Salesforce announced it has signed a definitive agreement to acquire Fin, formerly Intercom, an industry-leading customer agent company. Under the terms of the agreement, Salesforce will acquire Fin for approximately $3.6 billion, subject to customary purchase price adjustments.

Fin’s core offering, its AI Agent, resolves complex customer queries end-to-end, across every channel, including live chat, email, WhatsApp, SMS, phone, and Slack. The AI Agent is powered by the company’s proprietary AI model, Apex, which is purpose-built for customer support and has demonstrated industry-leading resolution rates that outperform top commercially available frontier models.

“We’re thrilled to welcome Fin to Salesforce as we enable every company to become an agentic enterprise,” said Marc Benioff, Chair and CEO, Salesforce. “Fin brings proven agent technology, a deep commitment to customer success, and an incredible AI team that will complement Agentforce with powerful service agent capabilities. Together, we’ll help companies of every size seize this opportunity — accelerating time to value with trusted agents that deliver measurable outcomes at scale.”

Also Read: Payment Experience Is the Foundation of B2B Loyalty

“This is a major win for consumers of the world,” said Eoghan McCabe, Chief Executive Officer and Co-Founder of Fin. “Our technology has defined this category and set the new standards for what great customer service looks like today. By joining forces with Salesforce, we can deploy it far and wide at a rate far faster than we could have ever achieved on our own.”

Building on the strength of Agentforce, which reached $1.2 billion in ARR in Q1 FY27, up 205% year-over-year, Fin’s packaged offerings and proprietary models will complement Agentforce’s deeply customizable platform with additional fast-to-value deployment options for service organizations.

Upon close, Salesforce and Fin will give customers more ways to deploy AI agents across their customer service operations, with fast time-to-value options especially well-suited for SMB and some commercial organizations that need to launch quickly, integrate with existing systems, and deliver measurable outcomes. Together, Salesforce and Fin will support customers at every stage of AI adoption, from rapidly deployable support agents to more tailored, enterprise-scale transformations built on trusted data, security, governance, and integration.

Also Read: Is AI About to Make Media Buying an Endless Experiment?

Fin’s AI agent technology will help organizations improve autonomous resolution, reduce cost-to-serve, and accelerate AI adoption across their service organizations. The AI Agent has already demonstrated strong customer outcomes, including examples of AI agents resolving on average 76% of support volume end-to-end. The acquisition will also bring a long-tenured technical AI team and an established global customer base of more than 30,000 companies to Salesforce.

The post Salesforce to Acquire Fin for $3.6 Billion appeared first on MartechView.

Gopuff, the leader in instant commerce, and GrowthLoop, a pioneer in agentic AI-powered marketing solutions, today announced a partnership to enable custom audience segmentation for Gopuff and BevMo! advertising partners. With GrowthLoop’s composable commerce media solution, brands advertising with Gopuff and BevMo! will be able to leverage new audience targeting, activate campaigns quickly, and gain visibility into campaign metrics.

“Retail media is one of the fastest-growing channels in advertising, but too many networks are limited by disconnected data and slow execution,” said Anthony Rotio, co-founder and co-CEO of GrowthLoop. “Gopuff has a unique advantage with high-intent customers making purchase decisions and a Gen Z audience that brands are eager to understand. By leveraging GrowthLoop, Gopuff’s partners can act instantly on first-party data to reach the right customers at the moment they’re ready to buy.”

With the new integration, Gopuff and BevMo!’s ad partners can move beyond broad category targeting to segment buyers by geography, purchase history, and order frequency, then activate those audiences via sell-side platforms (SSP) or demand-side platforms (DSP). For example, within a few days, a brand could request and launch an audience based on frequent buyers of its top five competitors living in college markets. Later this year, brands will also be able to target these audiences for sponsored product ads in the Gopuff app and utilize the new capabilities to target customers in the U.K.

“At Gopuff, our entire business model is built around precision and speed,” said JR Crosby, Director of Data Partnerships at Gopuff. “But historically, creating custom audience cohorts isn’t something brands can turn around quickly. By integrating GrowthLoop, we are now able to give brands exactly what they’ve been asking for: the ability to build and test with really granular audience segments both on Gopuff and through our marketplace partners, all with a turnaround time that was previously impossible.”

Also Read: AI Is Supercharging Returns Fraud. Retailers Are Behind.

GrowthLoop’s platform unifies audience creation, campaign activation, and measurement directly on Snowflake’s AI Data Cloud, eliminating the need to move data across systems and ensuring campaigns are powered by the most complete and up-to-date customer information. The integration is now live for Gopuff’s beta partners, with the new capabilities rolling out to all brands over the coming weeks.

As Gopuff continues to expand its retail media capabilities, its partnership with GrowthLoop supports its broader mission to deliver more personalized, relevant experiences to customers while creating new opportunities for brand partners to connect with its growing community.

The post Gopuff and GrowthLoop Expand Retail Media Targeting appeared first on MartechView.

Enterprise systems are complex because the real world is complex. Every automation, workflow, data model, and report reflects a decision the business needed at the time. The real challenge now is to make that complexity legible enough for both people and AI agents to act on it safely, keeping the valuable parts and dropping the rest.

Sweep, the agentic layer for enterprise systems, announced the launch of its cross-platform agent for Salesforce, Snowflake, and Data 360. The new capability allows teams to reason across all three systems simultaneously through a continuously updated dependency graph, so they can see how enterprise data, workflows, metadata, permissions, and business logic connect before making any change.

