# Sa · Support Agent — element 35 of 58 > Answers at 3 a.m. Turns tickets into resolved conversations. - **Group:** 8 · Support & Success - **Necessity:** Essential - **Price band:** $$ · $30–150/mo - **Maturity:** Stable - **Edition:** v2026.Q3 · verified 2026-09-13 ## Leading tools (v2026.Q3) - **Fin** — the benchmark, per-resolution - **Sierra** — enterprise outcomes on contract - **Decagon** — operator tooling, consumer scale - **Pylon** — b2b support, slack-native - **Lorikeet** — pay only when resolved ## Our take The most proven agent category in production. Resolution rates, not deflection rates, are the metric that matters. ## Combines with Dc, Cr, Vs - Guide: [Best AI support agent for early-stage startups (2026)](https://elems.ai/best/best-ai-support-agent-for-startups.html) ## The top 5 — deep dossier (verified 2026-09-13) Fin, for almost everyone — it publishes the numbers the category is judged on (76% average resolution across 12,000+ customers, 2M resolutions a week), charges a transparent $0.99 per resolved outcome, and runs on Intercom, Zendesk, or Salesforce alike. Sierra is the pick when you're an enterprise buying outcomes on contract — $200M ARR, 40% of the Fortune 50, and a $15.8B valuation say the market agrees. Decagon wins for high-volume consumer ops teams who want the deepest tooling (AOPs, simulations, A/B tests); Pylon when your support lives in Slack and your customers are businesses; Lorikeet when you're a regulated fintech/healthtech that literally refuses to pay for an unresolved ticket. The incumbents' AI (Zendesk, Agentforce) wins only when procurement already decided. 1. **Fin** (Intercom (rebranded Fin) → Salesforce (agreed Jun 2026, closing FY2027 Q4)) — $0.99 per resolved outcome · $9.99 per qualification · standalone from $49/mo (50 resolutions) · +$19/seat if you take the Intercom helpdesk. Best for: Any team that wants the category benchmark with transparent per-resolution pricing — on Intercom, Zendesk, Salesforce, or standalone. Why: The reference point everyone else is measured against: 76% average resolution rate across 12,000+ customers, ~2M resolutions per week, G2 category leader at 90% satisfaction (Aug 2026). Fin crossed $100M ARR at 350% annual growth, and its outcome pricing pushed Intercom's net revenue retention from 112% to 146% — the clearest evidence yet that pay-per-resolution aligns vendor and buyer. Watch: Salesforce agreed to acquire Fin for roughly $3.6B in June 2026, closing in its FY2027 Q4 — a rank-1 seat whose owner changes mid-contract, and whose acquirer sells a competing agent. Resolution rates are self-reported on Fin's own metric definition — no independent cross-vendor benchmark exists. At scale, $0.99 per outcome plus $9.99 qualifications outruns flat-fee rivals, and Intercom steers hard toward its full helpdesk bundle. [https://fin.ai](https://fin.ai) 2. **Sierra** (Sierra (Bret Taylor, Clay Bavor)) — Outcome-based enterprise contracts — pay per resolution/conversation, no public list price, multi-year with services bundled. Best for: Enterprises that want conversation quality and outcome-based contracts with white-glove deployment — the Fortune 500's AI-agent vendor of record. Why: The enterprise standard-setter: $200M ARR by May 2026 (from $26M end-2024), agents for 40% of the Fortune 50 (SiriusXM, ADT, Cigna, Nubank, Ramp), and the most-funded pure-play — a $950M Series E at ~$15.8B (Tiger Global, GV, May 2026). Ghostwriter (Apr 2026) builds production agents from SOPs and transcripts, compressing months of deployment. Watch: No self-serve, no public pricing — you negotiate everything. ~79x ARR valuation is priced for perfection, and deployments remain services-heavy. Publishes no aggregate resolution rate, so quality claims rest on references. [https://sierra.ai](https://sierra.ai) 3. **Decagon** (Decagon) — Per-conversation (volume discounts) or per-resolution (premium, charged only on resolve) — enterprise quotes only. Best for: High-volume consumer scale-ups (fintech, marketplaces, subscription) that want the deepest operator tooling: AOPs, simulations, live experiments. Why: The operator's platform: Agent Operating Procedures in natural