Scout Academy Newsletter: AI Agents & Sales Ops

Your roundup of what actually matters in AI-powered sales, March 2026.

---

February was the month sales tools finally started talking to each other. Gong and Outreach both shipped Model Context Protocol (MCP) support in the same week, and Salesforce reported $800 million in Agentforce revenue. Meanwhile, every major vendor quietly added the governance controls that signal one important thing: AI agents in sales are no longer experimental. They're running in production, closing deals, and finally getting the safeguards they need.

The bigger picture is coming into focus. Goldman Sachs, Gartner, and CIO are all converging on the same thesis: 2026 will be the year agents stop being demos and start being infrastructure. But the gap between hype and production is wider than most people realize.

Here's what you should know.

---

🔗 Link Roundup

The Big Picture: Where AI Agents Are Actually Headed

Before we get into sales-specific updates, here are four articles worth reading that help to frame the broader landscape.

What To Expect From AI in 2026 — Goldman Sachs
Goldman's CIO Marco Argenti makes seven predictions for 2026, and the most interesting one isn't about models getting larger. It's about context becoming the new frontier. His argument is that we've already essentially trained models on the entirety of the internet, so the next leap is giving them better memory of what you specifically need. He also predicts the rise of an "agent-as-a-service" economy where companies bill by tokens consumed instead of hours worked. That's a pricing model shift that will reshape how sales teams buy and deploy AI tools. → https://www.goldmansachs.com/insights/articles/what-to-expect-from-ai-in-2026-personal-agents-mega-alliances Our take: Argenti's "context is the new frontier" framing is the most useful mental model for anyone building agents right now. The models are good enough, but the real bottleneck is providing them with the right context at the right time. If you're building in Scout, this is exactly what your databases and collections solve: they give your agents a personal library of context that generic models don't have.
AI Agent Trends 2026: From Chatbots to Autonomous Business Ecosystems — Gapps Group
This article provides a practical breakdown of five trends shaping agentic AI this year. The standout insight is the idea that every employee will become a "human supervisor," managing a team of specialized agents. Gapps frames this as a shift from instruction-based computing ("do this step by step") to intent-based computing ("here's what I want, figure it out"). Its also highlights MCP as the connective tissue that makes multi-agent workflows possible across tools such as BigQuery, Cloud SQL, and CRM systems. → https://www.gappsgroup.com/blog/ai-agent-trends-2026-from-chatbots-to-autonomous-business-ecosystems Our take: The concept of a human supervisor is entirely accurate, and it's already visible in sales. Your best AEs aren't manually researching accounts anymore. They're reviewing what their research agent found and deciding what to do with it. The teams that adopt this manager-of-agents workflow will outperform those that are still debating whether to adopt AI at all. The Gapps article is also a helpful piece if you're explaining agentic AI to someone on your team who still thinks AI means chatbots.
2026: The Year of Multi-Agent Systems — RT Insights
If 2025 was the year of the single AI agent, 2026 will be the year they start working together. RT Insights makes the case that multi-agent orchestration (where specialized agents collaborate on complex processes) is the next breakthrough. The article discusses how organizations are moving from isolated task automation to coordinated, end-to-end workflows where agents seamlessly hand off tasks to each other like a well-run team. → https://www.rtinsights.com/if-2025-was-the-year-of-ai-agents-2026-will-be-the-year-of-multi-agent-systems/ Our take: This tracks with what we're seeing in Scout. The most effective setups aren't those where one agent does everything. Instead, they're the ones with a research agent feeding information to a writing agent, which then hands over to an outreach agent — each agent is specialized, each doing one thing well. The challenge isn't building multi-agent systems. It's designing the handoffs. When Agent A finishes, what exactly does Agent B need to pick up? This is a workflow design problem worth solving, and it relies more on clear thinking than technical complexity.
Agentic AI in 2026: More Mixed Than Mainstream — CIO
This report offers the reality check the industry needs. CIO highlights that Gartner expects 40% of enterprise apps to include AI agents by the end of 2026 (up from less than 5% in 2025). But, importantly, the article suggests that only 23% of organizations experimenting with agents are scaling them effectively. The rest are stuck in a never-ending cycle of pilots. The article also flags that up to 40% of Global 2000 job roles will involve working alongside AI agents this year, which means the workforce impact is real whether your organization is ready or not. → https://www.cio.com/article/4107315/agentic-ai-in-2026-more-mixed-than-mainstream.html Our take: This is the most honest piece in the roundup. The 23% scaling figure aligns with Deloitte's finding that only 11% of AI agents make it to production. The pattern is clear: most teams can build a demo, but few can successfully ship a production agent. The difference isn't the AI itself. It's governance, data quality, and clear workflow design. If you're in the 77% that's stuck in pilots, start smaller with one agent, one workflow, and one measurable outcome. You can scale after you know it works.