Generic AI was built for clean, greenfield environments. But Enterprise AI has to work inside real systems that carry years of complexity, which often require first-hand context to understand and build upon. 

The gaps between systems, particularly, are where complexity compounds, explains Sweep CEO and co-founder, Ido Gaver.

Also Read: Your ERP Is Holding You Back. Here’s How to Fix It.

“Every platform wants to be the agentic layer for its own domain. None of them can see what happens in between,” says Gaver. “Enterprise system complexity grows with the business. Sweep gives teams a way to embrace that complexity, apply context, and use AI to discover, design, build, and monitor change across the systems that actually run the enterprise.”

When agents act with incomplete context, ordinary business logic can become operational risk. For example, an agent might approve an expedited order because they can see the customer record in Salesforce, but miss a credit hold stored downstream or a data-quality exception surfaced in Snowflake. Another team might change a picklist or workflow in Salesforce without seeing how that value feeds a Data 360 audience, a Snowflake report, or a downstream compliance process. In both cases, the issue is not the agent’s reasoning ability. It is the absence of cross-system context.

Sweep addresses that problem by making the connected stack understandable to both teams and agents. Instead of stitching together one-off queries, data catalogs, and platform-specific copilots, Sweep creates and maintains a live map of dependencies across Salesforce, Snowflake, and Data 360. Teams can ask plain-language questions, understand impact, and govern changes before they ship.

Also Read: Why the CMO Now Owns the Privacy Problem

How Sweep’s Cross-platform Agent Works

• Discover: Sweep automatically indexes metadata, schema, permissions, automations, and dependencies across connected platforms, creating a shared map of how revenue and data infrastructure actually operate.
• Design: Teams can use natural language to then ask cross-platform questions before making a change, such as which reports, data products, automations, audiences, or processes depend on a field, object, workflow, or dataset.
• Build: Sweep supports governed execution in Salesforce, where teams can turn approved plans into changes while preserving the context and impact analysis behind them.
• Monitor: Agentic Governance continuously scans connected systems for drift, permission inconsistencies, policy violations, and dependency risks, providing teams with a single source of truth across a distributed architecture.

Sweep’s internal benchmarks, derived from analysis of more than 12,000 anonymized chats between their customers and agents, show that what typically takes enterprise teams 11 hours of cross-system investigation is compressed to roughly 1 hour. That’s because dependencies are already mapped and traced, and the downstream impact is already visible before the change ships.

The post Sweep Launches AI Agent for Salesforce and Snowflake appeared first on MartechView.

Modernization Is Now a Business Priority

Enterprise modernization has reached a point where delays carry business risk, and careless speed poses equal danger. Many organizations still depend on ERP landscapes shaped by years of local decisions, customizations, acquisitions, compliance needs, and short-term fixes. They run the business, yet often make it harder to see clearly, respond quickly, and scale.

Selective Transformation Offers a More Practical Path

The answer is not a reckless replacement of everything that exists. It is also not a technical conversion that carries yesterday’s complexity into a newer environment. A more responsible path sits between those extremes. Selective transformation gives leaders a way to protect what still has value while removing what has become a burden. Historical data, proven controls, and essential operating knowledge can be retained. Outdated code, fragmented processes, and avoidable variation can be reduced.

A Clean Core Creates Room for Progress

A clean core is not a technology slogan. It is a management discipline. It asks the enterprise to standardize common processes, limit unnecessary customization, govern integrations, and keep the business’s core ready for improvement. When the core is crowded with exceptions, every upgrade becomes harder. Every report becomes debatable. Every innovation effort starts with a cleanup.

For senior leaders, the clean core should be viewed in business terms. Finance needs trusted numbers. Operations need reliable signals. Supply chain teams need visibility across demand, inventory, suppliers, and plants. Compliance teams need evidence that controls are working. Employees need systems that do not force them into manual workarounds. Customers need commitments that can be met. None of this is possible when the foundation is unstable.

Also Read: Is AI About to Make Media Buying an Endless Experiment?

Public Cloud Readiness Requires Honest Assessment

Public cloud readiness belongs in the same conversation. Public cloud ERP models are attractive because they encourage alignment with standard thinking, shorter deployment cycles, predictable upgrades, and a lower infrastructure burden. They can help growing enterprises move away from fragmented systems and toward a more consistent operating platform. Yet, public cloud should be adopted honestly. It works best when the business is ready to accept standard ways of working.

That distinction matters. Some processes are truly differentiating. Others are merely familiar. A mature modernization program separates the two. The enterprise should not customize a new core simply because the old system carried a certain practice for years. At the same time, critical regulatory, manufacturing, quality, or industry-specific needs should not be dismissed casually. Good leadership wisely uses standardization to create speed, control, and scale.

Industry Needs Should Shape the Modernization Journey

Industry realities make this balance more important. Manufacturers may need better costing across products and plants. Utilities may need modernization without disruption to regulated asset operations. Life sciences companies may need validation discipline, audit readiness, and data integrity. Automotive suppliers may need faster carveout execution, partner integration, and production visibility. Food and dairy businesses may need traceability and recall readiness. Chemical companies may need batch insight, yield control, and margin protection. These challenges are different, but they point to the same principle. The enterprise core must be standardized enough to scale and flexible enough to respect real operating needs.

