Pricing

No per-user fees.
Pay only for AI work.

Bring your whole team in. Every plan has unlimited users. You buy credits for the work your AI agents do, and nothing else.

Starter

Everything a growing team needs to run great support.

$49/ month
Start free trial
  • Unlimited users, always
  • All agents, all work tiers
  • 3 mailboxes
  • Shared inbox, views, assignment
  • Knowledge base, unlimited articles
  • Automations and workflows
  • Basic reporting and data export
  • Email support
Most popular

Pro

Routing, SLAs, and deeper reporting to hit your targets.

$199/ month
Start free trial
  • Everything in Starter, plus:
  • 10 mailboxes
  • CSAT surveys
  • Round robin and load balancing
  • SLAs and escalation rules
  • Private and multilingual knowledge base
  • Custom actions and integrations
  • Custom reports
  • SSO

Scale

Advanced security, compliance, and control for larger teams.

$449/ month
Start free trial
  • Everything in Pro, plus:
  • Unlimited mailboxes
  • Skills-based routing
  • Custom roles and permissions
  • Real-time dashboard
  • Sandbox environment
  • Extended API limits
  • SSO/SAML and audit log
  • HIPAA

Enterprise

Volume pricing and tailored terms at the largest scale.

Custom
Talk to sales
  • Everything in Scale, plus:
  • Volume credit pricing
  • Annual credit pooling
  • SCIM provisioning
  • Custom retention and residency
  • Custom connectors
  • Uptime SLA
  • Dedicated support and migration

Estimate

What will this cost us?

Two questions: how many conversations you handle, and how hard you want the AI working. Company size never comes into it.

Who you support
How hard the AI works
%

Most queues, with a few (security reviews, escalations) still handled by hand.

5,456

Credits you'd use a month

AI works 800 of 1,000 conversations. The rest are handled without it.

3,600 Low actions3,600
200 Medium actions800
88 High actions1,056

A 7,000-credit block costs

Starter$399
Pro$499
Scale$649

Credits

What a credit buys

Credits are priced on complexity: how much context the AI has to gather and how many systems it touches, not on what the answer turns out to be.

Low
1credit

Context is already in the thread. No tools, no search, no skill.

  • Summarize this ticket
  • Draft a reply
  • Make it shorter and less formal
  • What haven't we answered yet?
  • Rewrite this in Spanish
Medium
4credits

One skill, one or two systems, a bounded search.

  • Is this contact usually this frustrated?
  • Have we seen this before?
  • Why were they charged $412 this month?
  • Is this our bug or their config?
  • File this as a Linear issue
High
12credits

Several skills chained, multiple systems, cross-account scans, and a judgment call at the end.

  • Account health before I reply
  • How many other accounts have this?
  • Do they qualify for a refund, and prepare it
  • Should I worry about this account before renewal?

Compare

Every plan, side by side

Everything in the AI block is identical across plans. Everything below it is what you're buying up into.

FeatureStarterProScaleEnterprise
From$49$199$449Custom
Credit blocks500 – 32,0002,000 – 32,0004,000 – 32,000Volume
UsersUnlimitedUnlimitedUnlimitedUnlimited
AI, identical on every plan
Assist
Support
Background agents
Low, Medium, High work
Skills, tools, guidance
Rollover, spend cap, auto top-up
Helpdesk
Mailboxes310UnlimitedUnlimited
Inbox, views, assignment, snooze
Knowledge basePublicPrivate, multilingualPrivate, multilingualPrivate, multilingual
Automations and workflows
CSAT–
Round robin, team office hours–
SLAs and escalation–
Skills-based routing––
Sandbox environment––
Admin and security
SSO–SSO/SAMLSSO/SAML
Audit log––
Custom roles––
HIPAA––
Data retention and residency––StandardCustom
Reporting, API and support
ReportingBasicCustom reportsReal-time dashboardCustom
API limitsStandardStandardExtendedCustom
SupportEmailPriorityPriorityDedicated

FAQ

Questions, answered

How credits work, what they cost, and what happens at the edges.

One credit is one simple AI action, like drafting a reply or summarizing a ticket. Work that takes more effort costs more: a lookup across one or two of your systems is 4 credits, and a full investigation that reads across everything and reaches a judgment is 12. The full table is on this page, and it does not change without notice.

Multiply your monthly conversations by about 7 for B2B support, or about 4 for B2C. A B2B team handling 1,000 conversations a month lands near 7,000 credits. The estimator above does this properly, including how much of your queue you want AI touching.

Yes, on every plan including the $49 one. Add your whole team, your engineers, your success managers, whoever needs to see a conversation. Nothing about your bill changes when you hire.

Because the plan is the helpdesk. Every plan includes a block of credits, and the price also covers the inbox, the knowledge base, the integrations, and the mailboxes, routing, security and admin features at that tier. Starter is the whole product for a small team. Pro and Scale exist for operations that need SSO, more mailboxes or audit logs, and those cost something to run whether or not the AI is busy.

Per-resolution pricing only charges when AI closes a ticket, which sounds fair until you notice it charges nothing for the AI that helps your team all day, and then sells that back as a separate per-agent add-on. Helply prices every kind of AI work in one currency. Drafting is cheap, deep investigation costs more, and it all draws from the same block you already bought.

The next block buys automatically at the same rate you are already paying, never a penalty rate, up to a spending cap you set yourself. You get a notice at 80%, at 95%, and at the cap. Turn auto top-up off and the AI simply pauses at the cap while your team keeps working. Nothing about the helpdesk stops.

Yes. Every account sets a monthly ceiling, and it defaults to your plan price, so nothing can surprise you unless you raise it yourself.

Unused credits carry into the next month and expire at the end of it. A quiet month funds a busy one. Annual plans pool credits across the whole year instead, which suits seasonal businesses.

Retries, actions that fail, drafts your team rejects before they reach a customer, spam and no-touch tickets, knowledge base indexing, and anything you do to set the AI up: writing skills, connecting tools, changing guidance. You pay for output, not for configuration.

A thread with a customer, however many messages it runs to and whichever channel it arrives on. Credits are counted per AI action rather than per conversation, so a long thread that AI touches once costs one action.

Move up at any time and pay the difference pro rata. Move down at your next renewal. Changing your credit block does not change your plan, and changing your plan does not change your credits.

Yes, and it works in favour of consumer teams. B2C conversations are mostly simple, so they run about 4 credits each against roughly 7 for B2B. Higher volume, lower cost per conversation.

Yes. Set a monthly allowance per team or per mailbox so one queue cannot spend the whole block.

The table is versioned and dated. If weights ever change, existing customers keep their current table until renewal and we tell you before it happens. We will not reprice work you have already built a workflow around.

Above 32,000 credits a month, which is roughly 4,500 B2B conversations or 9,000 B2C. Enterprise adds volume pricing, annual credit pooling, SCIM, custom data residency and a dedicated team.

14 days with 1,000 credits and no card. That is enough to put the AI on a real queue rather than a demo one.

Yes, at 15% off, with credits pooled across the year instead of monthly.