Chaos and Order

💡 왼쪽 원문을 읽으면서 오른쪽에 따라 써보세요. Tab 키로 힌트를 받을 수 있습니다.

원문 렌더가 준비되기 전까지 텍스트 가이드로 표시합니다.

Prologue — "Ticketing tools" became "relationship OS" in 2026

Customer support in 2026 is no longer "replying to emails that came in."

- AI agents handle 50-80% of first responses (Intercom Fin, Decagon, Sierra).

- Slack, Discord, in-app messengers, live chat, email, social DM channels — all flow into one queue.

- For B2B SaaS, per-customer Slack-Connect channels have long been the actual "ticketing system" (the Pylon model).

- Korea and Japan run messaging-first models — KakaoTalk, LINE, Channel Talk — as the standard.

- The "tier-1 bot, tier-2 human" model died. "Agent and human collaborating in the same inbox" took over.

The problem is **too many choices**. In the 2010s, "Zendesk vs Intercom" was the whole conversation. In 2026, 30+ tools live in the same category, each with a different persona, pricing model, and AI strategy. The category itself is shifting — Intercom moved to "performance pricing" (per-resolution), Plain insists it is "not a helpdesk but a CS workflow platform."

This piece maps the whole landscape — each camp's key players, AI-first newcomers, regional players in Korea and Japan, and recommendations from solo founder to enterprise.

1. The 2026 customer support map — SMB / enterprise / AI / open source, four camps

To fit 30 tools into your head, split them into four camps first.

┌─────────────────────────────────────────────────────────────────────┐

│ 2026 Customer Support / Helpdesk — 4 camps │

│ │

│ ┌────────────────┐ ┌────────────────┐ ┌─────────────────┐ │

│ │ SMB / Startup │ │ Enterprise │ │ New generation │ │

│ │ │ │ │ │ (Dev/B2B/Collab)│ │

│ │ - Intercom │ │ - Zendesk │ │ - Plain │ │

│ │ - HelpScout │ │ - Salesforce │ │ - Pylon │ │

│ │ - Crisp │ │ Service Cloud│ │ - Front │ │

│ │ - Hiver │ │ - ServiceNow │ │ - Hiver │ │

│ │ - Freshdesk │ │ - HubSpot │ │ │ │

│ │ - Tidio │ │ Service Hub │ │ │ │

│ │ - LiveChat │ │ - Kustomer │ │ │ │

│ │ - Re:amaze │ │ (Goldman) │ │ │ │

│ │ - Customerly │ │ - Gladly │ │ │ │

│ └────────────────┘ └────────────────┘ └─────────────────┘ │

│ │

│ ┌────────────────┐ ┌────────────────────────────────────────┐ │

│ │ AI-first │ │ Open source / Free │ │

│ │ │ │ │ │

│ │ - Decagon │ │ - Chatwoot (OSS, MIT) │ │

│ │ - Sierra │ │ - Tawk.to (free) │ │

│ │ - Ada │ │ - UVdesk, FreeScout │ │

│ │ - Cresta │ │ │ │

│ │ - Forethought │ │ │ │

│ └────────────────┘ └────────────────────────────────────────┘ │

│ │

│ Regional: │

│ - Korea: KakaoTalk Channel, Naver TalkTalk, Channel Talk │

│ - Japan: KARTE Talk, Channel Talk JP, Mercari self-built support │

│ - Europe: Crisp (FR), Userlike (DE), LiveChat (PL) │

└─────────────────────────────────────────────────────────────────────┘

**The boundaries blur.** Intercom started in SMB but now lands enterprise too, and is AI-first via Fin. Salesforce Service Cloud is bolting on Einstein. Chatwoot is OSS but also runs a cloud SaaS. Still, **the starting point** continues to shape the tool's character.

Four axes determine which tool matters in 2026.

1. **Who is the customer** — B2C (tens of thousands of users, short conversations) vs B2B (hundreds of accounts, long and context-rich)

2. **Channel priority** — email-first vs chat-first vs messaging (Kakao, LINE) -first vs Slack-first

3. **AI strategy** — your own agent like Fin vs bolting on Decagon/Sierra vs DIY

4. **Operational scale** — 1-3 people vs 10-50 vs 100+

These four axes drive most decisions. Now into the camps.