AI Readiness Begins with the Foundation

Artificial intelligence has made this discussion more urgent. Intelligent tools depend on reliable data and disciplined processes. When data definitions differ across functions, recommendations become questionable. When workflows are unclear, automation may introduce errors more quickly. When integrations are fragile, digital operations become difficult to trust. AI readiness begins long before a model is introduced. It begins with architecture, data, process ownership, and governance.

Also Read: Brands Are Making ‘No AI’ Their Biggest Selling Point

Digital Workers Need Governance and Accountability

The same thinking applies to software agents and digital workers. These capabilities may assist with monitoring, analysis, exception handling, and workflow execution. Their value will depend on control. Each digital worker should have a defined purpose, a clear owner, approved boundaries, escalation paths, and performance measures. Human oversight should remain visible where decisions affect customers, employees, financial results, safety, or compliance. Without that discipline, automation can create risk.

Transformation Must Also Support People

Modernization also has a human side. Employees experience transformation through daily tasks, not through board presentations. They will judge success by whether the new process is clearer, whether training is practical, whether leaders explain the reason for change, and whether the system helps them do better work. When people are left to interpret change on their own, uncertainty grows. When they are supported, adoption becomes more natural.

This requires leadership beyond the technology function. Finance, operations, supply chain, human resources, risk, and business unit leaders must be involved early. Process owners must be willing to make decisions. Governance must be active. Change management must be practical. A modern ERP program should not be handed to IT and reviewed only at milestones. It should be a business transformation enabled by technology.

Business Outcomes Should Define Success

The measure of success should also become more grounded. A successful program should improve cost visibility, working capital control, reporting confidence, inventory accuracy, customer response, compliance quality, productivity, and resilience. These outcomes matter more than the number of features launched. They also matter more than speed if speed comes at the expense of readiness.

The current moment should be treated as a chance to simplify with courage. Aging ERP environments have forced many organizations to make decisions that had been delayed too long. That pressure can be uncomfortable, but it can also be useful. It gives leadership a reason to remove complexity and build a cleaner foundation.

Also Read: Are Brands Losing Credibility in the AI Era?

The Future Belongs to Enterprises That Modernize with Discipline

The organizations that move wisely will not confuse modernization with migration. They will protect continuity while reducing debt. They will adopt public cloud where standardization creates value. They will assess specialized operations carefully. They will prepare data before scaling AI. They will govern digital workers before granting autonomy. They will bring employees through the change with clarity and respect.

Modern ERP modernization is ultimately about confidence. Confidence that the business can change without losing control. Confidence that data can be trusted. Confidence that people can adopt new ways of working. Confidence that intelligent operations can scale responsibly. When clean core discipline, public cloud readiness, industry awareness, AI governance, and business continuity are aligned, the enterprise gains more than a new system. It gains a stronger foundation for the future.

The post Your ERP Is Holding You Back. Here’s How to Fix It. appeared first on MartechView.

Adzymic announced the launch of the AgenX Creative Agent, the first release in its agentic suite, marking a structural shift in the digital advertising ecosystem: from fragmented, manual workflows to interoperable, agent-to-agent execution built on open protocols. Alongside the Creative Agent, Adzymic is unveiling Agent as a Service (AaaS), a new subscription model that opens the AgenX platform to brands, agencies, and media owners globally.

Positioned at the intersection of creative technology and media infrastructure, AgenX extends Adzymic’s established role as a dynamic creative optimization (DCO) provider into an emerging paradigm where autonomous agents transact, optimize, and deliver advertising outcomes across channels.

Creative Agent anchors next-generation ad creation

At the core of the AgenX suite is its Creative Agent – an industry-first agentic creatives platform capable of generating rich media and interactive HTML ad units autonomously from a single campaign brief, producing assets across formats, sizes, and languages.

Beyond automation, the system incorporates brand governance at the point of creation. By ingesting brand guidelines, including visual identity systems, typography, color palettes, and tone of voice, the agent ensures that generated creatives remain consistent with brand standards while maintaining creation speed.

One of the platform’s pioneering adopters is long-standing client Cathay Pacific Airways. Karen Cheung, Performance Marketing Manager at Cathay Pacific, noted, “What stands out is the breadth and consistency of outputs generated at scale. From a single brief, we are seeing a wide range of high-quality, brand-compliant assets that are immediately deployable across formats. The ability to traffic these dynamically, whether programmatically, directly, or through agentic workflows, creates a far more agile testing environment for campaign performance.”

Another pioneering user of the platform is Mediacorp, Singapore’s national media network and largest content creator. Raj Parekh, Head of Growth & Partnerships, at Mediacorp, said, “At Mediacorp, we are always looking for ways to help partners and clients connect more meaningfully with audiences across our premium, brand-safe inventory. Our work with Adzymic has supported this by enabling us to deploy richer ad formats and use AgenX outputs to develop more responsive sales pitches. The speed of delivery and quality of creatives have been a game-changer, allowing us to bring more value to advertisers while strengthening the overall campaign experience. We are delighted to deepen our partnership with Adzymic, as we continue to explore new ways to help brands connect meaningfully with their audiences.”

Also Read: Hyper-Automation Is Over. Agentic AI Is What Comes Next.

The AgenX suite – what comes next

The Creative Agent is the first of three agents in the AgenX suite. The Sales Agent, designed to represent publisher inventory within agent-driven transactions, and the Buyer Agent, enabling autonomous media planning and activation, will each be the subject of dedicated launches as the agentic advertising ecosystem matures. Together, the three agents form a closed-loop agentic system across creative generation, inventory monetization, and media execution.

