Salesforce Buys the Best Independent AI Support Agent on the Market
Salesforce signed a definitive agreement in mid-June to acquire Fin, the renamed Intercom, for roughly $3.6 billion, folding it into Agentforce. Fin, rebranded from Intercom in May, runs an AI service agent on its proprietary model and resolves queries across live chat, email, WhatsApp, SMS, phone, and Slack, reportedly bringing around 30,000 AI customers with it. The deal is expected to close in Salesforce’s fiscal Q4 2027, subject to regulatory approval, with CEO Eoghan McCabe staying on. (The $3.6B figure comes from TechCrunch; Salesforce’s own release confirms the agreement.)
This is the biggest consolidation signal in CX tech this cycle, and it is worth being clear-eyed about what it removes. Fin was the most credible independent AI-support pure-play in the market, the one a mid-market company could adopt without buying into a mega-suite. Salesforce just took that option off the table and pointed it downmarket, at the SMB and commercial buyers it has always struggled to win. The message to the rest of the field is blunt: agentic service is now a suite feature, and the independents are acquisition targets.
For CX leaders, the practical fallout depends on where you sit. If you run Intercom or Fin, start planning for the Agentforce roadmap and the integration questions that come with any acquisition of this size. If you are on Zendesk or another stack, expect sharper Salesforce competitive pressure and pricing moves aimed at pulling you into the ecosystem. And if you are evaluating standalone AI-support vendors right now, factor consolidation risk into the decision, because the pattern this deal confirms is that today’s independent favorite is tomorrow’s acquired feature. The tool you pick is also a bet on who owns it in two years.
The $2 Resolution Arrives
Alongside the Fin deal, Salesforce made Agentforce Help Agent generally available with a pricing model that deserves your attention more than any feature: a flat $2 charged only when the AI actually resolves an issue end to end. No charge if the customer escalates to a human or leaves unhappy. The agent grounds itself on Salesforce Knowledge, ships with prebuilt actions for case management, order management, and scheduling, and turns on across voice, web, portal, and messaging from a single screen. Salesforce’s own deployment reportedly handled 4.3 million inquiries and resolved 70% of them.
Outcome-based pricing is the real product here, and it changes the CX budgeting conversation at a structural level. For years, support economics were a headcount math problem: seats, agents, cost per ticket. “Pay only when it resolves” reframes the whole thing around resolution economics, where your spend scales with successful outcomes instead of licenses. That is a very different way to build a business case, and it is friendlier to finance, because you are buying results rather than capacity you hope to use.
But read the fine print in the model itself, because the entire thing hinges on how you define a “resolution.” If the vendor gets to decide what counts as resolved, the incentive to mark borderline interactions as wins is obvious. CX leaders evaluating outcome-based AI need to interrogate the resolution definition the way they would interrogate any metric tied to a bill: what counts, who verifies it, and can you audit it. This is where the discipline of designing your actual customer experience matters more than the pricing headline. A cheap resolution that leaves the customer half-served is not cheaper. It is churn with a discount.
Zendesk Answers With “Verified Resolutions,” and the Definition Becomes the Battleground
Zendesk spent 2026 repositioning around the same idea from a different angle. Coming out of its Relate 2026 conference and rolling into its July release, Zendesk pushed an “Autonomous Service Workforce” built on AI agents billed per verified resolution rather than per seat, priced at roughly $1.50 per committed resolution or $2.00 pay-as-you-go. The tell is the word “verified”: every billed resolution is dual-checked by the AI agent and a separate AI evaluation model, with spam and routine exchanges excluded. Zendesk also made its Forethought AI agents available as an add-on and shipped new predictive routing. (The Relate vision unveiling was May 18-20 in Denver, so this straddles the window; the July availability keeps it current, and exact July feature details should be confirmed.)
When both Salesforce and Zendesk anchor to “you pay when it’s resolved,” you are not looking at two pricing experiments, you are looking at an industry pivot. And Zendesk’s dual-verification mechanism is the smartest part, because it targets exactly the anxiety the model creates. Buyers grew wary of paying for AI “resolutions” that were not resolutions. Zendesk’s answer is to make the billing auditable: two models have to agree it was resolved, and the obvious junk gets excluded before you are charged.
That is the new competitive front for CX leaders to watch, and it belongs in your next RFP. The question is no longer just “how good is the AI agent.” It is “how do you define a resolution, how do you verify it, and can I trust the number I am being billed against.” Vendors are about to compete on the trustworthiness of their billing, not just the capability of their bots, and that is a healthy development for buyers who have been burned by deflection metrics that looked great and served customers badly.
