Email Is Not Dead. It Is Becoming Programmable.

Martech moves at a pace that is genuinely hard to keep up with. Every week something gets renamed, ships an "agent," or swallows a category that used to be its own product. I run a marketing function for a living and I still catch myself a release or two behind.

So lately I have been doing my own digging — less to chase the announcements, more to work out what actually changed underneath them. This is where I landed on email.

The email programs that feel strongest usually send less than the ones that feel noisy. Not always. Often enough to be a pattern.

The difference is rarely creative. It is architecture. Sends tied to signals. Branches someone can explain. Agent-assisted drafts that still pass a human who knows the account.

That is the direction: less batch-and-blast, more programmable customer logic. The channel is not dying. It is becoming software.

Every serious platform is being rebuilt around that idea. Eloqua and Marketo hold the deep end, HubSpot owns the middle, and Salesforce — still Pardot to half the teams running it — is rewiring its whole marketing suite around a shared data layer. The names on the shortlist have not changed much. What sits underneath them has.

The platforms most teams still start with

Oracle Eloqua remains the deep end for B2B: complex nurture, account-level logic, integrations that survived mergers and reorgs. Eloqua's world is campaigns as programs, not messages. The teams that thrive on it treat operations as a discipline, not a side job for whoever learned the tool fastest.

Adobe Marketo Engage sits in similar territory with a different center of gravity: revenue alignment, scoring, sales handoffs marketing can stand behind. Marketo's AI features (suggested content, predictive nudges) only matter when the data underneath is honest. That part is harder than the demo.

HubSpot changed what "good enough" email feels like. Visual automation. CRM-native context. Reporting a VP can skim without a translator. Its agentic direction — AI assistants inside workflows, content generation tied to CRM objects, smarter send-time and audience tools — points at the real shift: the bottleneck is not template design anymore. It is decision speed.

Those three run the day-to-day for most enterprise teams. Salesforce belongs on the same shortlist — but it earns its own section, because it is the one incumbent rebuilding its entire foundation right now.

Salesforce, Pardot, and the migration that is not one

A lot of B2B teams still say Pardot. Salesforce calls it Marketing Cloud Account Engagement (MCAE) now — same product, clearer attachment to the CRM. The name change matters when you are reading release notes, not when you are drafting copy.

One step up from Pardot sits Marketing Cloud Engagement — the platform most teams still call ExactTarget. MCAE handles lead-centric B2B nurture; Engagement is the enterprise cannon, built for B2C-scale email, SMS, push, and cross-channel orchestration through Email Studio, Journey Builder, and Automation Studio. Teams that outgrew Pardot's contact-and-lead model have run their high-volume programs here for years, and it is not being retired: Salesforce's own line is that Engagement is where you operate today and Next is where marketing goes tomorrow.

The bigger shift sits above both. Salesforce has rebranded its marketing suite as Agentforce Marketing (the umbrella that used to be Marketing Cloud). The next-generation product inside it is Marketing Cloud Next: built natively on the Salesforce platform with Data Cloud and Agentforce at the center. It is the attempt to rebuild a decade of acquisitions — ExactTarget, Pardot, Datorama, Evergage — as one native application instead of the stitched-together stack most teams still run today.

For MCAE customers, the important part is you do not have to rip and replace. At Connections '25, Salesforce announced that Growth, Plus, Advanced, and Premium editions get Account Engagement+ permission set licenses — access to Next-gen features inside your existing instance. Run both apps side by side for the use cases that make sense.

What ships through Account Engagement+ today:

• Agentforce Campaign Creation — one prompt generates a campaign brief, email copy, and assets. Marketers review, upvote or downvote outputs, and edit before anything schedules.
• Segment creation via natural language — audiences pulled from Data Cloud, including data outside the CRM.
• SMS in journey builds — multichannel paths without a third-party SMS vendor. Plus, Advanced, and Premium editions with Digital Engagement can turn one-way SMS into two-way conversations with service agents.
• Cross-Object Merge Fields — personalization from related records (product purchased, webinar attended, case status), not just contact-level fields. A gap MCAE customers have wanted closed for years.
• Rule-Based Dynamic Content — subject lines, preheaders, and body blocks driven by Data Cloud attributes.

The catch is architectural, not licensing. Data Cloud is the prerequisite: Sales Cloud or Service Cloud Enterprise (or higher), Data Cloud provisioned, and the Account Engagement: Data Cloud connector wired. The PSLs are included in supported editions; running Agentforce, SMS, and advanced personalization consumes credits for AI requests, messaging, and Data Cloud segmentation.