AgenX is built on AdCP (Ad Context Protocol), the emerging open standard for agentic media transactions, alongside Prebid Sales Agent and MCP (Model Context Protocol). This means AgenX is natively interoperable with any AdCP-compliant buyer or supply agent, including the agent-to-agent buying frameworks that major buy-side players are now actively deploying. The suite operates globally across all Adzymic markets.

Deanson Lee, General Manager for Southeast Asia at Ebiquity, a global media investment analysis firm, noted, “Agentic advertising represents one of the most meaningful structural shifts in how the industry operates in recent years. While much of the early momentum has been driven by the US and Europe, Southeast Asia is now approaching its own turning point. We are encouraged to see a homegrown company like Adzymic stepping forward as both enabler and operator, bringing agentic capabilities to publishers, agencies and clients across the region, playing a catalytic role in shaping how this new operating model is adopted and scaled across Southeast Asia.”

Adzymic’s Co-founder, Travis Teo, concluded, “We are at the inflection point of a structural shift in how advertising operates. Major holding companies are actively testing and executing agent-to-agent media transactions, signaling that agentic infrastructure is no longer a future consideration but an immediate commercial reality. With AgenX, we are building the infrastructure for this next phase, enabling intelligent agents to collaborate across the entire value chain. Our ambition is to help shape an open, interoperable ecosystem and to be at the forefront of how this new model evolves.”

Agent as a Service expands access to AgenX

For brands and agencies running campaigns through Adzymic’s APX 36One Managed Campaign service, AgenX-powered ad generation is included at no additional cost as part of every engagement, alongside campaign operations and access to premium omnichannel inventory across Adzymic’s network.

In parallel, Adzymic today introduced Agent as a Service (AaaS), a standalone subscription offering that provides brands, agencies and media owners with direct access to the AgenX platform, independent of Adzymic’s managed services.

Designed for organizations operating their own programmatic infrastructure, the subscription model enables direct access to the Creative Agent, with Sales Agent and Buyer Agent capabilities progressively activated as market readiness increases.

The post Adzymic Launches AI Agent to Automate Ad Creation at Scale appeared first on MartechView.

Madhive, a DSP and AI solutions partner built specifically for local brands and agencies, today announced the expansion of its enterprise intelligence layer, Maverick AI, to include new agentic capabilities designed to effortlessly and efficiently drive performance. A new suite of Maverick AI agents embeds local-first intelligence directly into Madhive’s DSP, enabling seamless execution while surfacing real-time insights for advertisers at every step, from planning and activation through measurement.

Introduced last year, Maverick AI set a new industry standard, fueling Madhive’s entire platform for smarter, faster campaign planning and accelerated revenue. As the intelligence layer powering Madhive’s unified platform from beneath the surface, Maverick AI has evolved into a full suite of agentic capabilities that will roll out over the next year. 

“Local advertising has always been performance-driven at its core—defined by execution, accountability, and results,” said Jim Wilson, CEO of Madhive. “With Maverick AI agents embedded directly into our DSP, we’re enabling advertisers to close the loop between insight, activation, and performance in real time—unlocking a new level of efficiency and impact for local campaigns.”

Also Read: The CIO Who Says Governance Can Actually Speed Up AI

While most DSPs are not built for the complexity of local markets, Madhive’s AI agents are trained on more than a decade of proprietary data generated from its local-first platform. Ingesting signals from over 50,000 daily campaigns, these agents are uniquely equipped to navigate the nuances of local media, accounting for geographic fragmentation, budget variability, seasonality, and the pacing dynamics that define performance at the local level.

By embedding agentic AI across the platform, these agents serve as an operational thread connecting every stage of the campaign lifecycle – from initial discovery and pitching to planning, activation, optimization, and reporting. They work alongside media teams by day while continuously driving performance around the clock. 

Maverick AI agents enable:

• Faster media plan creation using historical performance and audience data
• Activate campaigns across local inventory with speed and precision
• Continuously optimize delivery and budgets in real time

Also Read: Peak Martech? The Landscape Has Plateaued, but the Real Story Lies Beneath

Alongside the AI agentic capabilities, Madhive has introduced its Model Context Protocol (MCP) server, an offering that allows multiple LLMs to connect directly to Madhive’s platform, exposing Madhive’s functionality through a standard interface that allows users to interact with campaigns, creatives, audiences, and publisher groups using natural language commands. The MCP server lays the groundwork for more advanced, agent-powered execution by allowing enterprises to bring local intelligence directly into their own tools, workflows, and systems. 

“Insights have always been at the core of how FOX drives performance in local markets,” said Michael Page, SVP, Digital Sales at Fox Television Stations. “We’re excited that Maverick AI agents will not only enhance that capability, but they will also fundamentally accelerate it, turning insight into immediate action and performance at scale.”

The post Madhive Expands Maverick AI for Local Ad Buying appeared first on MartechView.

For more than a decade, the enterprise world chased hyper-automation. Boards, consultants, and venture capitalists all shared the same dream: big productivity gains delivered by clever workflow tools, low-code connectors, and robotic process automation (RPA). Top-tier valuations, a parade of IPOs, and billion-dollar rounds crowned automation as the next trillion-dollar frontier.

But today that narrative is collapsing — not because automation isn’t useful, but because the architecture underpinning it is fundamentally obsolete. What the market once celebrated as “hyper-automation” is now being written off as incremental plumbing — not strategic leverage.