---

Your Sales Tools Now Speak the Same Language

The biggest structural change this month isn't a product launch. It's a protocol.

Gong and Outreach both adopted MCP in February, which is an open standard (originally from Anthropic) that lets AI tools share context without custom integrations. Your conversation intelligence can now directly inform your engagement platform, with no middleware required.

This matters because the sales stack of 2026 isn't about finding one platform that does everything. It's about four or five intelligence tools that can natively share context.

Gong Launches "Mission Andromeda" — Gong
Gong's biggest release ever bundles AI-powered enablement, a conversational assistant for querying call data, and MCP support for interoperability with Dynamics 365, Salesforce's Agentforce, and HubSpot. The AI call reviewer identifies skill gaps automatically, while the AI trainer runs role-play simulations, connecting training directly to revenue outcomes. → https://www.gong.io/platform/mission-andromeda
Outreach Ships MCP Server and AI Agents — Outreach
Outreach can now push insights, activity signals, and next-best actions into Salesforce's Agentforce, Microsoft Copilot, and Anthropic's Claude via MCP. It also launched a meeting prep agent (in beta) that auto-generates briefs with talking points and account context, plus a research agent that refreshes account data on daily, weekly, or monthly cadences. → https://www.outreach.io/resources/blog/february-2026-product-release
What Is Model Context Protocol? — Outreach
Outreach provides a clear explainer on MCP for sales teams who want to understand the standard without reading the technical spec. It's a good primer if you're evaluating how your tools will connect in 2026. → https://www.outreach.io/resources/blog/what-is-model-context-protocol-mcp

Salesforce's Agentforce: The Numbers Are Real

Salesforce reported its Q4 FY26 earnings, showing Agentforce at $800M annual recurring revenue (ARR), marking a 169% increase year over year and a 48% rise quarter over quarter. Salesforce closed 29,000 Agentforce deals in Q4 alone (up 50% from Q3). Combined Agentforce and Data Cloud ARR now exceeds $2.9 billion.

But here are the numbers that matter most: Salesforce deployed its own SDR agents internally, contacted 130,000 leads in four months, and generated 3,200 qualified opportunities. This isn't just vendor marketing. It's a demonstration of a company using its own solutions at scale.

Salesforce Q4 FY26 Earnings Breakdown — SalesforceBen
Provides a detailed breakdown of the Agentforce numbers, including Benioff's response to the "SaaSpocalypse" narratives. → https://www.salesforceben.com/huge-agentforce-growth-in-salesforce-q4-as-benioff-mocks-saaspocalypse-narratives/
Salesforce Crushed Earnings and Still Can't Get Respect — 24/7 Wall St.
A more skeptical take on the earnings. Worth reading for the investor perspective on whether Agentforce's growth is sustainable. → https://247wallst.com/investing/2026/02/26/salesforce-crushed-earnings-by-25-and-still-cant-get-any-respect/

Governance: The Boring Stuff That Actually Matters

Every major sales AI vendor shipped governance features in February. That's not a coincidence. It means enough customers are now running agents in production that vendors know what breaks.