The Money Keeps Flowing to Independent CX Players
Even after Salesforce swallowed Fin, capital is still finding independent CX vendors, especially in voice and regulated industries. Omilia, an enterprise voice-first agentic CX company, secured a $67M (€58.1M) Series B in early August, led by Expedition Growth Capital, reporting more than 10x live-ARR growth since its Series A to over $60M and plans to open its first US office in the second half of the year. Separately, India’s Kapture CX raised $10M in late June, led by Bajaj Finserv Ventures. (Omilia is well corroborated across multiple outlets; the Kapture terms come from a single funding tracker, so verify before citing.)
The signal to read here is that the vendor landscape is not collapsing to three incumbents, whatever the Fin deal implied. Specialized challengers, particularly in voice AI and compliance-heavy verticals like banking, insurance, and healthcare, are still raising real money and growing fast. Omilia’s 10x ARR jump without an interim raise is the kind of number that says enterprises with hard requirements are actively buying outside the mega-suites.
For CX leaders, that is useful leverage. The narrative that “everything is consolidating, just buy the suite” is convenient for the suites and only partly true. If you have specialized needs, regulated data, complex voice workflows, industry-specific compliance, there are well-funded alternatives that will fight for your business and shape their roadmap around your vertical. Consolidation at the top does not mean surrender at the edges. It means you should know which of your requirements are generic enough for a suite and which are specialized enough to justify a specialist, and buy accordingly.
Forrester Says North American CX Finally Turned the Corner
After years of the “CX quality is declining” storyline, Forrester’s 2026 CX Index suggests the trend may be reversing, at least in North America. Of brands measured in both 2025 and 2026, roughly 20% of CX Index scores improved versus about 8% that declined, and in Forrester’s broader 375-brand Total Experience set, 41% improved against just 3% that slipped. Europe and APAC stayed flat. But the improvement was not universal: US federal government logged its first-ever significant decline, with its average CX Index score falling from 61.9 in 2025 to 59.4 in 2026 on the 100-point scale, and sectors like airlines and credit-card issuers also slid. (The federal figure comes from a single Forrester blog; high-credibility source, but cross-check before quoting.)
The rebound gives CX leaders something they have not had in a while: a data-backed argument that investment is paying off. When the multi-year decline finally bends upward in your region, that is ammunition for the budget conversation, evidence that the work of the last few years produced measurable gains rather than just activity.
The sector-specific declines are the more instructive half, though. Airlines, card issuers, and federal government are cautionary tales about what erodes trust: friction, opacity, and experiences that treat the customer as a cost to be managed rather than a relationship to be kept. The lesson is not “CX is fixed.” It is that the gains are unevenly distributed, and the losers share a profile. The brands that slipped are the ones where the experience across the whole journey quietly degraded while attention went elsewhere. Improvement is possible and now proven. It just is not automatic, and it does not survive neglect.
Gartner Confirms What Every CX Leader Already Feels
If it seems like the pressure to “do something with AI” is coming from above, the data agrees. A Gartner survey found that 91% of customer service and support leaders report pressure from executive leadership to implement AI in 2026, with top priorities being CSAT, operational efficiency, and self-service success. Notably, 58% of service leaders aim to upskill agents into knowledge-management specialists as self-service scales. Gartner’s standing prediction is that by 2029, agentic AI will autonomously resolve 80% of common service issues and cut operational costs 30%. (The 91% survey published in February 2026, so it is a few months old, but the pressure it describes is very much the live 2026 story.)
The 91% quantifies a feeling every CX leader recognizes: the board wants AI, now, and the mandate arrived before the plan did. That pressure is real and it is not going away, so the useful move is not to resist it but to direct it toward outcomes that actually hold up.
The 58% upskilling stat is where the smart play hides. The reflexive read of “AI resolves 80% of issues” is headcount reduction. The more durable read, and the one Gartner’s own data points at, is redeployment: as AI absorbs the routine volume, your best people move into knowledge management, quality oversight, and the complex, high-emotion cases where human judgment is the entire value. The organizations that treat this as “cut the team” will hollow out the expertise that makes the AI good in the first place, because someone has to write and maintain the knowledge the agents ground themselves on. The ones that treat it as “move humans up the value chain” will end up with better automation and better hard-case handling. Same technology, opposite outcomes, decided by how you frame the mandate.