2. Intercom + Fin AI Agent — the most aggressive AI pivot

The biggest story in the 2026 customer support category is, without question, **Intercom's AI-first pivot**.

2.1 The evolution of Fin AI Agent

Intercom shipped Fin AI Agent in November 2023. At first it looked like "an AI chatbot," but it evolved fast through 2024-2026.

- **Fin 1 (2023)** — GPT-4 based RAG, pulls answers from help center docs.

- **Fin 2 (2024)** — Claude / GPT multi-model, action calling (refund, subscription change).

- **Fin 3 (2025)** — voice support, multi-turn reasoning, 49 languages.

- **Fin 4 (2026)** — agentic workflows, auto-integration with other systems (Stripe, Salesforce), "Fin decides when to hand off to a human."

Intercom's pitch is simple — **"more than 50% of our customers' conversations are resolved automatically by Fin."** Some customers (Anthropic, Lovable) report 70-80%.

2.2 Performance pricing — pay per resolution

The most interesting shift is **the pricing model**.

Traditional SaaS pricing: per-seat monthly ($25-$200/seat/month).

Fin pricing: **$0.99 per resolution** (resolution-based pricing). If Fin closes out a ticket that a human could have handled, you pay $0.99.

Why this matters:

- For customers, the alignment is "you don't pay if the AI doesn't do the job."

- For Intercom, **margin grows as model costs fall**.

- It is decoupled from seat count, so revenue holds even in a future where AI handles 100%.

In 2024-2025, large customers like Klarna and Linktree publicly reported their automation rates, which validated this model. By 2026, almost every competitor has a similar "outcome pricing" model (Decagon, Sierra, Ada all do).

2.3 Intercom strengths and weaknesses

**Strengths**

- Messenger, help center, ticketing, CRM, and tools in one package — a real "all-in-one."

- Fin is one of the most market-validated, strongest AI agents.

- The in-app messenger is still the category standard.

- Fast release cadence, aggressive market leadership.

**Weaknesses**

- Expensive. A full-stack setup easily runs hundreds to thousands of dollars per month for SMBs.

- Hard to make a UI "not look like Intercom" — the messenger widget is highly recognizable.

- Deep enterprise workflows still trail Salesforce / Zendesk.

- Frequent pricing model changes draw complaints (per-resolution, per-seat, usage-based all mixed).

**When to choose**: fast-growing B2C/B2B SaaS, chat-first support, you want to push AI automation hard.

3. Zendesk — enterprise heavyweight

Zendesk has been the market default since 2007. In 2022 it was taken private by Hellman & Friedman + Permira for $10.2B, and since then has clearly pointed itself at "enterprise market capture."

3.1 Zendesk in 2026

- **Zendesk Suite** — ticketing, chat, voice, messenger, help center, bot in one bundle.

- **Zendesk AI** — own AI agent (Ultimate.ai acquisition in August 2024) plus contextual analytics.

- **Zendesk Workforce Engagement Management** — scheduling, QA, coaching.

- **Sunshine Conversations** — multi-channel messaging (WhatsApp, Apple Business Chat, Instagram).

- **Built for Service Cloud** — deep Salesforce integration also available.

3.2 Strengths and weaknesses

**Strengths**

- **Massive marketplace** — 1500+ integrations, "everything plugs in."

- Enterprise security and compliance — SOC 2, ISO 27001, HIPAA, FedRAMP in progress.

- Most mature multi-brand, multi-region, multi-language operations.

- Full stack including workforce management (QA, scheduling).

- Owns its own AI (Ultimate), no external dependency.

**Weaknesses**

- **Heavy UI** — "using Zendesk" is often agonizing day-to-day for agents.

- Pricing is expensive and complex — five tiers from Suite Team to Enterprise, each add-on separate.

- Overkill for fast-moving SMBs.

- AI integration is often judged less seamless than Intercom Fin.

**When to choose**: 100+ employees, multi-country / multi-channel ops, strict compliance, large contact centers.

4. Front — email collaboration (Series D)

Front comes from a different starting point — the bet that **email is not dead**.

4.1 The Front model

Front is not a helpdesk. It is a **shared inbox**.