Public markets and valuations aren’t bluffing — they are signaling a tectonic shift. 

Public Market Reality: When Automation Valuations Drop Hard

One of the poster children of hyper-automation was UiPath. Valued at over $35 billion in late-stage private funding, UiPath’s public market debut was among the largest software IPOs of the 2020 cohort. But today, the stock trades 70% below its peak, reflecting a stark reset in what public markets are willing to pay for legacy automation that fails to expand margins or deliver sustained enterprise outcomes. 

This sell-off isn’t isolated — it’s a shift in valuation narrative. Capital now demands results, not engineering complexity. Ask yourself: if automation fundamentally transformed productivity across every business unit, why haven’t legacy automation stocks maintained their valuations in an era obsessed with AI growth?

The answer is simple: legacy automation still operates like software from the 1980s — static, linear, brittle workflow logic under the hood — while the world around it has become exponentially more dynamic.

Also Read: E-commerce Doesn’t Have a Data Problem. It Has a Speed One.

The Funding Frenzy That Built Fragile Architecture

While UiPath faced valuation compression, private markets saw a secondary boom in no-code and security automation startups — each promising to elevate hyper-automation through low-code workflows and drag-and-drop playbooks.

Consider Tines, a Dublin-based no-code automation platform focused on security workflows, with hundreds of millions in funding across multiple rounds. 

Meanwhile, Torq — another “hyper-automation” platform — raised a massive $140 million in its Series D at a $1.2 billion valuation in 2026, bringing its total raised to $332 million. 

These companies embody exactly the market the last decade valued: drag-and-drop workflows that connect tools, handle alert triage, and automate repeatable tasks. Yet the core architecture for these systems remains deterministic-first — defined by prewritten steps assembled into stories or playbooks that attempt to anticipate every possible state of the world. That’s fine for checklists — less so for real business outcomes. 

Why This Architecture Is Headed Toward Zero

Here’s the uncomfortable truth:

Hyper-automation didn’t redefine how work actually gets done. It repackaged deterministic workflows with prettier UIs and AI buzzwords. The entire paradigm assumes:

• Enterprise processes are static enough to be defined in workflows.
• Users can anticipate every possible branch in a decision tree.
• You can write your way around complex human-machine interaction.

But the 2020s enterprise is not static. The business landscape, cloud ecosystems, security threats, and digital environments change by the minute. Legacy architecture cannot adapt because it starts with rules rather than objectives.

When markets recognize that a product’s DNA can’t survive the world it claims to automate, they repriced those assets accordingly. That’s exactly what’s happening with companies built on the old stack. 

Public Sentiment Is Shifting — Investors Want Outcomes, Not Workflows

The public markets — and savvy late-stage investors — are increasingly separating automation as a product from real enterprise leverage. The headlines today aren’t about bots mimicking clicks anymore — they’re about agents that act autonomously on business objectives.

Investors are no longer content to pour capital into incremental workflow stitching. Capital is chasing systems that can:

• Interpret intent, not just follow static rules
• Plan and execute across systems, not just trigger steps
• Adapt to uncertainty, not crash when conditions change

The evidence is clear: when the underlying architecture is rigid, valuations get compressed. Investors won’t buy another round of “better connectors” if the fundamental utility is low-margin and rigid. 

Also Read: Contextual Advertising: What It Is and Why It Matters

Agentic Automation Is What Comes After Hyper-Automation

The next wave isn’t a prettier drag-and-drop canvas. And it definitely isn’t another workflow builder packaged as the next big thing.

It’s agentic automation — systems built for objectives, not sequences; for dynamic coordination across tools, not fixed playbooks. An agentic system understands intent, negotiates tasks across APIs, adapts to context, and achieves outcomes without human-modeled paths for every permutation of work.

We’ve moved beyond single-thread dragons to intelligent meshes of agents that collaborate to solve ambiguous problems. This isn’t a minor upgrade — it’s an architectural rewrite of how work gets done. 

Legacy Automation Architecture Is Becoming the New Mainframe

Here’s the real market truth:

Legacy process automation, even with AI badges slapped on workflows, is rapidly approaching the same fate as outdated computing paradigms — valuable in a historical context, obsolete in a strategic context.

Public markets already charge legacy automation names discount multiples compared to true AI-driven growth platforms. Venture capital is tightening its focus on workflow stitchers while expanding its backing of agentic computation pioneers. And boards are starting to ask deeper questions about outcome economics instead of demo bells and whistles.

If legacy automation is mainframe-era tools rebranded for the cloud, then agentic systems are the next computational substrate for business execution.

That’s the disruption that will wipe out the old market cap — and unlock orders-of-magnitude value for enterprises that embrace the new paradigm.

The automation story isn’t over — it’s rewritten.

The post Hyper-Automation Is Over. Agentic AI Is What Comes Next. appeared first on MartechView.

Publicis Groupe has agreed to acquire LiveRamp, the global data collaboration platform, for a total enterprise value of $2.167 billion in an all-cash transaction — its most significant strategic bet since the $4.4 billion acquisition of Epsilon in 2019 and a clear signal of where the world’s third-largest advertising group believes the next competitive frontier lies.

The deal, priced at $38.50 per share and representing a 29.8 percent premium to LiveRamp’s closing price on May 15, has been unanimously approved by the boards of both companies. It is expected to close before the end of 2026, subject to regulatory approval and LiveRamp shareholder sign-off.