Qualtrics Reminds Everyone the Customers Are Quietly Leaving
For a needed counterweight to the AI-everything momentum, Qualtrics XM Institute research warns that poor experiences could put nearly $3 trillion in global sales at risk, with 34% of consumers cutting spending after a bad experience and 13% cutting it entirely. The most pointed finding: nearly one in five consumers who used AI for customer support saw no benefit at all, a failure rate roughly four times higher than other AI use cases. On top of that, only about three in ten customers now give direct feedback on why they leave, down year over year, and 71% of CX practitioners rate their own organization at one of the two lowest maturity stages. (The $3T figure is Qualtrics’ own modeled estimate; the study draws on 27,000-plus consumers.)
Put the two halves together and you get the uncomfortable truth of this month. AI support is being deployed under intense board pressure, and it is specifically the AI use case customers say is underdelivering, while most of them do not even bother telling you why they left. That is silent churn happening right underneath an AI rollout that leadership assumes is a win.
The takeaway for CX leaders is to hold your AI deployments to a measurable benefit bar, not just a deflection rate. A bot that deflects 60% of tickets while quietly annoying a fifth of the people who touch it is not a success, it is a churn engine with a good dashboard. Instrument the silent departures: figure out who is leaving without complaining, and whether your automation is part of why. That is the experience work that does not show up in a deflection metric but shows up in retention, which is the number that actually pays the bills. Deflection is easy to celebrate. Retention is the one that is true.
Our Take on the August 2026 CX News
The tension running through every story this month is the same: the industry is racing to automate customer service at exactly the moment its own research says customers are unimpressed by automated customer service.
Salesforce bought the best independent AI agent and priced resolution at $2. Zendesk answered with verified resolutions and a billing model designed to be trusted. Investors keep funding voice and regulated-industry specialists. Gartner says 91% of leaders are under board pressure to deploy AI. And then Qualtrics quietly notes that AI support is the use case customers find least helpful, that a third of them cut spending after a bad experience, and that most of them will not even tell you why they walked. Forrester offers the one truly hopeful data point, North American CX is improving, but the sectors that declined all share the profile of neglected, friction-heavy journeys.
The throughline is that the pricing revolution and the capability race are both real and both beside the point if the underlying experience is bad. Outcome-based pricing is a better way to buy. Verified resolutions are a better way to trust the bill. Neither one makes a poorly designed customer journey good. A $2 resolution that leaves the customer half-served still costs you the customer, and the model will happily bill you for it. The vendors are competing on how they charge for AI. The customers are still just deciding whether they felt served.
So the move for CX leaders this month is to take the board’s AI mandate and aim it at the thing that actually retains customers: a coherent journey where the automation handles what it is good at, humans handle what requires judgment and empathy, and the whole experience is instrumented well enough to catch the people leaving in silence. Use the outcome-based pricing, it is a real improvement. Interrogate the resolution definitions, they matter. But do not confuse a cheaper, better-billed bot with a better experience. The experience is still the product. The AI is just how you deliver more of it, or, if you are not careful, how you disappoint people faster and get invoiced for the privilege.
August 2026 CX Events
August is a thin month for flagship CX conferences (most cluster in June and again from September through November), so this is a short but real list.
Contact.io 2026 August 23-25 | Hyatt Regency Denver at Colorado Convention Center, Denver, CO A customer-contact and call-marketing conference for operators and growth leaders focused on turning conversations into revenue: inbound call programs, outbound lead response, contact-center performance, compliance, and AI-powered customer contact. Draws 800-plus attendees across pay-per-call marketing, contact-center leadership, and CX tech. https://contact.io/
Support Driven Expo US 2026 August 24-26 | DePaul University Student Center, Chicago, IL A practitioner-run customer-support community conference with 100-plus peer-led talks, panels, and workshops. Deliberately vendor-neutral, no sales demos or product pitches, focused on support careers, operations, and leadership. A strong pick for building your team’s craft rather than shopping for tools. https://www.supportdriven.com/support-driven-us-expo-2026
Forrester CX Summit APAC 2026 August 25 | Sheraton Grand Sydney Hyde Park, Sydney, Australia Forrester’s one-day flagship APAC customer-experience summit for CX professionals, digital leaders, and B2C marketers, themed “The Total Experience: Your Brand, Their Journey.” Focused on CX strategy, measurement, and leadership alignment, and a chance to see Forrester’s Total Experience research applied to the region. https://go.forrester.com/event/cx-apac/