- support@, sales@, billing@ group emails collaborated on like Slack.

- Comments on emails, assignments, mentions — Slack's collaboration patterns brought to email.

- SMS, messenger, live chat, WhatsApp also handled in the same inbox.

- Customer context shown in a side panel like a CRM.

Front's strength is that **"conversations," not "tickets," are the unit**. Classic helpdesks turn everything into a numbered ticket; Front lets you collaborate on email threads as they are. So sales, ops, CS, and legal can share the same tool.

4.2 Funding and position

- Series D in 2022, $5.6B valuation (Sequoia, Salesforce Ventures, etc.).

- Some headcount adjustments and an AI-first realignment reported in 2024-2025.

- As of 2026, around 8,500+ customers; ARR not public.

4.3 Strengths and weaknesses

**Strengths**

- Perfect for email-centric teams (B2B sales, logistics, real estate, boutique agencies).

- Collaboration UX is very smooth (comments, mentions, shared drafts).

- Built-in AI reply drafts, automated routing.

- Great for integrated sales + CS + ops teams.

**Weaknesses**

- Awkward for chat-first support (in-app messenger is weak).

- AI agent not as strong as Intercom Fin or Decagon.

- Weak on large contact center workflows (QA, scheduling).

- Pricing can be steep for SMBs.

**When to choose**: email-centric B2B, sales and CS share the tool, want UX that "does not feel like a ticketing system."

5. HelpScout — the indie SaaS favorite

HelpScout has been the indie SaaS and SMB darling since 2011. Design, pricing, and philosophy were all "the opposite of Intercom."

5.1 The HelpScout philosophy

- "Replies that look human" — refuses ticket numbers, auto-signatures, anything that looks like a bot.

- Clean, light UI — agents learn it in five minutes.

- Email first, chat (Beacon) second.

- Public company values — "Be Excellent," remote-first, B Corp certified.

5.2 HelpScout Beacon

Beacon is HelpScout's in-app widget. The alternative to the Intercom messenger.

- Help center search + chat + email form.

- Lightweight and fast, low impact on page load.

- AI responses added in 2024-2026 for first-line automation.

5.3 Where it stands in 2026

- In 2025, Endurance International Group reportedly divested its stake, with a new PE owner taking over.

- Even so, product philosophy remained steady — the "not expensive, not ugly" default for SMB and indie SaaS.

- AI features added, but the company is not going "AI-first" — intentional positioning.

5.4 Strengths and weaknesses

**Strengths**

- Reasonable pricing — Standard starts at $25/user/month.

- Clean UX with almost no learning curve.

- Perfect for brands where "human-feeling support" is a core value.

- Docs (KB), Beacon, and inbox integrate cleanly.

**Weaknesses**

- Weaker enterprise features (workflow automation, deep analytics).

- AI agent not as strong as Intercom Fin or Decagon.

- Weak on multi-brand, multi-region.

- "You grow out of it eventually" is a common opinion.

**When to choose**: 1-50 person SaaS, indie makers, design-first brands that want to "not look like Intercom."

6. Plain — developer-friendly (Series A)

Plain started in the UK in 2020. The core claim was simple — **"existing helpdesks are not developer-friendly."**

6.1 What makes Plain different

- **API-first** — every feature exposed via GraphQL. More "support workflow backend" than helpdesk.

- **TypeScript SDK** — first-class. The SDK feels like a library, not a helpdesk.

- **Deep Slack integration** — automatic two-way sync with customer Slack channels.

- **Workflow automation** — route, tag, escalate tickets with custom code/flows.

- **Fast UI** — keyboard shortcuts as a first-class citizen, Linear-like feel.

6.2 Who uses it

- Developer-tools companies (Vercel, Linear, Sentry-style SaaS).

- B2B SaaS with technical customers.

- Teams who think "the helpdesk should be part of our workflow."

6.3 Funding

- Series A in 2023, $13M, Index Ventures and angel round.

- Additional round reported in 2025 (amount undisclosed).

- Fast growth in 2026, quickly becoming the default in the developer-tools category.

6.4 Strengths and weaknesses

**Strengths**

- API/SDK is really well built — automation and integration are very easy.

- Fast, modern UI (Linear-influenced design).

- Deeper Slack integration than other tools.

