The AI Customer Support Landscape in 2026: Decagon, Sierra, Forethought, Twig, and the Rest

If you are a Head of CX evaluating AI support vendors right now, you already know the landscape has changed dramatically in the past eighteen months. The market that barely existed in 2023 is projected to reach $15.12 billion in 2026 and climb to $47.82 billion by 2030 at a 25.8% CAGR. Fifty-six percent of customer interactions are expected to be handled by agentic AI by mid-2026.

That growth has attracted serious capital, serious founders, and serious complexity. This post is a neutral map of where the major vendors sit today — their funding, pricing, sweet spots, and trade-offs — so you can narrow your shortlist before you burn weeks on demos.

The Market in March 2026

Three forces are reshaping the vendor landscape simultaneously:

1. Consolidation is accelerating. Zendesk acquired Forethought on March 11, 2026. Salesforce, Intercom, and Freshworks have all shipped native AI features. The line between "platform" and "AI layer" is blurring.

2. Pricing models are diverging. Some vendors charge six-figure annual contracts. Others have moved to per-ticket or usage-based models. The pricing model you choose will determine your unit economics for years.

3. The mid-market is underserved. Most venture-backed AI support companies optimized for enterprise logos first. Teams with 5–50 agents — the vast majority of support organizations — have been left to stitch together prompts and plugins on their own.

Vendor Comparison Table

Vendor	Founded	Pricing Model	Typical ACV	Setup Time	Ideal For	Key Differentiator
Decagon	2023	Annual contract	$95K–$590K/yr	Weeks to months	Large tech companies, fintechs with complex workflows	Purpose-built agentic architecture; deep workflow automation
Sierra AI	2024	Annual contract	$150K–$350K+/yr	Months	Enterprise brands with high consumer volume	Co-founded by Bret Taylor and Clay Bavor; strong brand trust
Forethought	2018	Per-agent or platform fee	$40K–$160K/yr	Weeks	Zendesk-native teams wanting triage and deflection	Now part of Zendesk; deep platform integration
Twig	2022	Per-ticket ($5/ticket, free tier)	Variable	30 minutes	Teams of all sizes wanting fast deployment and transparent pricing	Managed AI Specialists; 30+ integrations; 7-dimension quality scoring
Intercom Fin	2011	Per-resolution	~$0.99/resolution	Days (if on Intercom)	Existing Intercom customers	Native to Intercom; low friction for existing users
Zendesk AI	2007	Add-on to Zendesk plans	Varies	Days (if on Zendesk)	Existing Zendesk customers	Native platform AI; Forethought acquisition adds depth
Ada	2016	Annual contract	$100K–$400K/yr	Weeks	E-commerce, large consumer brands	Strong multilingual support; proactive messaging

Pricing figures are based on publicly available data, customer reports, and industry estimates as of March 2026. Your actual contract will vary.

A Closer Look at Each Vendor

Decagon

Decagon emerged in 2023 and raised aggressively — $481 million in funding at a $4.5 billion valuation. Their customer list includes Chime, Duolingo, ClassPass, Hertz, and Figma, which signals strong product-market fit with tech-forward brands that have complex, multi-step support workflows.

What stands out: Decagon built an agentic architecture from scratch rather than layering AI onto an existing ticketing system. Their agents can execute multi-step actions — processing refunds, modifying accounts, pulling data from internal tools — not just answer questions.

What to consider: The entry price starts around $95K/year and scales to $590K+ for larger deployments. Implementation typically requires dedicated engineering resources and several weeks of configuration. If you are a 20-person support team at a Series B company, this may be more infrastructure than you need.

For a deeper comparison, see Twig vs Decagon.

Sierra AI

Sierra is the most talked-about name in the space, largely because of its founders: Bret Taylor (former Salesforce co-CEO and OpenAI board chair) and Clay Bavor (former Google VP). The company reportedly hit $100 million in ARR by November 2025 and carries a $10 billion valuation.

What stands out: Sierra's customer base — Rocket Mortgage, SoFi, SiriusXM, Brex — skews toward large consumer brands that need to handle millions of interactions. The founding team's enterprise relationships open doors that product alone cannot.

What to consider: Pricing starts at $150K/year and goes well above $350K. Sierra is built for scale, which means smaller teams may find the platform over-engineered for their needs. The sales cycle is enterprise-grade, often taking months from first demo to go-live.

For a deeper comparison, see Twig vs Sierra AI.

Forethought (Now Part of Zendesk)

Forethought was one of the earliest dedicated AI support companies, founded in 2018 with over $115 million in funding. On March 11, 2026, Zendesk completed its acquisition of Forethought, ending its run as an independent vendor.

