[object Object]

The Bet

Salesforce opens Agentforce 360 to ISVs so partners can build and distribute AI agents using platform capabilities. The vision: partner-built AI-native companies powered by Salesforce but going to market independently. The platform provides reasoning (Atlas), grounding (Data 360), trust (Einstein Trust Layer), distribution (AgentExchange), billing, and audit. The partner provides the vertical depth, the customer relationships, and the eval discipline. The model resembles a managed marketplace more than a traditional ISV channel — closer to how Stripe or Shopify operate than how AppExchange has historically worked.

Virtual Employee Economy

Salesforce’s term for this emerging category. Agents as “virtual employees” — they work, they deliver outcomes, they bill per outcome. Partners productize specific agent jobs (sales development representative, procurement coordinator, AR collector, legal review analyst, technical recruiter). Pricing typically follows headcount-displacement math: charge a fraction of a junior employee’s loaded cost in exchange for capacity that scales without hiring lag.

Per-job pricing benchmarks (illustrative, mid-2026):
  SDR-style outreach agent      $1.50-$3.00 per qualified meeting booked
  Tier-1 support agent          $0.40-$1.20 per resolved ticket
  AR collections agent          0.5-1.5% of dollars collected
  Recruiter screening agent     $5-$15 per qualified candidate forwarded
  Procurement coordinator       $25-$75 per PO routed

The Coding Shop Opening

Five-person coding shops and early-stage startups get explicit focus. The $50M AgentExchange Builders Initiative funds them with grants up to $250k and SAFE notes up to $1M. Engineering support comes through the Forward Deployed Engineering Partner Network. Salesforce wants the long tail of vertical innovation, not just the established AppExchange majors. Two early case studies: a four-person team building a HIPAA-compliant utilization-review agent for health plans, and a six-person team building a SOX-compliant revenue-recognition agent for SaaS finance teams.

Risk Factors

Partners building on Salesforce depend on platform stability, pricing predictability, and Salesforce’s continued commitment to the Open Agentforce thesis. Past platform pivots (Heroku pricing changes, MuleSoft repositioning, ExactTarget rebrand) stung partners. Due diligence includes “what if Salesforce deprecates X” — diversify model dependencies where possible (don’t hardcode to Salesforce xGen if you can swap to Claude or GPT), keep your core domain logic in portable services, and negotiate floor-and-ceiling platform fees in your partner agreement before scale.

Portability checklist for AI-native ISVs:
  Domain logic in portable code (not buried in Flow XML)
  Eval set runnable outside Agentforce Testing Center
  Customer data model documented and exportable
  Billing system independent of Salesforce checkout (optional fallback)
  Model contracts negotiated separately where economics allow

What Strong AI-Native Companies Look Like

Narrow vertical, deep eval coverage, a named design partner producing real outcome data, a unit-economics model with positive contribution margin from day one, and a clear plan for the next vertical to expand into once the first is proven. Vague horizontal pitches don’t get funded and don’t get adopted.

What to Do This Week

If you’re considering an AI-native build on Salesforce, write a one-page thesis: vertical, eval metric, design partner, unit economics, portability plan. Pressure-test it against the Builders Initiative criteria before you write code.

[object Object]
Share