What LiveRamp Is

LiveRamp is not a conventional advertising technology company. It is infrastructure — a data collaboration platform that enables organizations to unify, manage, and activate data across the digital ecosystem without exposing the underlying sensitive information that makes that data valuable.

Its network spans more than 25,000 publisher domains and over 500 technology and data partners across 14 markets. Thousands of brands, retailers, media platforms, and data providers use its clean room technology to collaborate on data they could not share through conventional means. With 1,300 employees and a business anchored in highly recurring revenue, LiveRamp has delivered a compound annual revenue growth rate of 13 percent over the past five years.

Why Publicis Is Buying It

The acquisition is a direct response to what Publicis identifies as the defining constraint on enterprise AI adoption: most companies lack the right data to make their AI systems genuinely effective. According to figures cited by Publicis, 93 percent of companies do not currently have the data infrastructure required for AI success.

LiveRamp addresses that gap through what Publicis calls data co-creation — the process by which companies combine multiple high-value data sources across partners in secure, governed environments to generate new proprietary data assets that no single organization could build alone.

Combined with Epsilon’s identity resolution capabilities, LiveRamp’s collaborative infrastructure is designed to enable clients to build AI agents that are more capable, more differentiated, and more commercially effective than anything a single organization’s data could support independently.

“After acquiring Epsilon in the name of personalization at scale and enabling our clients to take back control of their data from the walled gardens,” said Arthur Sadoun, Chairman and Chief Executive of Publicis Groupe, “once again we are looking ahead to what’s next. By building the future of data co-creation, we’re empowering our clients to generate new, exclusive and proprietary data — to build the smartest, most differentiated AI agents on top of the leading large language models.”

Also Read: Peak Martech? The Landscape Has Plateaued, but the Real Story Lies Beneath

The Use Cases

Publicis has been explicit about what data co-creation enables in practice, offering three illustrative examples.

A bank could build a wealth management agent that draws on tokenized customer data from retail banking, credit cards, and wealth management, combined with partner data from merchants, payment networks, and travel providers — without exposing underlying customer records. The result is an agent capable of cross-selling across multiple business lines with far greater precision than any single dataset would allow.

A retailer could connect loyalty, in-store, and retail media data with partner signals to measure the incremental value of each customer touchpoint and build new, proprietary shopper journeys — turning retail media from a cost center into a measurable growth driver.

A pharmaceutical company could build a therapeutic area optimization agent that draws on clinical, commercial, supply chain, and de-identified patient data simultaneously — enabling more efficient field-force deployment and better product lifecycle management across an entire portfolio.

The Strategic Architecture

The LiveRamp acquisition completes what Publicis describes as an end-to-end capability stack for agentic business transformation. Publicis Sapient provides the technology modernization layer that makes enterprise infrastructure AI-ready. Epsilon’s identity resolution connects agents to real people, behaviors, and deterministic transactions. LiveRamp enables secure data collaboration across partners to generate the co-created data that fuels smarter agents. Marcel, Publicis’s internal agentic platform, activates that data across enterprise functions.

Each component addresses a different layer of the same problem. Together, Publicis is arguing, they constitute a capability set that no competitor can currently match end to end.

Financial Implications

The transaction is expected to be earnings-accretive from the first year of consolidation, excluding transaction-related costs. Publicis is funding the acquisition with cash on hand and debt, maintaining financial leverage within existing investment-grade rating parameters, with full deleveraging expected within two years of closing.

The deal also allows Publicis to raise its medium-term financial guidance. The Groupe now targets net revenue growth of 7 to 8 percent and headline earnings per share growth of 8 to 10 percent at constant currency in 2027 and 2028 — up from prior targets of 6 to 7 percent and 7 to 9 percent respectively.

Also Read: Push Notifications Are Broken. Here Is What Comes After Them

What Changes — and What Does Not

LiveRamp will continue to operate as a neutral, interoperable platform following the acquisition. Chief Executive Scott Howe will remain in his role, reporting directly to Sadoun. The company’s data will continue to be protected in accordance with existing contractual commitments. Pricing and commercial practices will remain unchanged outside the normal course of business.

The commitment to neutrality is not incidental. LiveRamp’s value depends on the trust of the 500-plus technology and data partners and 25,000-plus publishers in its network — a trust that would erode quickly if the platform were perceived to operate in the exclusive interest of its new parent company.

“Our customers and partners have always been our North Star,” said Howe. “By joining forces with Publicis, we will have greater resources and flexibility to scale our business, continue innovating our platform, and help them unlock even greater value from their data.”

The transaction is expected to close before year-end 2026.

The post Publicis Acquires LiveRamp for $2.2B in AI Data Push appeared first on MartechView.

Sendbird, the AI customer experience company powering conversations for more than 300 million people every month, has launched Agent Steward, an AI agent designed to take full, end-to-end ownership of complex customer issues — and to hold that ownership across systems, channels, and time without requiring human coordination at every step.

The launch is accompanied by two platform updates — Trust OS 2.0 and Voice 2.0 — that together represent a significant expansion of Sendbird’s Delight.ai platform, pushing it from reactive AI support toward autonomous improvement and proactive customer engagement.

The Problem With Reactive AI

The case for Agent Steward begins with a structural critique of how enterprise AI currently operates. Despite rapid adoption, AI agents remain largely channel-specific and reactive — waiting for a prompt, handling the immediate query, and handing off to human teams when anything more complex arises. Mistakes are identified and corrected after the fact, if at all.