- Positioning maps exactly onto the developer persona.

**Weaknesses**

- Overkill for non-developer teams — not a fit for marketing, sales, traditional ops.

- Marketplace / ecosystem smaller than Zendesk.

- AI agent features not as mature as Intercom Fin.

- Pricing can be steep for SMBs.

**When to choose**: developer-tools companies, you need API-first integration, you like the Linear / Vercel aesthetic.

7. Pylon — B2B + Slack-Connect

Pylon started in 2022 and exploded in 2024-2026. The core insight is **"for B2B SaaS, the customer Slack channel is the real support system."**

7.1 The Pylon model

Reality of traditional B2B support:

- 5 customers → 5 Slack Connect channels.

- 100 customers → 100 Slack channels. The CS team loses their mind.

- Conversations are scattered across channels, making SLA tracking, analytics, handoff impossible.

What Pylon solves:

- Merge every Slack channel into one inbox.

- Auto-detect channel messages as tickets, apply priority, assignment, SLA.

- Replies sync back to Slack — CS team works in Pylon, customers see Slack.

- Email, in-app messenger, MS Teams in the same inbox.

7.2 Additional features

- Help center (similar to Plain and Intercom).

- AI reply drafts and automated routing.

- Linear, Jira, GitHub auto-sync (critical for B2B).

- Analytics — per-account response time, frequently asked topics.

7.3 Funding and growth

- 2023 Seed, $4.5M.

- 2024 Series A, $17M (Andreessen Horowitz).

- 2025 Series B, $25M+ reported.

- Customer roster grows fast — Hightouch, Common Room, Rippling, Deel.

7.4 Strengths and weaknesses

**Strengths**

- Pinpoints the real B2B SaaS pain (Slack-Connect chaos).

- Deep workflows — priority, SLA, automated routing.

- Strong engineering-tool integrations (Linear, Jira).

- Fast growth, becoming the category default.

**Weaknesses**

- Less valuable for teams not B2B-Slack-centric.

- Messenger widget (in-app live chat) is weaker than Intercom.

- Pricing is expensive (B2B SaaS pricing model).

- AI agent still maturing.

**When to choose**: B2B SaaS, Slack Connect is your main channel, you need deep engineering-team integration.

8. Freshdesk / Crisp / Hiver — other SMB tools

Beyond the five above, several other SMB tools matter.

8.1 Freshdesk (Freshworks)

- Started in India in 2010, IPO'd on Nasdaq as Freshworks in 2021.

- Started as a cheaper alternative to Zendesk; now a full stack (Freshchat, Freshcaller, Freshsales included).

- Strengthened with AI (Freddy) in 2024-2026, own model plus external LLM.

- Used across SMB, mid-market, and some enterprise.

- **Strengths**: great pricing (free plan available, Pro from $49/agent/month), integrated suite.

- **Weaknesses**: deep workflows don't match Zendesk, AI doesn't match Intercom Fin.

8.2 Crisp (France)

- Started in France in 2015, popular in Europe.

- Multi-channel messenger (web chat, WhatsApp, Messenger, LINE, Telegram, Instagram).

- Video and co-browsing built in.

- **Strengths**: very good pricing (free plan, paid from $25/team/month), strong multi-channel.

- **Weaknesses**: weak enterprise features, AI is basic.

8.3 Hiver (Gmail-based)

- Started in India in 2011, "helpdesk inside Gmail."

- Popular with Google Workspace teams — collaborate inside Gmail with no external tool.

- Shared inbox, assignment, labels, SLA, analytics all integrated into the Gmail UI.

- **Strengths**: almost zero adoption friction for Google Workspace teams.

- **Weaknesses**: tied to Gmail (no Outlook or other mail systems), weak multi-channel.

8.4 Quick mentions

- **Tidio** — SMB e-commerce chat, strong AI chatbot, from $29/month.

- **LiveChat** (Poland) — classic live chat since 2007 (now LiveChat Software AG).

- **Olark** — lightweight live chat, popular for indie sites.

- **Re:amaze** — e-commerce friendly (Shopify, BigCommerce).

- **Customerly** — European SMB friendly, Intercom alternative.

- **Userlike** (Germany) — GDPR-friendly, popular in German-speaking markets.

