How much does Help Scout cost?

Free tier (5 users, 1 Inbox, 1 Docs site); Standard $25, Plus $45, Pro $75 per user/mo, 16% off annual; AI Answers add-on at $0.75 per resolution with a 3-month free trial; extra Inboxes $10/mo, extra Docs sites $20/mo; 15-day trial on Standard and Plus.

What are the best alternatives to Help Scout?

Based on documented capability coverage across the Agentic Index 14 point taxonomy, the closest alternatives to Help Scout among customer support agents are LivePerson, Fini, My AskAI, Parahelp. Each is graded on the same public evidence, and no vendor pays for placement.

Does Help Scout support MCP and tool calling?

Full under the Agentic Index verification standard. Help Scout publishes public evidence of Model Context Protocol (MCP) support, tool calling, and API or SDK extensibility for connecting agents to external tools. Evidence basis: A documented developer platform for Help Scout's own product: the Mailbox API v2 with read and write access and per-plan rate limits of 200, 400 and 800 calls a minute, the Docs API, an Apps framework rendering inside the conversation sideb. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Does Help Scout have SOC 2, SSO and role based access control?

Full under the Agentic Index verification standard. Help Scout publishes public evidence of enterprise security and identity controls, covering a compliance posture such as SOC 2 or ISO 27001, single sign on (SSO or SAML), and role based access control (RBAC). Evidence basis: SAML single sign-on works against any compatible identity provider, with published setup guides for Okta, OneLogin and generic providers and a force SAML option, alongside IP restrictions limiting the account to defined address ranges and a. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Does Help Scout require human approval before an agent acts?

Full under the Agentic Index verification standard. Help Scout publishes public evidence of human approval steps and guardrails, so an agent action can be held for review before it executes. Evidence basis: Guardrails are a shipped, vendor-owned gate before the action: the operator lists topics the agent must not engage, and every Beacon running the agent carries a setting defining the response and immediate escalation to the team the moment o. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Back to vendors

Help Scout

Also known as: AI Answers, AI Agents, Beacon, Help Scout Docs, AI Drafts

Visit site

Customer support agentindependentVerified 2026-09-05

Entry priceFrom $25/user/moFull pricing detail

Email-first help desk for small and mid-sized teams, pairing a shared inbox and knowledge base with an AI agent that answers customers in the web widget and drafts replies for the team.

Help Scout is a customer support platform built for small and mid-sized teams that want a help desk without the heavy, ticket-system feel. Founded in 2011 in Boston and certified as a B Corp, it centers on a shared inbox where customer conversations from email, live chat, and social channels like Facebook Messenger and Instagram appear as threaded, email-like exchanges rather than ticket numbers, so replies feel personal and come from real agent addresses.

Around that inbox sit two more pillars: Docs, a self-service knowledge base, and Beacon, an embeddable widget that combines live chat, FAQ search, and email contact on a website or in an app. Help Scout's design philosophy favors gradually layering AI onto human support rather than replacing the team.

Its AI is a set of four features. AI Answers is the customer-facing piece: an autonomous chatbot inside the Beacon widget that draws on a company's knowledge base articles and any website sources it has been given, answers common questions conversationally around the clock, replies in many languages, and hands off cleanly to a human agent when it cannot resolve an issue. Teams can review every AI Answers conversation and see reporting on which questions were resolved, left unhelpful, or escalated.

The other three features assist human agents inside the inbox. AI Drafts generates a first-draft reply from past conversations and knowledge base content that the agent can edit before sending. AI Assist adjusts the tone, length, grammar, or language of a reply or knowledge base article with one click. AI Summarize condenses long back-and-forth threads into a quick set of bullets so an agent can get up to speed fast.

All of it runs on OpenAI's foundation models, with lightweight adapters Help Scout trains itself for voice, tone, and formatting. Your account data is not used to retrain the base model, though Help Scout does say limited samples may go toward tuning those adapters and evaluating feature quality. Operators set guardrails naming topics the agent must not engage, and it escalates to the team the moment one comes up. A Test tab lets you simulate customer conversations before a change goes live, and test conversations are not billed.

For buyers weighing it: Help Scout is SOC 2 Type 2 certified, runs annual white-box penetration testing, and will sign a BAA for HIPAA workloads. SAML single sign-on and IP restrictions are included on Pro and available as add-ons below it, and permissions run across a four-role model with per-inbox access. It hosts on AWS in the United States only, with no regional choice, which matters for teams carrying EU or other residency requirements.