language, QA simulations, and A/B experimentation that CX teams actually run themselves. Case evidence is strong — Chime at 70% chat+voice resolution, Duolingo at 80% deflection — and investors priced it at $4.5B in January 2026 (Coatue, Index), reconfirmed by a March tender offer. Watch: $35M ARR (Oct 2025) against a $4.5B valuation is a ~128x multiple — steeper than Sierra's. Headline metrics mix resolution and deflection, which are not the same thing, and it fights Sierra for the same enterprise logos with a fraction of the war chest. [https://decagon.ai](https://decagon.ai) 4. **Pylon** (Pylon) — Platform $59–139/seat/mo (annual, 3-seat min) · AI agents from $100/mo or $0.50/ticket · AI assistants $50/seat. Best for: B2B startups whose support lives in Slack, Teams, and shared channels — helpdesk, success, and AI agents in one system. Why: The B2B-native answer: 1,500+ business customers including Cognition, Linear, Modal, and Airtable, with support agents that investigate across Salesforce, Jira, and Linear rather than just answering FAQs. Relaunched as "the agentic support platform" (Jul 15, 2026) with assist, background, Slack, and support agents on one substrate. Watch: AI pricing is per-ticket, not per-resolution — you pay whether or not the agent resolves. Young platform vs the enterprise players, and consumer-scale B2C volume is not its game. Post-2024 funding beyond the $17M Series A is not independently confirmed. [https://usepylon.com](https://usepylon.com) 5. **Lorikeet** (Lorikeet (Sydney)) — Start $1,500/mo (18k credits/yr) · Scale $4,000/mo (48k credits/yr) · ~1 credit per resolved chat/email ticket · no seat or platform fees. Best for: Regulated fintechs and healthtechs that need 40-step workflows executed correctly — and refuse to pay for unresolved tickets. Why: The purest outcome pricing in the category: you're charged only for successfully resolved tickets, and disputed resolutions are free. Built compliance-first (ISO 27001, SOC 2, HIPAA) for complex, high-stakes journeys — customers include Flex, Linktree, Magic Eden, and Eucalyptus — where a wrong answer is a regulatory event, not an inconvenience. Watch: Small company, small war chest (~$16M disclosed) against billion-dollar rivals. Performance stats (97% faster resolution, 99% accuracy) are vendor-reported without an aggregate resolution rate, and the credit system needs modeling before you can compare cost to $0.99-per-resolution rivals. [https://lorikeetcx.ai](https://lorikeetcx.ai) ### How to choose - If You want the proven default and a price you can put in a spreadsheet → Fin at $0.99 per resolved outcome — it runs on whatever helpdesk you already have, and its 76%-average resolution claim is the number every rival gets compared to. - If You're an enterprise buying a multi-year outcome contract, not a tool → Sierra — 40% of the Fortune 50 already did, and Ghostwriter cuts the deployment slog that used to take quarters. - If You're a consumer scale-up doing millions of tickets and your CX team wants to run experiments themselves → Decagon — AOPs, simulations, and A/B tests are the best operator tooling in the category; negotiate per-resolution, not per-conversation. - If Your customers are companies and half your support happens in shared Slack channels → Pylon — B2B-native ticketing plus agents that investigate across your CRM and issue tracker, from $59/seat. - If You're regulated (fintech, health) and a hallucinated answer is a compliance incident → Lorikeet — resolved-only billing, HIPAA/ISO 27001/SOC 2, and 40-step workflow execution; or Gradient Labs if you want a financial-services specialist. ### The field (21 more) Zendesk AI Agents / Resolution Platform, Agentforce, Ada, Parloa, Gradient Labs, Maven AGI, Crescendo, Gorgias, Siena, Duckie, Freshworks (Freddy AI), Kustomer, Yellow.ai, Aisera, Netomi, DigitalGenius, Forethought (acquired), Ultimate.ai (acquired), Moveworks (acquired), Solvvy (dead), Intercom Resolution Bot (renamed) Full dossier data: https://elems.ai/e/sa.json --- Source: [elems.ai](https://elems.ai/e/sa.html) — the periodic table of the AI-led startup. Data: https://elems.ai/elements.json (CC BY 4.0, cite elems.ai).