9. Salesforce Service Cloud / ServiceNow / HubSpot — enterprise

The enterprise market plays a different game. More "customer data platform" than "helpdesk," more "enterprise-wide workflow" than "support."

9.1 Salesforce Service Cloud

- Shipped in 2009, the support module integrated into Salesforce CRM.

- In 2026 it is the enterprise default — for companies already on Salesforce.

- **Einstein** AI — Salesforce's own AI platform, rebranded to Einstein Copilot in 2024, evolved into Agentforce (announced September 2024).

- **Agentforce** — autonomous AI agent platform, aggressively pushed in 2024-2026.

- **Strengths**: one body with Salesforce CRM, enterprise compliance, global ops.

- **Weaknesses**: expensive, complex adoption, "you need Salesforce consultants" tooling.

9.2 ServiceNow

- Started in 2003, came out of IT Service Management (ITSM).

- In 2026 it is the ITSM default, with strong Customer Service Management (CSM).

- Covers workflow automation, internal ops, and external customer support all together.

- AI: Now Assist (own AI, integrated Generative AI).

- **Strengths**: enterprise ITSM integration, deep workflows.

- **Weaknesses**: overkill for general B2C support, heavy UI, expensive.

9.3 HubSpot Service Hub

- Support module integrated into HubSpot CRM.

- Run marketing, sales, and CS on one platform.

- AI features (Breeze Intelligence) strengthened in 2024-2026.

- **Strengths**: natural for SMB and mid-market already on HubSpot, reasonable pricing.

- **Weaknesses**: deep helpdesk features don't match Zendesk.

9.4 Kustomer (Goldman) / Gladly

- **Kustomer**: acquired by Meta for $1B in 2020, divested in 2023, reportedly sold to Goldman Sachs Asset Management in 2024. Strong B2C multi-channel.

- **Gladly**: voice, messenger, email integrated, "one customer view per agent," popular with B2C luxury brands (JetBlue, Crate & Barrel).

10. Chatwoot / Tawk.to — open source / free

For teams under cost pressure or with self-hosting needs, there are open-source / free options.

10.1 Chatwoot (open source)

- Started in India in 2017, MIT license.

- Positioned as "open-source Intercom" — live chat, help center, multi-channel.

- Choice of self-hosting or cloud SaaS (Chatwoot Cloud).

- Multi-channel: web chat, email, WhatsApp, Facebook, Instagram, Telegram, LINE, SMS.

- Active AI integration (external LLM connection) in 2024-2026.

- **Strengths**: free (self-hosted), open code, data sovereignty, active community.

- **Weaknesses**: you have to run self-hosting, AI needs to be DIY via external integration.

10.2 Tawk.to (free)

- Started in 2008, live chat that is free forever.

- Revenue model: paid add-ons (remove branding, hired chat agents).

- Very lightweight widget, 5-minute adoption.

- **Strengths**: actually free, 5-minute adoption.

- **Weaknesses**: no deep features, no (or weak) AI, data sovereignty concerns.

10.3 Other OSS

- **UVdesk** (PHP) — started in India, email-centric.

- **FreeScout** (PHP) — Help Scout clone, self-hosted.

- **Erxes** (Node.js) — open-source marketing + CRM + CS bundle.

11. AI customer support — Decagon (Series B) / Sierra (Bret Taylor) / Cresta / Forethought / Ada

The hottest category in 2024-2026 is **AI-first support agents** — not helpdesks themselves, but the AI layer bolted on top of (or replacing) the helpdesk.

11.1 Decagon

- Started in 2023, quickly raised Series A ($35M, 2024) and Series B ($65M+, 2024).

- Enterprise AI agent — large customers like Klarna, Bilt, Webflow.

- Strengths: deep workflows, action calling (refunds, subscription changes), multi-language.

- Positioned as "the enterprise competitor to Fin."

11.2 Sierra (Bret Taylor)

- Co-founded in 2023 by Bret Taylor (former Salesforce co-CEO, OpenAI board chair) and Clay Bavor (former Google).

- Series valuation around $4.5B in 2024, $10B+ reported in 2025.

- Enterprise AI agent, heavy "voice-first" emphasis.

- Customers: ADT, Sonos, SiriusXM, Casper — large B2C brands.