Beyond a catalog of more than 100 apps there is a read/write API, webhooks, a custom apps framework, and Help Scout's own MCP server for connecting an outside assistant. Plans run from a free tier through Standard at $25, Plus at $45, and Pro at $75 per user per month, with the AI Answers chatbot billed separately at $0.75 per resolution.

Vendor details

Canonical URL

https://helpscout.com

Subcategory

Support — assistant + agent

Company status

independent

Use cases & customers

Primary use cases

customer supportshared inboxdocumentation

Target customers

SMBsupport teams

Deployment options

SaaS

In practice

Your small team gets the same handful of how-to questions all day. Help Scout's AI Answers sits in the Beacon widget, answers them from your Docs around the clock, and hands off to an agent when it can't.

A new support hire writes replies that don't quite match your brand voice. Help Scout's AI Assist adjusts tone, length, and grammar with one click, and AI Drafts gives them a first reply pulled from past conversations.

An agent picks up a long, messy email thread and has to read the whole history. Help Scout's AI Summarize condenses it into a few bullets so they get the context in seconds.

Agentic Index coverage score

8.5 / 14 capabilities · 61%

Integrations & Tool Calling	Full
A named catalog of 100+ apps, plus documented custom apps, a read/write Mailbox API v2 with per-plan rate limits, webhooks and OAuth apps, so the customer is not confined to a fixed connector list. Integrations take authenticated action in outside systems rather than only reading from them: Jira creates a new issue from a conversation and Linear links, tracks and creates issues from any conversation, alongside Salesforce, HubSpot and Aircall. Sourcehelpscout.com/pricingread 2026-09-05
Workflow Orchestration	Partial
Rules-based Workflows organize, assign and reply to conversations automatically, with advanced workflows triggering actions on customer properties, round robin and load-balanced routing and SLA policies, tiered from 150 basic to unlimited advanced across the plans. These workflows are single step inbox automation rather than multi step or multi agent execution. The AI Agent is documented as a configuration surface for knowledge, behavior and guardrails that does not act independently, and no task chaining, branching or agent to agent handoff runtime is documented. Sourcedocs.helpscout.com/article/1757-ai-agents-in-help-scoutread 2026-09-05
Knowledge Grounding & RAG	Full
Knowledge sources are ingested into a maintained structure rather than assembled per run: websites taken as a single page, a whole-site crawl or a custom selection, PDF, Word, Excel and Google Docs content, a Help Scout Docs site from the account, and operator-authored Improvements layered on top. Each source syncs on add, the sync time scales with site size, and the operator resyncs manually when the underlying content changes; the store is reused across every Beacon sharing the agent. The transparency hub confirms relevant snippets are retrieved against it at request time. Sourcedocs.helpscout.com/article/1757-ai-agents-in-help-scoutread 2026-09-05
Human Oversight & Guardrails	Full
Guardrails are a shipped, vendor-owned gate before the action: the operator lists topics the agent must not engage, and every Beacon running the agent carries a setting defining the response and immediate escalation to the team the moment one is raised. Help Scout states the agent does not act independently and operates only within the permissions set for it, copilot output is generated into a draft for a person to edit or discard before anything reaches a customer, AI Answers is off by default until enabled per Beacon, and every AI surface is labeled to the customer. Sourcedocs.helpscout.com/article/1757-ai-agents-in-help-scoutread 2026-09-05
Security, Identity & Governance	Full
SAML single sign-on works against any compatible identity provider, with published setup guides for Okta, OneLogin and generic providers and a force SAML option, alongside IP restrictions limiting the account to defined address ranges and a four role model of Account Owner, Administrator, User and Light User with per inbox grants and owner configurable permission sets. On compliance, Help Scout holds SOC 2 Type 2 for Security and Availability with the report on request, runs annual white box penetration testing by an independent firm with the summary on request, and covers HIPAA with a signed BAA and an AI feature healthcare addendum, GDPR, UK GDPR and CCPA, with an executable DPA and a published subprocessor list. Sourcedocs.helpscout.com/article/15-user-roles-and-permissionsread 2026-09-05
Observability & Auditability	Partial
Pre-built and custom reports with CSV and Excel export cover volume, response times and happiness with up to all-time history, the Beacon Insights tab breaks AI Answers conversations into contact helped, contact not helped and human escalation, and the agent's Connections view shows where AI is running across the account. All of this reports what happened, not why. No decision trace, no retrieval log naming which source produced a given answer, and no configuration audit trail is documented. Sourcedocs.helpscout.com/article/1736-manage-ai-answersread 2026-09-05