What is coming has mostly arrived. Adaptive Websites — pages that reshape themselves from customer interaction across marketing, commerce, sales, and service — shipped at the end of 2025. The Summer '26 release, rolling out through July 2026, is where the agentic layer gets concrete for MCAE-adjacent teams:

• Conversational Email — Agentforce reads replies to a broadcast send and answers in-thread. Someone asks about sizing, shipping, or warranty and gets a contextual response without a human touching it; high-intent replies get captured instead of dying in an unmonitored inbox.
• Consent mapping between Engagement and Next — subscription and publication-list consent syncs automatically when a subscriber changes it. Email mapping lands by July 13, 2026; SMS and WhatsApp by August 17.
• Distributed Marketing Agent — sales users send marketing-approved templates with engagement insight attached, personalized per recipient, without going off-brand.
• Journey Decision Agent — picks the next step in a journey from real-time behavioral and profile data, not a static branch someone drew a year ago.
• Business Units in Marketing Cloud Next — up to 50 per org, each mapped to its own data space, so campaigns, audiences, and reporting stay isolated where they should be.

B2B nurture stays rules-based at its core. The upgrade is agents that propose campaigns, read richer CRM data into personalization, answer replies, and extend the channel mix without leaving Salesforce.

If you are on MCAE, the practical unlock is whether Data Cloud is provisioned yet — because that is what gates Agentforce, cross-object personalization, and the Next-gen features Salesforce is shipping alongside Account Engagement.

The underdogs doing the interesting work

The four incumbents own the enterprise center. One tier down, a smaller set of tools is doing the more interesting work — faster, partly because they are not carrying twenty years of technical debt:

• Braze for cross-channel orchestration: email as one surface in a real-time system, not the whole system.
• Iterable and Customer.io for product-led teams that need event-driven messaging without an integration army.
• Klaviyo where commerce signals (browse, cart, purchase, repeat) drive the program instead of the calendar.
• ActiveCampaign and Brevo for mid-market teams that want automation depth without procurement theater.
• Mailchimp at the long tail. Still relevant, with AI-assisted content raising the floor for teams that will never buy Eloqua.

The pattern: underdogs win on time-to-first-intelligent-send, not feature checklists. Which of them still matters in five years is an open question. Right now they are useful pressure on the incumbents.

Agentic email: what is real vs. slideware

Email is a natural home for agents. Repetitive work, structured rules, measurable outcomes. The version that holds up is not "AI writes our newsletter." It is closer to:

• A signal agent that watches product usage, web behavior, and support themes, then proposes triggered paths with sample copy and an estimated lift.
• A QA agent that checks links, tokens, compliance language, and dark-mode rendering before anything schedules.
• A performance agent that reviews cohort results, flags fatigued segments, and drafts test hypotheses. Humans still pick what ships.

HubSpot, Eloqua, Marketo, and Salesforce are all telling variations of this story. The gap between vendors is not "does AI exist." It is how cleanly the AI connects to the data that should drive a send.

We dreamed about this back when dynamic content was a checkbox in a demo. The difference now is agents operating across CRM, product analytics, and support, not only inside the ESP's walled garden.

What separates programs that compound from programs that spam

1. Signals before segments. Start from what changed in the customer's world, not what month it is.
2. Fewer journeys, clearer ownership. A hundred automations nobody remembers is debt, not sophistication.
3. Humans at the relationship moments. Agents draft and route. People send what requires trust.
4. Deliverability treated like brand protection. Reputation, frequency caps, list hygiene. Not "ops details" you defer forever.

Where this lands

Email is not going anywhere as an owned channel. B2B and a lot of B2C still live here. What is changing is the comparison set: recipients now measure your nurture against product-native lifecycle programs, not just your competitor's newsletter.

HubSpot, Eloqua, Marketo, and Salesforce will keep consolidating the enterprise center. Braze, Iterable, Klaviyo, and the rest will keep pushing them to move faster. Agentic AI sits in the middle, removing friction between insight and send.

The winners will not have the most templates. They will be the teams whose email program behaves like software: versioned, tested, observable, and honest about what each message is for.

That is where my own digging landed this month. I run these programs in the real world, not just in the release notes — so when the next update changes the answer, I will come back and say so.