Consumer expectations reflect the gap. According to new data from Sendbird, 57 percent of consumers say the ability for AI to correct its own mistakes and reverse decisions would increase their trust. Nearly two-thirds — 59 percent — say the ability to stop or override an AI agent is very important to them. The signal is clear: what customers want from AI is not just competence. It is accountability.

Also Read: The CIO Who Says Governance Can Actually Speed Up AI

Agent Steward: One Owner, Start to Finish

Agent Steward is built specifically for long-horizon, multi-step workflows — the category of customer issues that current AI systems handle poorly. Rather than routing a problem across multiple agents and teams, Agent Steward acts as a single point of accountability from first contact through to resolution, coordinating across backend systems, chat, SMS, and email while escalating to a human only when genuine judgment is required.

The practical implications are significant. A customer who receives a damaged product from a third-party logistics partner currently faces a resolution process that can span days, involve multiple handoffs, and result in a chargeback. With Agent Steward, the issue is identified proactively, owned by a single agent throughout, and resolved within hours — with human involvement triggered only at decision points that require it.

“Most AI systems are effective at handling routine queries, but break down when it comes to complex, multi-step issues,” said John Kim, co-founder and chief executive of Sendbird. “The goal isn’t to replace human agents, but to elevate them — freeing them to focus on judgment and exceptions while AI owns coordination, follow-through, and continuous improvement.”

Trust OS 2.0: AI That Fixes Itself

The launch of Trust OS 2.0 introduces what Sendbird calls Zero-Touch Improvement — a capability that enables AI systems to identify, diagnose, and correct their own errors in real time, without requiring manual intervention from engineering or operations teams.

The contrast with traditional AI management is stark. Under conventional frameworks, identifying a model error, rewriting the relevant rules, testing the fix, and deploying it can take days or weeks. Zero-Touch Improvement compresses that cycle to real time, shifting AI governance from a supervisory burden to an autonomous function.

Also Read: Is Your CRM Making Your Customer Service Worse?

Voice 2.0: Reaching Customers Before They Ask

Voice 2.0 extends the platform’s capabilities to proactive outbound engagement, allowing companies to initiate contact with customers when a potential issue is identified, rather than after it has already created a problem. The capability supports more than 100 languages, enabling global deployment without localization overhead.

The use cases are immediate and practical. An airline can notify a traveler of a disruption before they reach the airport. A financial services company can send a payment reminder before a deadline is missed. The shift from reactive support to anticipatory engagement represents a meaningful change in how AI interacts with customers across the full relationship lifecycle.

Agent Steward, Trust OS 2.0, and Voice 2.0 are available now as part of the Delight.ai platform, following their debut at Sendbird’s annual customer experience event, Delight Spark.

The post Sendbird Launches AI Agent That Owns Issues End to End appeared first on MartechView.

For most of the past decade and a half, the annual marketing technology landscape had one reliable characteristic: it grew. Sometimes by a lot. Sometimes by a merely extraordinary amount. Even in the years when analysts confidently predicted consolidation was finally upon us, the landscape found another gear. This year, it did not.

The State of Martech 2026, which debuted at MartechDay on May 5 by Scott Brinker of chiefmartec and Frans Riemersma of MartechTribe, puts the total number of marketing technology products at 15,505 — up just 121 from the 15,384 counted last year. That is growth of 0.79%, rounding to effectively zero. After a run of more than 10,000% expansion since 2011, when the landscape counted just 150 products, the market appears to have hit a ceiling — or at least a plateau.

But flat, as the report and keynote make abundantly clear, is perhaps the most misleading word one could use.

A Market Metabolizing, Not Stagnating

Beneath that near-zero headline number, the market is moving with real intensity. In the past 12 months, 1,488 new products were added to the landscape while 1,367 were removed. The volume of new entrants dropped 40% year on year — down from 2,489 in 2025 — while the removal rate climbed 13%. For the first time in the post-pandemic era, additions and removals are nearly canceling each other out.

Riemersma’s framing at MartechDay was direct: “Peak Martech is a myth. Martech is entering its Darwin phase. The martech landscape is renewing. Value is growing.” The era of accumulating tools, both argued, is giving way to an era of replacing them. At the core of that transition is a structural change in where value actually lives: SaaS platforms are no longer the primary source of differentiation. They are becoming infrastructure — systems of record, workflow engines, and integration layers. The real value is moving on top of that foundation. AI is becoming the value layer.

The companies exiting the market tell their own story. More than half of this year’s removals — 51.7% — came from the 2010–2019 wave of software-as-a-service startups, the first great generation of martech builders. The exits are concentrated among smaller firms: 41.2% had between one and ten employees; 38.7% had between 11 and 50. By revenue, the $1 million to $10 million band accounts for 45.5% of removed products — companies that found enough early traction to survive past zero revenue, but not enough to build a truly defensible position, caught between incumbents bundling AI features from above and AI-native startups attacking from below.

The Content Marketing Bust

Perhaps the single most striking data point concerns content marketing tools. When generative AI went mainstream in 2023, content marketing was one of the first categories to feel the full force of the wave, nearly doubling in two years from 575 tools to 1,102. In 2026, it leads all subcategories in a less coveted ranking: the highest net product removal of any category, at minus-37, with 176 removed and only 139 added.