Memory & State Persistence	Not documented
The AI Agent configuration surface covers knowledge, identity, guardrails, testing and where the agent runs, with no memory layer of stated scope or lifetime among them. The transparency hub states that relevant context is sent to the model with each request and is not absorbed into it, and that Improvements are knowledge the operator writes rather than state the agent retains between conversations. Sourcehelpscout.com/company/legal/ai-transparencyread 2026-09-05
Deployment & Data Residency	Not documented
Help Scout states on its own security page that it is hosted on AWS servers in the United States while serving customers worldwide. There is no region menu, no EU or other regional option, and no VPC, single-tenant or self-hosted deployment anywhere in the documentation or the plan matrix. A buyer with a different residency requirement has no other option. Sourcedocs.helpscout.com/article/1263-security-at-help-scoutread 2026-09-05
Prebuilt Agents, Templates & Packs	Partial
Adoptable assets exist but the customer assembles them: an unlimited library of saved replies inserted in two clicks, tags, workflow automations and the Docs sites the team writes. The AI Agent is created through a configuration wizard the operator fills in rather than chosen from anything, and no prebuilt agent, template gallery or marketplace of packaged agents is documented. Sourcedocs.helpscout.com/article/1757-ai-agents-in-help-scoutread 2026-09-05
Triggers & Channel Coverage	Full
Email, live chat, Facebook Messenger, Instagram and WhatsApp reach the desk natively, with phone, voice and SMS through integrations, plus the embeddable Beacon widget and contact forms, webhooks and OAuth apps for event delivery, and workflows triggered on conversation state and customer properties. The autonomous AI Answers agent serves the Beacon widget specifically, while the assist features work the shared inbox across every channel above. Sourcehelpscout.com/pricingread 2026-09-05
Model Flexibility & Routing	Not documented
The AI transparency hub names OpenAI as the sole foundation model provider, with lightweight LoRA adapters Help Scout trains itself for voice, tone and formatting on top of it. No model selection, routing or bring-your-own-model control appears anywhere in the operator's configuration surface, which covers knowledge, identity, guardrails, testing and deployment. Sourcehelpscout.com/company/legal/ai-transparencyread 2026-09-05
APIs, SDKs & MCP Extensibility	Full
A documented developer platform for Help Scout's own product: the Mailbox API v2 with read and write access and per-plan rate limits of 200, 400 and 800 calls a minute, the Docs API, an Apps framework rendering inside the conversation sidebar, a JavaScript SDK, a React UI Kit matching the product's own interface, webhooks and OAuth apps. Help Scout also runs its own MCP server, letting an external assistant reach conversations, customer records, mailbox data and Docs content, read-only for new connections and scoped to the permissions of the user who connects. Sourcedeveloper.helpscout.comread 2026-09-05
Testing, Debugging & Optimization	Full
A Test tab lets the operator chat with the agent and simulate the conversations customers may have before a change reaches production, Preview covers identity edits, and Improvements submitted from those tests feed straight into live AI Answers behavior; test conversations are excluded from billed resolutions, so nothing discourages running them. The loop closes after deployment on the Insights breakdown of resolved, unhelpful and escalated conversations, which measures the agent rather than the content. There is no scored test suite or automated evaluation harness, so the verdict on any single test is the operator's own reading. Sourcedocs.helpscout.com/article/1757-ai-agents-in-help-scoutread 2026-09-05
Browser & Computer Use	Not documented
The agent answers from synced knowledge sources inside the Beacon widget, and every route out to another system runs through a published API, app or webhook rather than by operating an interface. No browser control, hosted session, virtual desktop or computer use is documented. Sourcedocs.helpscout.com/article/1757-ai-agents-in-help-scoutread 2026-09-05

The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Pricing

From $25/user/mo

hybrid

Free tierTrial available

Included quota

Standard (entry paid, $25/user/mo annual): unlimited contacts, 2 shared inboxes, basic workflows, standard integrations, 2-year reporting; AI Assist included. AI Drafts/Summarize require Plus+. AI Answers (the chatbot) is NOT included - separate $0.75/resolution. Free: 5 users, 1 inbox, 1 Docs, 100 contacts/mo.

What is public

Help Scout runs two billing models in parallel. New signups get per-user pricing (the current default on the live pricing page): Free ($0, 5 users, 1 inbox, 1 Docs, 100 contacts/mo), Standard $25/user/mo, Plus $45/user/mo, Pro $75/user/mo (10-user min, sales demo) - annual saves ~15-20% - with AI Answers as a separate $0.75/resolution add-on. A legacy contact-based model ('contacts helped per month,' editions $0-$90, 'no per-seat or AI usage fees') is maintained for existing customers and still appears on some third-party listings (e.g. G2), which is the source of the apparent conflict.

