Salesforce, Snowflake, Databricks Ship AI Agents That Carry Quota
SaaStr AI 2026 ran end to end on agents in production. Not demos. Not prototypes. Agents carrying quota, writing to systems of record, and changing how sales teams operate.
What Shipped
Rubrik's Ruby agent handles capacity planning that used to take customers a full day. The pattern: LLM generates the plan, deterministic code executes it. Keep the probabilistic part away from production systems.
Webflow now runs an answer engine optimisation agent that detects technical AEO gaps, drafts content, and executes autonomously. The number that matters: customers saw a 75% bump in organic traffic. Answer engines now drive 50% of site visitors, up from 10% last year. If your product does not show up in agent responses, you are invisible to buyers.
Glean went from information retrieval to agentic performance. It now operates as a single MCP server feeding company context into Claude, Cursor, and other tools. The spillover effect: external usage pulls users back into Glean's own surfaces.
Harvey, the legal AI platform valued at $9 billion, deploys legal engineers alongside forward deployed engineers. Most of Harvey's engineers practiced law for eight to ten years before joining. Selling AI into law firms is harder than selling security software, because partners often pay out of pocket.
What This Means for Sales Teams
The buying decision is moving from humans to agents. If your product is not the default an agent recommends, your AE does not get a shot. Webflow called it agent-led growth. The panel called it the new moat.
Accountability shifted. When an agent takes core actions, the vendor stays responsible for what customers do with it, even when nobody predicted the workflow. Build guardrails before you ship autonomy.
Comp structures will follow. If agents are carrying quota, expect comp plans to adjust for agent-assisted deals, agent-sourced pipeline, and what happens when the agent closes without a human touch.
The Numbers
Salesforce restructured its sales org in 2024 to focus on AI-driven efficiency. Snowflake employs roughly 3,000 in sales globally, with expanding ANZ presence in Sydney and Melbourne. Databricks employs around 10,000 globally, including a growing Sydney team. Harvey raised $100 million in Series B in 2025 and is scaling its sales team to target enterprise legal firms.
The frame from SaaStr founder Jason Lemkin: old SaaS may be dying, but B2B plus AI is in a Cambrian explosion. The goal is not a cheap CRM clone from 2006. The goal is a CRM that puts deals on your calendar, worth $50,000 to $100,000 even to a small company.
Agents are in production. Comp plans will catch up.