- Strengths: strong on voice call automation, brand voice controlled via code.

11.3 Cresta

- Started in 2017, "real-time agent assist" first for large contact centers.

- During a live call, the AI suggests the next utterance for the human agent.

- From 2024 expanded into fully automated territory as "Cresta AI Agent."

- Customers: Verizon, Brinks, Cox — large contact centers.

11.4 Forethought

- Started in 2017, strong on AI triage and auto-response.

- An AI layer laid on top of Zendesk, Salesforce, Intercom.

- Shipped Forethought Solve Lite in 2024, expanded into SMB.

11.5 Ada

- Started in Toronto in 2016, pivoted from chatbot to AI agent.

- Rebranded as Reasoning Engine in 2024-2026.

- Strengths: no-code builder, 50+ languages, multi-channel.

- Customers: Square, Verizon, Shopify Plus.

11.6 Business model of AI-first support

Almost all are **outcome-based pricing** (per resolution) or **conversation-based pricing**.

- Decagon, Sierra — $0.50 to $2.00 per resolution (negotiated by customer).

- Ada, Forethought — per conversation or per seat, mixed.

- Cresta — per agent seat (because it's a human-assist model).

The 2026 trend is rapid convergence on "pay more when AI resolves well, pay nothing if it doesn't."

12. Korea — KakaoTalk Channel, Naver TalkTalk, Channel Talk

Korean customer support is **messaging-first** — messaging apps come before email and live chat.

12.1 KakaoTalk Channel

- The de facto standard in Korea — direct reach to 49 million KakaoTalk users.

- Evolved from "Yellow ID" to "KakaoTalk Channel."

- Auto-reply (chatbot), ad delivery, 1:1 chat, order/booking integration.

- Kakao Sync removes member-signup friction.

- Cost: chat is free, ad delivery (Friend Talk, Alimtalk) is metered.

- Nearly every B2C operating in Korea runs a KakaoTalk Channel.

12.2 Naver TalkTalk

- Direct "TalkTalk inquiry" link straight from Naver search results.

- Integrates naturally with Naver Smart Store.

- One of the twin pillars of Korean B2C alongside KakaoTalk Channel.

12.3 Channel Talk (Channel Corp)

- Started in Korea in 2014; in 2026 one of the largest chat tools across Korea and Japan.

- Positioned as "the KR/JP version of Intercom," design also similar.

- Deep localization for Korea, Japan, and Southeast Asia.

- In Japan it grew quickly under the name Channel Talk JP and is one of the defaults for Japanese SaaS.

- Manager chat, AI chatbot (ALF), marketing messages, voice — all integrated.

- A Series round on the order of 100 billion KRW reported in 2023.

12.4 Korea market recommendations

- B2C ops: KakaoTalk Channel is required, Naver TalkTalk strongly recommended.

- B2C SaaS: Channel Talk + KakaoTalk Channel.

- B2B SaaS: Channel Talk or Intercom + Slack Connect.

13. Japan — KARTE Talk, Channel Talk JP, Mercari self-built support

In Japan, LINE plays the role KakaoTalk does in Korea, and there are domestic CRM / CX companies.

13.1 KARTE Talk (Plaid Inc)

- The chat module of KARTE, a CX platform from Plaid Inc (listed in Japan), started in 2015.

- Combines user behavior analytics (every click and scroll on the site) with chat.

- Behavior-based triggers like "auto-pop a chat for users hesitating on this page."

- KARTE has deep penetration in Japanese large EC and media.

13.2 Channel Talk Japan

- Channel Corp (Korea)'s Japan expansion.

- One of the fast-growing SaaS in Japan; reasonable pricing and native Japanese support.

- Local office in Tokyo.

13.3 LINE Official Account + LINE WORKS

- LINE is the Japanese equivalent of KakaoTalk.

- Reply to chats and broadcast marketing via LINE Official Account.

- Active chatbot integration via LINE's Messaging API.

13.4 Mercari self-built support

- Mercari, Japan's large C2C marketplace, built its own support system.

- Complex workflows (identity verification, dispute resolution, refunds) could not be done by external tools, so it stayed in-house.

- Expanding its own AI support stack in 2024-2026.