Clay added credit spend limits and estimated cost warnings.
HubSpot shipped audit trails showing exactly what an agent changed in your CRM.
Salesforce built "review before send" controls and approval flows for agent actions.
Gong launched MCP with clear controls for data access, usage, and provenance.

Gartner forecasts that by the end of 2026, 40% of enterprise apps will embed AI agents (up from 5% in 2025). But it also warns that more than 40% of agentic AI projects will get canceled by the end of 2027 because of escalating costs, unclear ROI, or missing risk controls. The projects that survive will be those that started with proper infrastructure and governance before adding agents on top, while the ones that fail will have started with the agent and tried to figure out the rest later.

Sales AI Tool Updates Q1 2026 — Prospeo
Summarizes Apollo, Clay, and Gartner data in one place. It includes Apollo's waterfall enrichment results: 5% more emails, 7% more phone numbers, and 45% fewer bounces. → https://prospeo.io/s/sales-ai-tool-updates
Salesforce State of Sales Report — Salesforce
The source for the vital statistics: 9 in 10 sales teams are currently using AI or planning to do so within two years, leading to 43% higher win rates and 37% faster sales cycles. It also reveals that only 20% of organizations report material bottom-line impact despite 78% adoption. → https://www.salesforce.com/sales/state-of-sales/

---

📊 The Numbers

Sales teams using AI or planning to within two years — 9 in 10

Higher win rates for teams using AI sales tools — 43%

Faster sales cycles — 37%

Top performers more likely to use AI agents vs. underperformers — 1.7x

Organizations reporting material bottom-line impact from AI — 20%

Organizations experimenting with agents that are scaling effectively — 23%

AI agents that make it to production (Deloitte) — 11%

Salesforce Agentforce ARR — $800M

Agentforce year-over-year growth — 169%

Apollo bounce rate reduction (waterfall enrichment) — 45%

Average AI agents per organization — 12

Projected agent growth by 2027 — 67%

Enterprise apps expected to include AI agents by the end of 2026 — 40%

Global 2000 job roles involving AI agents in 2026 — 40%

Two important statistics stand out: 78% of organizations have adopted the technology, but only 20% report a significant material impact. While 75% plan to invest in agentic AI, only 11% are running agents in production. The technology works, but execution remains the bottleneck.

---

💡 Quick Tip

If you're building sales agents in Scout, start with governance. Set up audit trails and approval flows before you automate outreach. It's tempting to ship the agent first and add controls later, but Gartner's data is clear: that's how 40% of projects end up getting canceled. Build the guardrails first, then let the agents run.

---

👀 What To Watch in April

Multi-agent orchestration is emerging as the next frontier. Goldman's Argenti predicts companies will shift from deploying individual agents to "human-orchestrated fleets of specialized multi-agent teams." Organizations currently average 12 agents each, but this is projected to increase by 67% by 2027. The question isn't whether you'll run multiple agents. It's how they'll coordinate.

The enablement gap is widening, with 65% of CSOs saying their enablement functions are stretched thin. Frontline managers now spend just 9% of their time coaching. AI enablement tools, such as Gong's new AI trainer, are helping to fill that gap, but only for teams that deploy them.

The agent-as-a-service pricing model is coming. Goldman predicts billing will shift from hours worked to tokens consumed. If that happens, the economics of sales automation will change completely. Keep an eye out for early movers in Q2.

Gen Z sales reps are losing about two hours per week to manual data entry compared to their senior colleagues. If you're hiring junior reps and not giving them AI tools, you're essentially paying them to do admin work.

---

📬 That's a Wrap

The theme this month is clear: AI sales agents work. The hard part isn't the technology. It's deploying them with the right controls, connecting them to the right data, and measuring what actually matters.

The CIO piece summed it up best, saying agentic AI in 2026 will be "more mixed than mainstream." That's an honest assessment. And it means there's still a real advantage in figuring out production deployment while everyone else is stuck in pilot mode.

If you've got questions or ideas for topics you'd like us to cover, just reply to this email.

Until next time, The Scout Academy Team