Billing mechanics

Per-user model: pay per seat per month (billed in advance), with unlimited contacts on paid plans; AI Answers (the customer-facing chatbot) bills separately in arrears at $0.75 per AI-resolved conversation - one resolution max per conversation, charged only if the AI fully resolves without the customer escalating or asking for more help. New accounts get 3 months of unlimited AI Answers free; a monthly spending cap auto-disables AI when hit (predictable billing, but stops automation mid-peak). A parallel legacy contact-based model bills on contacts helped with no per-seat or AI usage fees; the two models coexist (new signups on per-user, existing customers grandfathered on contact-based) rather than one having replaced the other.

Cost watchouts

AI Answers add-on ($0.75/resolution - ~$375/mo at 500, ~$750 at 1,000), extra inboxes ($10-$12) and Docs sites ($20), the 26-agent cliff (forced Standard->Plus, ~87% jump), onboarding fees ($1,500-$5,000 reported), HIPAA/SSO gated to Plus/Pro, Pro requires a 10-user minimum + sales demo

Variable cost rationale

Under the per-user model, seat cost is predictable but AI Answers adds usage-based per-resolution spend (capped if you set a cap); the legacy contact-based model instead ties exposure to monthly contact volume

Additional watchouts

The published model is unsettled - confirm whether your account is per-user or contact-based before budgeting; AI Answers is a separate usage cost not in any base plan; the 26-agent forced upgrade is a sharp cost step

Overage / add-ons

AI Answers bills per resolution ($0.75) in arrears with an optional monthly spending cap (auto-disables AI at the cap). Extra inboxes (~$10-$12/mo each) and extra Docs sites (~$20/mo) bill as add-ons. Crossing 25 agents forces a Standard->Plus upgrade.

Sales call required

Mixed (some tiers require a call)

Free / trial

Free plan (5 users) + 15-day trial

Lowest paid plan

Standard $25/user/mo

Commercial notes

Email-first shared-inbox helpdesk; AI features run on GPT-4-class models; AI Answers is read-only/KB-grounded (can't check orders or take actions); Help Scout for Good (nonprofit discounts up to 100%) and a startup program (up to ~6 months free); 99.99% uptime

Key ambiguities

Help Scout runs per-user and contact-based billing in parallel - new signups see per-user (Standard $25/user + $0.75/res AI), while existing customers may be grandfathered on the legacy contact-based model ($0-$90 editions, 'no per-seat or AI usage fees'); some third-party listings (e.g. G2) still show the contact-based version, which is the main source of confusion.

Cancellation / refund

Monthly or annual (annual ~15-20% off; multi-year unlocks more); users added/removed within 24h aren't billed; AI Answers billed in arrears; alternate payment (ACH/wire) for invoices over $2,500 prepaid annually

Support SLA / resale

Pro adds a dedicated onboarding specialist, HIPAA, SSO, and a dedicated support relationship (10-user min); Plus/Pro add advanced reporting, AI Drafts/Summarize, and key integrations (Salesforce, Jira, HubSpot, WhatsApp inbound)

Missing data

Contact-based tier pricing above the $50/mo Standard base isn't published publicly (visible only to active account holders), so the full contact-based ladder can't be reconciled; exact annual-discount % and onboarding fees vary by source. Seed '$25/user +$0.75/res' matches the current per-user default for new signups.

Agentic Index verified 2026-09-05

Official pricing page Docs

Related vendors

Alternatives to Help Scout

The closest documented capability profiles to Help Scout among customer support agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.

LivePerson8.5 / 14Fuller documented coverage on Workflow Orchestration
Fini8.5 / 14Fuller documented coverage on Workflow Orchestration and Observability & Auditability
My AskAI6.5 / 14A lighter documented profile than Help Scout
Parahelp8.5 / 14Fuller documented coverage on Workflow Orchestration and Prebuilt Agents, Templates & Packs
Parloa7.5 / 14Fuller documented coverage on Workflow Orchestration
Tidio9.5 / 14Adds documented Deployment & Data Residency

Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded

Compare Help Scout with other vendors →Compare all customer support agents →See where Help Scout ranks →

Data confidence: high

Head to head

Help Scout vs Freshworks Help Scout vs Fin Help Scout vs Zendesk