13.5 Others in Japan

- **PKSHA Communication** — chatbot solutions from Japanese NLP company PKSHA.

- **Mobilus SupportTech** — Japanese contact-center solution.

- **Tayori** — PR Times subsidiary, SMB helpdesk.

- **Zendesk JP** — Japanese-language support, popular in Japanese enterprise.

13.6 Japan market recommendations

- B2C ops: LINE Official Account is required, plus Channel Talk JP or Crisp on site.

- Large EC: KARTE Talk + own CS stack.

- Japanese SaaS going global: Intercom or Channel Talk JP.

14. Who should pick what — solo / SMB / B2B SaaS / enterprise

Now to wrap up. Recommendations by persona.

14.1 Solo founder / indie maker

- **Start free**: Tawk.to or Crisp free plan.

- **Paid**: HelpScout Standard ($25/user/month) — if email-centric.

- **Chat-centric**: Crisp Pro or Intercom Starter.

- **E-commerce**: Tidio or Re:amaze.

- **Developer tools**: Plain Free or a small Pylon plan.

14.2 SMB (10-50 people)

- **Full-stack SaaS**: Intercom (messenger + Fin AI) — expensive but integrated.

- **Email-centric**: Front or HelpScout.

- **AI-first**: Intercom + Fin, or Freshdesk + Freddy.

- **Europe**: Crisp full plan.

- **Korea**: Channel Talk + KakaoTalk Channel.

- **Japan**: Channel Talk JP + LINE Official Account.

14.3 B2B SaaS (50-500 people)

- **B2C-like UX**: Intercom + Fin AI.

- **B2B Slack-centric**: Pylon. Top recommendation.

- **Developer tools**: Plain — quickly growing default.

- **Email / multi-channel**: Front.

- **Low-cost**: Freshdesk + Freddy.

14.4 Enterprise (500+ people)

- **Already on Salesforce**: Salesforce Service Cloud + Agentforce.

- **Contact-center centric**: Zendesk + Ultimate AI, or Salesforce.

- **ITSM**: ServiceNow + Now Assist.

- **Already on HubSpot**: HubSpot Service Hub.

- **AI automation first**: Decagon or Sierra (on top of Zendesk).

- **Voice contact center**: Sierra (voice automation) + Cresta (agent assist).

- **B2C luxury brand**: Gladly.

14.5 Cost pressure / need OSS

- **Self-hosting**: Chatwoot (the most mature OSS).

- **Free**: Tawk.to.

- **PHP stack**: FreeScout (Help Scout clone).

14.6 AI-first (bolt on top of helpdesk)

- **B2C automation**: Decagon, Ada, Forethought.

- **B2B automation**: Decagon, Sierra.

- **Voice automation**: Sierra, Cresta.

- **DIY**: build LLM + RAG + workflow yourself (see Inkeep, Mendable for reference).

14.7 Decision checklist

Five questions to answer before picking a tool.

1. **Who is the customer** — B2C tens of thousands / B2B hundreds of accounts / enterprise.

2. **What is the main channel** — email / chat / messaging (Kakao, LINE) / Slack Connect.

3. **How important is AI automation** — 30% / 50% / 80% automation target.

4. **Budget / preferred pricing model** — per-seat / per-resolution / usage-based / self-hosted.

5. **What CRM or ITSM are you already on** — Salesforce / HubSpot / ServiceNow / none.

These five answers determine 90% of the choice.

15. Closing — customer support in 2026, the next 3 years

Customer support in 2026 sits at a clear inflection point.

- AI handling more than half of first responses has become the standard.

- Pricing models are moving from per-seat to per-resolution.

- B2B is going mainstream with new channels like Slack Connect.

- Regions like Korea and Japan run on messaging (Kakao, LINE) as the standard, and global tools have to follow.

- Human agents focus on "complex cases" and "cases that need empathy."

What to expect over the next three years (2026-2029):

- **AI-first tools consolidate**. Decagon / Sierra / Ada either grow much larger or get acquired by a giant (Salesforce, ServiceNow).

- **The line between helpdesk and CRM blurs further** — the Intercom / HubSpot model becomes more common.

- **Voice automation explodes** — companies like Sierra automate large contact-center work.

