Decagon AI vs Salesforce Agentforce: Pricing, Architecture, When Each Wins (2026)
Two of the most-cited enterprise AI agent platforms in 2026. Decagon AI is the standalone, model-agnostic, enterprise-focused option; Salesforce Agentforce is the Salesforce-native, multi-pricing-model option. They compete in some deals and not in others - this page maps where each wins.
Headline pricing comparison
| Dimension | Decagon AI | Salesforce Agentforce |
|---|---|---|
| Pricing model | Custom enterprise contract (per-conversation default) | Per conversation ($2), Flex Credits ($500/100K), per user ($5-$150), Agentforce 1 Edition ($550/user/mo) |
| Published pricing | No price card; custom enterprise quote only | Partially published ($2/conversation, Flex Credits, editions) |
| Company scale | $4.5B valuation, $250M Series D (Jan 2026) | Part of Salesforce (NYSE: CRM) |
| Entry commitment | Enterprise contract; no free tier or self-serve trial | $0 (Foundations free tier for Enterprise+) |
| Architecture | Standalone, model-agnostic agentic platform | Salesforce-native, Einstein + Data Cloud |
| Best fit customer profile | High-volume enterprise B2C, complex policy-driven support, multi-system data | Existing Salesforce customer, CRM-centric workflows, varying volume |
Decagon does not publish a price card; we do not substitute an aggregator ACV estimate for a vendor-published rate, so no Decagon price is stated. Decagon funding (Series D, $4.5B, Jan 2026) from decagon.ai. Agentforce figures from Salesforce's published Flex Credits, per-conversation, and Agentforce 1 Edition list pricing. Reviewed September 2026.
When Decagon wins
- +Multi-system data context. If your AI agent needs to reason over data spread across CRM, billing, support, and product analytics, Decagon's model-agnostic architecture is more flexible than Salesforce-native Agentforce. Agentforce works hardest when most of the data lives in Salesforce already.
- +Model portability requirement. If you need to run the same agent definitions against OpenAI, Anthropic, Google, and an internal LLM (for cost, latency, or compliance reasons), Decagon is built around that flexibility. Agentforce is Einstein-and-Salesforce-Data-Cloud-centric.
- +Regulated industry with complex policy-driven resolution. Decagon's public customer base includes regulated B2C and fintech. The platform's reasoning over structured policy documents is the marketed differentiator.
- +High-volume customer support with material deflection economics. At 100,000+ resolved conversations per month with high agent loaded cost, the custom-contract economics typically beat Agentforce per-conversation pricing.
When Agentforce wins
- +Already a Salesforce shop. Native CRM integration, shared identity, single billing relationship, single procurement contract. The consolidation value is real for organisations already deeply on Salesforce.
- +Variable / unpredictable conversation volume. Per-conversation or Flex Credits consumption pricing scales cleanly with volume; Decagon's custom-contract structure does not work as well when monthly volume is highly variable.
- +Mid-market scale below the Decagon enterprise band. At 5,000 to 50,000 conversations per month, Agentforce per-conversation pricing is typically cheaper than a Decagon enterprise contract.
- +Want to start small. Agentforce Foundations (free tier) and Flex Credits let you start with no commit. Decagon is sold via enterprise contract from the start.
Frequently asked questions
How much does Decagon AI cost in 2026?
Decagon does not publish a price card publicly. It prices on a custom enterprise contract basis, typically per conversation (the default) or per resolution, and the only ACV figures in circulation are buyer-reported aggregator estimates rather than vendor-published rates, so none is stated here. What is on the record: Decagon raised a $250M Series D in January 2026 at a $4.5B valuation (led by Coatue and Index Ventures), so this is an enterprise-scale product sold through a sales process, not a self-serve rate. Expect an 8-16 week implementation alongside licensing.
Decagon vs Salesforce Agentforce: which is right for me?
Decagon is the right choice when (1) you want an AI agent platform that is genuinely model-agnostic and not tied to a single LLM provider, (2) you are running a customer-support volume that requires fully agentic resolution rather than mostly deflection, (3) your data is spread across multiple sources of truth and the agent needs deep integration into all of them. Agentforce is the right choice when (1) you are already Salesforce-centric and want native CRM integration, (2) per-conversation or per-credit consumption pricing fits your volume better than a custom enterprise contract, (3) you want to consolidate the AI agent line item into an existing Salesforce relationship rather than add a separate vendor.
Who actually uses Decagon AI?
Decagon's publicly-disclosed customers in 2025-2026 include enterprise B2C brands (Eventbrite, Bilt, Vanta, Notion, Curology, Substack), regulated industries with complex policy-driven support (Bilt rewards, fintech, regulated SaaS), and mid-to-large e-commerce. The pattern is high-volume customer support where each ticket carries real complexity and the savings from deflection are material at scale.
What is different architecturally about Decagon versus Agentforce?
Decagon is a standalone AI agent platform built around model-agnostic agentic reasoning - meaning the same agent definitions can run against multiple LLMs (GPT, Claude, Gemini, internal models) without rewrites. Agentforce is built natively on the Salesforce Data Cloud and Einstein layer, deeply integrated with the broader Salesforce stack. Decagon trades the deep Salesforce integration for portability and model choice; Agentforce trades portability for native-Salesforce data context and consolidated billing. Neither is structurally better - the answer depends on whether your data and workflows live primarily inside Salesforce or across many systems.
What pricing model do agentic AI platforms generally use in 2026?
Three dominant models. (1) Per resolved conversation - Intercom Fin $0.99/resolution, HubSpot Breeze $0.50/resolved conversation, Sierra AI custom. (2) Per conversation or per credit / consumption - Salesforce Agentforce $2/conversation, Flex Credits $500/100K, Decagon (custom contract structure). (3) Per seat with usage caps - older-generation chatbot vendors and migrating enterprise CS platforms trying to add agentic capability. The market is consolidating around per-conversation, per-resolution, or per-credit consumption pricing for genuinely agentic platforms; pure per-seat models are increasingly seen as legacy.
Is Decagon AI worth the price?
For organisations with sustained customer support volume above roughly 50,000 conversations per month and complex resolution requirements (multiple data sources, policy-driven decisions, regulatory documentation), Decagon's ROI math typically works at a 50 to 70% ticket-deflection rate. Below that volume threshold, simpler platforms (Intercom Fin, HubSpot Breeze, Zendesk AI) are usually more cost-effective. The break-even moves with your loaded support agent cost per ticket; in high-cost regions (US, UK, Australia) the break-even sits lower than in lower-cost regions.