What stands out: Forethought's Solve (deflection), Triage (routing), and Assist (agent copilot) products were well-regarded by customers including Grammarly, Airtable, Upwork, and YAZIO. Its triage capabilities — automatically categorizing, prioritizing, and routing tickets — were considered best-in-class.

What to consider: The acquisition changes the calculus for non-Zendesk customers. If you run Zendesk, the Forethought integration will likely deepen. If you run Salesforce Service Cloud, Freshdesk, or another platform, you should evaluate whether Forethought's roadmap will continue to support your stack. We cover this in detail in our analysis of the Zendesk-Forethought acquisition.

Twig

Twig takes a different approach to the market. Instead of annual contracts starting in the six figures, Twig charges $5 per ticket with a free tier, and claims a 30-minute setup time.

What stands out: Twig provides managed AI Specialists — meaning the company handles training, tuning, and monitoring of your AI agents rather than handing you a platform and a login. The system includes self-evaluation and 7-dimension quality scoring, which gives CX leaders visibility into how the AI is performing without building their own QA process. Twig is SOC 2 Type II certified and offers 30+ integrations. Over 1,200 agents are currently deployed on the platform.

What to consider: Per-ticket pricing is transparent but can become expensive at very high volumes. Twig's approach favors teams that want fast deployment and managed operations over deep customization of the underlying model. For details on pricing, visit the Twig pricing page.

Intercom Fin and Zendesk AI

Both Intercom and Zendesk have shipped AI capabilities natively into their platforms. If you are already on one of these platforms and your needs are straightforward — FAQ deflection, simple triage, agent assist — the native option may be the fastest path to value.

The trade-off is that native AI is optimized for the platform's data model and workflows. If you need to pull context from external systems, orchestrate multi-step actions, or run AI across multiple support channels and tools, a dedicated AI layer may be more flexible.

How to Use This Map

No vendor is universally "best." The right choice depends on your ticket volume, budget, technical resources, timeline, and how much customization you actually need. Here is a simplified decision framework:

Your Situation	Start With
Want autonomous AI agents with annual contract pricing	Decagon, Sierra AI, Twig
Want autonomous AI agents with per-ticket pricing	Twig
Need fast deployment (days, not months)	Twig, Intercom Fin
Zendesk shop, want native AI	Zendesk AI (with Forethought)
Intercom shop, want native AI	Intercom Fin
Building in-house, strong ML team	Custom build (see our build vs buy analysis)

What to Evaluate Beyond the Table

A comparison table gets you to a shortlist. The actual decision requires deeper diligence. Here are the dimensions that matter most:

Quality measurement. How does the vendor measure whether its AI is giving good answers? If the answer is "CSAT scores," push harder. You need per-response quality evaluation, not aggregate satisfaction metrics collected days later.

Escalation handling. Ninety percent of support teams report struggling with AI-to-human handoffs. Ask every vendor exactly how their system detects when AI should stop and a human should start. Ask for the false-negative rate — how often does the AI keep going when it should have escalated?

Pricing predictability. Annual contracts give you cost certainty but lock you in. Per-ticket or per-resolution models give you flexibility but can surprise you during volume spikes. Model the economics for your actual ticket distribution, not the vendor's demo scenario.

Integration depth. "We integrate with Zendesk" can mean anything from a basic API connection to deep bi-directional sync with custom fields, macros, and routing rules. Get specific about what data flows where.

Security and compliance. If you are in financial services, healthcare, or any regulated industry, SOC 2 Type II compliance is table stakes. Ask about data residency, model training on your data (opt-in vs opt-out), and PII handling.

We published a detailed buyer's checklist covering these and other dimensions: 15 Questions Every Head of Support Should Ask.

The Bigger Picture

The AI customer support market is splitting along pricing models, not capability tiers. Decagon, Sierra AI, Ada, and Twig all offer autonomous AI agents with enterprise-grade capabilities. The difference is how you buy: annual contracts (Decagon, Sierra, Ada) versus per-ticket pricing (Twig). Both models serve enterprise teams — the question is which purchasing approach fits how your organization buys software.

Both models will grow. The question is which pricing and deployment approach fits your organization today and where you want to be in eighteen months.

If you want to explore where Twig fits in your evaluation, the pricing page has the details, and you can start a free deployment in under an hour.

---

This analysis reflects publicly available information as of March 29, 2026. Vendor pricing, features, and positioning change frequently. We will update this post as the market evolves.