- **Regional incumbents strengthen** — Channel Talk (Korea/Japan/SEA), KARTE (Japan), Hiver (India) dig deeper into their markets.

- **OSS renaissance** — tools like Chatwoot grow on the back of EU GDPR and data-sovereignty demand.

Tool choice is not forever. Worth reviewing every two to three years. But **migration cost is high** — help-center content, macros, integrations, and agent training are all baked in. Answering the five questions seriously the first time gives the largest ROI.

References

- Intercom Fin AI Agent — [https://www.intercom.com/fin](https://www.intercom.com/fin)

- Intercom Pricing — [https://www.intercom.com/pricing](https://www.intercom.com/pricing)

- Zendesk — [https://www.zendesk.com/](https://www.zendesk.com/)

- Zendesk Ultimate AI acquisition — [https://newsroom.zendesk.com/](https://newsroom.zendesk.com/)

- Front — [https://front.com/](https://front.com/)

- Front Series D — [https://techcrunch.com/2022/01/26/front-raises-65m-at-1-7b-valuation/](https://techcrunch.com/2022/01/26/front-raises-65m-at-1-7b-valuation/)

- Help Scout — [https://www.helpscout.com/](https://www.helpscout.com/)

- Plain — [https://www.plain.com/](https://www.plain.com/)

- Pylon — [https://usepylon.com/](https://usepylon.com/)

- Crisp — [https://crisp.chat/](https://crisp.chat/)

- Hiver — [https://hiverhq.com/](https://hiverhq.com/)

- Freshdesk (Freshworks) — [https://www.freshworks.com/freshdesk/](https://www.freshworks.com/freshdesk/)

- Salesforce Service Cloud — [https://www.salesforce.com/service/](https://www.salesforce.com/service/)

- Salesforce Agentforce — [https://www.salesforce.com/agentforce/](https://www.salesforce.com/agentforce/)

- ServiceNow — [https://www.servicenow.com/](https://www.servicenow.com/)

- HubSpot Service Hub — [https://www.hubspot.com/products/service](https://www.hubspot.com/products/service)

- Kustomer — [https://www.kustomer.com/](https://www.kustomer.com/)

- Gladly — [https://www.gladly.com/](https://www.gladly.com/)

- Chatwoot — [https://www.chatwoot.com/](https://www.chatwoot.com/)

- Chatwoot GitHub — [https://github.com/chatwoot/chatwoot](https://github.com/chatwoot/chatwoot)

- Tawk.to — [https://www.tawk.to/](https://www.tawk.to/)

- Decagon — [https://decagon.ai/](https://decagon.ai/)

- Sierra — [https://sierra.ai/](https://sierra.ai/)

- Cresta — [https://cresta.com/](https://cresta.com/)

- Forethought — [https://forethought.ai/](https://forethought.ai/)

- Ada — [https://www.ada.cx/](https://www.ada.cx/)

- Tidio — [https://www.tidio.com/](https://www.tidio.com/)

- LiveChat — [https://www.livechat.com/](https://www.livechat.com/)

- Olark — [https://www.olark.com/](https://www.olark.com/)

- Re:amaze — [https://www.reamaze.com/](https://www.reamaze.com/)

- Userlike — [https://www.userlike.com/](https://www.userlike.com/)

- FreeScout — [https://freescout.net/](https://freescout.net/)

- KakaoTalk Channel — [https://business.kakao.com/info/kakaotalkchannel/](https://business.kakao.com/info/kakaotalkchannel/)

- Naver TalkTalk — [https://talk.naver.com/](https://talk.naver.com/)

- Channel Talk — [https://channel.io/](https://channel.io/)

- KARTE — [https://karte.io/](https://karte.io/)

- Plaid Inc (KARTE) — [https://plaid.co.jp/](https://plaid.co.jp/)

- Channel Talk Japan — [https://channel.io/ja](https://channel.io/ja)

- PKSHA Communication — [https://www.pkshatech.com/](https://www.pkshatech.com/)

- LINE Official Account — [https://www.linebiz.com/jp/](https://www.linebiz.com/jp/)

- a16z Pylon investment — [https://a16z.com/](https://a16z.com/)

- Bret Taylor on Sierra — [https://sierra.ai/blog](https://sierra.ai/blog)