Salesforce Agentforce Commerce and Connected Journeys: A Marketing Ops Playbook
What Salesforce actually shipped at Connections 2026
At Salesforce Connections 2026 in Chicago on April 15-16, the Agentforce roadmap pivoted hard toward commerce. Two products carry the announcement: Agentforce Commerce and Connected Journeys. The 2025 Agentforce could reply to a support ticket. The 2026 Agentforce initiates conversations, captures payment inline, and decides on its own whether to keep the thread open.
Agentforce Commerce runs two-way conversations across email, SMS, and WhatsApp simultaneously under a single agent context. The agent can drop a payment link inside an email thread, then switch to WhatsApp when the shopper replies there. For ops teams who have been duct-taping an ESP, an SMS platform, and a WhatsApp BSP together, the consolidation is the real story. The channel-agnostic memory is what lets Salesforce claim agent autonomy with a straight face.
Connected Journeys is the rewrite of SFMC Journey Builder. Each step used to be either a template send or a decision split. Now each step can be a human-authored send or an agent-executed conversation, and the agent can branch the journey based on reply sentiment. That shifts the canvas from a static decision tree to a hybrid execution surface, which maps much more cleanly to how lifecycle teams actually think.
The Qualified acquisition, announced in early April, folds Qualified’s inbound site conversation agents into Agentforce. Site conversations and outbound re-engagement now share a lead record, which used to require two vendors and a messy handoff.
Where Agentforce Commerce fits in your stack
Agents are not rule engines. They decide against unified profiles in Data Cloud, which means Data Cloud is a hard prerequisite. If SFMC currently runs standalone without the CDP, your first line item is Data Cloud enablement plus data modeling, not the agent license.
For product and inventory, the native source is Commerce Cloud. Shopify and Commercetools customers get an official connector that syncs data one way into Data Cloud. Two-way writeback, where the agent modifies an order on the shopper’s behalf, stays Commerce Cloud only in the 2026 release. Most DTC brands on Shopify Plus should scope agent responsibilities to read, converse, and collect payment, and leave order mutations to the service desk.
SMS and WhatsApp ride on Twilio or MessageBird. If Twilio is already in place, wiring the channel takes about half a day. WhatsApp template approvals and opt-in compliance remain your responsibility. The agent generates content but does not own BSP governance.
Inline payment is the piece most teams underestimate. The agent can embed a single-use checkout link inside an email or SMS thread, routed through Salesforce Payments, Stripe, or Adyen. Cutting the return-to-site step out of the recovery flow is where a lot of the reported cart-abandon lift comes from.
Three scenarios that earn the license cost
The first is cart abandonment. Traditional three-email sequences hit single-digit open rates in 2026. Agentforce sequences across channels with reply-aware branching: T+30 minutes email, T+6 hours SMS, T+24 hours WhatsApp, each personalized to the prior reply. If the shopper replies “not sure about sizing,” the follow-up drops a size guide instead of another discount nudge. Salesforce pilot data reports a 35 percent lift in recovered revenue against the legacy three-email pattern, and a 22 percent CSAT increase attributed to personalized cross-channel threading.
The second is VIP re-engagement. When a Tier-1 customer lapses past 60 days, the agent composes a bundle proposal from purchase history, Data Cloud propensity scores, and live Commerce Cloud inventory, then sends by email with an SMS follow-up 48 hours later if unopened. This replaces the choice between a manual CRM motion and a rule-based recommender that never quite knew the margin floor.
The third is B2B lead nurturing, which is where the Qualified acquisition pays off. Inbound site agents and outbound email agents share one lead profile. When the MQL threshold is crossed, the handoff is silent because there is no cross-system replay. Enterprise-focused B2B teams with long consideration cycles are the natural buyer.
Anatomy of a Connected Journeys canvas
Here is a production-shape cart-abandon canvas as it sits in Connected Journeys:
| Step | Type | Trigger | Content | Branch |
|---|---|---|---|---|
| 1 | Trigger | Cart idle 15 min | Enter journey | - |
| 2 | Agent Step | T+30 min | Email, inline payment link | Read reply sentiment |
| 3 | Decision | On reply | - | Positive to 5, negative to 4, none to 6 |
| 4 | Agent Step | Immediate | Handle objection, propose alternative SKU | Read reply |
| 5 | Human Send | Immediate | Ops-authored thank-you plus discount code | End |
| 6 | Agent Step | T+6 h | SMS follow-up | Read reply |
| 7 | Agent Step | T+24 h | WhatsApp last touch | End |
Step 5 is a human-authored send, and that is the point. Connected Journeys does not require every step to be agent-driven. Lifecycle teams retain editorial control where it matters most: discount mechanics, brand voice moments, compliance-sensitive content. Agents own the steps that need dialogue and judgment. The hybrid pattern is the only production-safe mode in 2026. Pure agent journeys carry too much compliance and brand risk for most buyers right now.
Each Agent Step exposes guardrails in Agentforce Studio: discount ceiling per conversation, max messages per thread, SKU exclusion lists, topic allowlists. The journey canvas invokes the configured agent but does not redefine its permissions. Keep the guardrails centralized so that one audit covers every journey that references the agent.
Pricing, competitors, and a decision framework
Pricing shifted from roughly $0.10 per action in 2025 to about $2 per conversation-action in 2026. A conversation-action is one complete agent-initiated or agent-response interaction. For a mid-market brand running 3,000 to 7,500 conversation-actions per month, that lands at $5-15k in Agentforce spend alone, before Data Cloud, Twilio, and WhatsApp BSP template fees.
Competitive landscape, honestly framed:
| Dimension | Salesforce Agentforce | Klaviyo Composer | Attentive AI Pro | HubSpot Breeze |
|---|---|---|---|---|
| Cross-channel orchestration | Unified agent across email, SMS, WhatsApp | Email primary, SMS secondary | SMS primary, email secondary | Email primary |
| Agent autonomy | High, can converse and capture payment | Medium, rules-driven content | Medium, strong in SMS | Low, recommendation-oriented |
| CDP requirement | Data Cloud required | Built-in light CDP | Built-in | Built-in |
| Inbound site agent | Yes, via Qualified | No | No | Yes |
| Fit | Mid to large, already on Salesforce | SMB to mid DTC | Mid to large DTC | Mid B2B |
Agentforce wins on cross-channel agent autonomy. It loses when a brand is not already on the Salesforce stack, because the switching cost dwarfs the incremental lift from the agent itself. Do not treat Agentforce as a reason to migrate your CDP or ESP.
The pragmatic filter. SFMC is already the primary ESP, Data Cloud is already in production, monthly GMV clears roughly five million USD, and SMS plus WhatsApp are core to your international mix. If all four hold, the ROI math works. If any fail, the 2026 move is to layer Klaviyo Composer or Attentive AI onto the existing stack and revisit Agentforce once consolidation is no longer the blocker.
Related Articles
Attentive Thread 2026: What the Agentic AI and RCS Upgrades Mean Before BFCM
At Thread 2026 on May 26, Attentive revealed that nearly half its customers are already running AI agents. Before BFCM, Brand Voice 2.0, AI Email Template Generation, RCS Visibility AI, and AI Grow 2.0 are all shipping. Here is what each one does and how to prepare.
Intercom Fin for E-Commerce: How AI Turns Support Into a Sales Channel
Launched May 8, 2026, Fin for Ecommerce handles pre-purchase discovery and post-purchase support in a single conversation thread. Shopify integration takes minutes. Early numbers show 10% conversation-to-order conversion and 20% higher AOV on Fin-assisted orders.