridul Ghosh and Andy Lee respectively. No product names or performance figures were disclosed.
Should reactive support staffing shrink as proactive AI rolls out?
The sourced advice is no — reactive channels handle categories, like complaints and billing disputes, that proactive automation isn't suited to at all.
Source: CX Today — "Why You Shouldn't Abandon Reactive Support For Proactive AI," 28 August 2026.
Proactive customer service promises a more seamless experience, enabling brands to flag delays, prevent problems, and offer assistance before customers need to ask. Yet, as more organizations use AI to interpret behavioural data and predict intent, the difference between a useful interaction and an intrusive one is becoming harder to manage. CX leaders…