Three forces converged. The major AI laboratories absorbed the core functionality; incumbent platforms such as Adobe, HubSpot, and Salesforce rapidly embedded generative AI into existing workflows; and many first-wave tools solved the problem of generating content fast without solving the harder problem of generating content that actually works. The report describes this as a natural selection event: not the end of AI-powered content technology, but the clearing out of an undifferentiated first generation in favor of a more mature second.

The Stack Is Stratifying, Not Consolidating

One of the most significant conclusions from the MartechDay keynote — drawing on a survey of 208 marketing and marketing operations leaders across 70 specific AI use cases — is that the long-running debate between platform consolidation and best-of-breed diversification has a 2026 answer: neither. Instead, the stack is stratifying into layers with different competitive physics. 

AI-native tools are largely winning creation — copy ideation, pitch decks, visual production, competitive intelligence — tasks where the primary input is a prompt and model quality is the product. Incumbent SaaS platforms such as HubSpot and Salesforce are largely holding on to orchestration: lead scoring and routing, pipeline management, and channel delivery. These systems increasingly serve as infrastructure for other commercial and custom AI agents.

The survey also revealed a striking divergence between B2B and B2C adoption patterns. Conventional wisdom holds that B2C leads technology adoption. On AI, the data inverts that pattern: B2B shows broader adoption across more use cases, with consistently lower non-adoption rates — likely because B2B teams are chronically understaffed relative to their content and operational demands, and a decade of CRM, MAP, CDP, and revenue intelligence investment had already built natural docking stations for AI capabilities. When B2C does adopt, it builds deeper: the customer-facing AI output is the brand experience, and the differentiation lives in the final 20% — brand voice calibration, proprietary guardrails, custom data integration — that off-the-shelf tools cannot provide.

The AI Agent Paradox

A central tension running through the MartechDay findings is the gap between AI enthusiasm and AI deployment, as researchers described it. Some 90.3% of marketing organizations now use AI agents in some capacity, yet only 23.3% have deployed them in full production. The rest are piloting, experimenting, or running agents in narrow workflows with a human approving every output. The report identifies this as the “Trust Wall”: currently, 80.6% of marketing organizations refuse to let AI agents operate autonomously, requiring a human in the loop for every final decision.

Governance is moving in the right direction — 73% of respondents now report having a formal generative AI policy, up from 52% in 2024 — but the gap between having a policy and having the infrastructure to enforce it remains wide.

Where Growth Is Actually Happening

If content marketing is the cautionary tale, content management systems and e-commerce platforms are the 2026 growth story. CMS and web experience management grew 21.4%, jumping from 504 to 612 products. E-commerce platforms grew 19.9%, from 547 to 656. These are not new categories. They are being reshaped. CMS is evolving into a machine-readable infrastructure for AI agents. E-commerce is adapting to AI-driven discovery. iPaaS is becoming the orchestration layer that connects everything. Growth is happening where AI changes the job to be done.

The explanation lies in a fundamental shift in who—or what—digital properties are built for. For two decades, marketing teams designed experiences primarily for human visitors and search engine crawlers. That audience now includes AI search assistants, agentic browsers, shopping agents, and procurement systems that arrive not to browse but to extract, evaluate, and act. Other fast-growing subcategories follow the same logic: mobile and web analytics grew 11.3%, call analytics 8.9%, data integration 8.0%, and marketing automation 5.9% — the last a sign that AI is reinventing what campaign orchestration can look like, attracting builders who see agentic marketing automation as a meaningful step beyond rule-based systems.

SEO Becomes AEO — but Visibility Is Shrinking

Search engine optimization, widely eulogized as AI assistants swallowed the top of the funnel, is in fact metamorphosing rather than dying. The SEO and answer engine optimization subcategory posted a net positive result this year — 44 added, 38 removed — and has grown for three consecutive years. The market is reflecting a shift in the underlying discipline: from making brands findable by search crawlers to making them findable, credible, and actionable across AI search assistants, answer engines, and agentic browsers. The challenge, the report notes, is that the tools are improving while the marketer’s visibility is shrinking — when a customer consults an AI assistant about which product to buy, that conversation is entirely invisible to conventional tracking.

The Transformation Beneath the Numbers

What ties these shifts together is a structural transformation of marketing itself. As Brinker argued in the lead-up to MartechDay: “AI doesn’t eliminate constraints. It moves them. When content becomes abundant, the bottleneck shifts to relevance. When integrations get easier, the bottleneck shifts to orchestration.” The organizations pulling ahead are those that have recognized where the new bottleneck sits and invested in context engineering, governance, and strategic coherence — rather than continuing to optimize against constraints that AI has already dissolved.

The best stacks are not the most feature-rich. They are the most aligned — focused on a small number of high-impact use cases where SaaS enables, and AI amplifies. Integration is no longer just technical. It is a strategic asset.

Whether 2026 marks the peak of martech or simply a pause before the next expansion remains genuinely uncertain. Brinker and Riemersma’s own position is the latter. The cost to build keeps falling, AI keeps opening new niches, and the minimum viable scale for a sustainable martech business keeps shrinking. The landscape is metabolizing — not dying. But the shape of whatever emerges from the chrysalis will bear little resemblance to what went in.

The State of Martech 2026 was debuted by Scott Brinker and Frans Riemersma at MartechDay on May 5, 2026, and is available free at chiefmartec.com.

The post Peak Martech? The Landscape Has Plateaued, but the Real Story Lies Beneath appeared first on MartechView.