AI Add-Ons Trend

Companies that sell AI capabilities as paid extras — either as add-ons to existing plans or as entirely separate SKUs. This page covers both `ai_packaging: add-on` (16 companies) and `ai_packaging: separate-sku` / `separate` / `separate-product` / `standalone` (9 companies). The distinction is degree of separation: add-ons layer onto existing plans, separate SKUs stand alone.

Companies — Add-On (`ai_packaging: add-on`)

Enterprise & CRM

• salesforce — Einstein AI and Agentforce sold as add-on with flex-credits system
• zendesk — AI agents sold as paid add-on to support plans
• pendo — AI analytics as add-on
• okta — AI identity features as add-on

Analytics & Data

• datadog — AI observability features as add-on
• posthog — AI features as add-on to analytics
• segment — AI features as add-on to CDP

Collaboration

• notion — Custom Agents add-on at $10 per 1,000 Notion credits
• asana — AI features as paid add-on
• dropbox — AI features as add-on to storage plans
• backlog — AI features as add-on to project management
• loom — AI features as add-on to video messaging

Communication & IT

• aircall — AI features as add-on to phone system
• pagerduty — AI features as add-on to incident management
• tailscale — AI features as add-on to VPN/networking

SEO

• ahrefs — AI features as add-on to SEO tools

Companies — Separate SKU (`ai_packaging: separate-sku` and variants)

These companies sell AI as a distinct product line, not just an add-on toggle.

• github — Copilot sold as separate subscription ($10-39/user/mo), independent from GitHub plans. MCP now bundled on Free tier.
• clickup — Dedicated AI section: Brain at $9/user/mo, Everything AI at $28/user/mo, plus PAYG credits. Separate from core plans.
• cloudflare — Workers AI as separate product line from CDN/security plans
• vercel — v0 and AI features as separate products from hosting
• atera — Rebranded AI product as "Robin by Atera" — autonomous agent sold separately
• dialpad — New AI Agent tab with conversation-credit pool, separate from voice plans
• digitalocean — GPU Droplets and AI/ML products as separate catalog from core infra
• docusign — AI agreement management as separate product from e-signature
• optimizely — Opal AI as standalone product alongside experimentation suite

The Pattern

Add-on/separate-sku is the second most common approach (~25 companies). The strategic logic differs from bundling:

1. Established products protect margins — Salesforce, Zendesk, Notion already have large installed bases paying for non-AI features. Adding AI cost to the base price would dilute margins across all customers, including those who don't want AI.
2. Price discrimination — sell AI to customers who value it most at a premium. GitHub Copilot at $10-39/seat generates revenue specifically from developers who use AI, without raising GitHub's base price.
3. Usage cost isolation — AI inference is expensive. Notion's credit system ($10/1,000 credits) and ClickUp's PAYG credits let the company pass variable AI costs directly to heavy users.
4. Hedging the bundle bet — some companies (Notion, Asana) may eventually bundle once AI costs drop. The add-on is a temporary monetization vehicle while unit economics stabilize.

The trend: `separate-sku` companies are getting bolder. ClickUp, Atera, and Dialpad built dedicated AI product lines rather than simple feature toggles. This suggests confidence that AI is a standalone value prop worth its own pricing page.

See also: ai-bundling, ai-credits, no-ai