Chat Support Virtual Assistant Software: Which Fits a VA Business?

A VA business serving several client shops should start with Help Scout, a shared inbox that gives each client its own inbox from $25 a seat, and add live chat (Crisp or LiveChat) only for clients who need it. For a three-person team, AI reply drafts are worth paying for once they save about an hour a month.

Two things change the answer. The first is whose login owns the account: tools that charge per workspace (Crisp) suit clients who buy their own plan and invite you in, while per-seat tools suit a VA who runs everything under one roof. The second is which plan tier holds the AI drafts, because at Help Scout that is Plus, not Standard.

Follow me on Instagram@sagnikteaches

Decide who owns each client's account before comparing prices

Every price below changes once you settle one question: whose login owns the helpdesk? A VA business tends to drift into one of three set-ups without choosing, and each one favours different software.

Connect on LinkedInSagnik Bhattacharya

The second set-up is the safest for data, because each client's customer messages never sit in your system. It is also the messiest to work in: a VA covering five florists, bookshops and driving schools ends up with five tabs and five sets of notifications. For email-led support I would default to the first set-up and accept the export duty at the end. For live chat I would default to the second.

Subscribe on YouTube@sagnikteaches

Whichever you pick, put it in the service agreement: who owns the account, who pays, what happens to message history on the last day, and how quickly you will hand over an export. The contract wording matters more than the software when a client leaves on bad terms.

Six tools, priced from their own pricing pages

These are list prices read from each vendor's pricing page on 1 November 2026. Annual billing is cheaper almost everywhere, and prices move, so check the page before you buy.

The realistic starting points for a VA business are Help Scout for email-led work and Crisp or LiveChat for chat on a shop's site. Freshdesk is a fair alternative if a client already uses it. Intercom is worth knowing about because clients ask, and the comparison of Intercom's inbox and its Fin agent explains why it rarely suits a small shop.

Help Scout: the inbox-per-client sum for eight clients

Help Scout is the closest fit to how a VA actually works, because its price is a seat fee plus a small fee per extra inbox. Standard includes two inboxes, Plus includes five and Pro ten. Extra inboxes cost $10 a month paid annually. Light users, who can read and collaborate without taking a full seat, come with Plus (five of them) and Pro (fifteen), which is handy for letting a client's owner look at their own conversations.

Take a VA business with three agents and eight client shops, say two florists, two independent bookshops, two photography studios and two driving schools. Everyone shares inboxes, and each client gets one inbox.

• Standard: three seats at $25 is $75. Eight inboxes minus the two included leaves six at $10, which is $60. Total $135 a month.
• Plus: three seats at $45 is $135. Eight inboxes minus the five included leaves three at $10, which is $30. Total $165 a month.

So at eight clients the step up to Plus costs $30 a month more across the whole team, and it brings unlimited AI Drafts, round-robin assignment and internal teams. The gap depends on how many clients you have, which the next table shows for three agents. Seat prices are as Help Scout lists them (its page offers a discount for annual billing without labelling which rate the listed prices are, so confirm at checkout), and extra inboxes are at the annual-billing $10.

Once you pass five clients the premium for Plus stays flat at $30 a month for three seats, because the extra inboxes you would have paid for on Standard come bundled. Spread over eight clients that is under $4 a client, and the whole inbox bill for the team works out at about $17 a client on Standard and $21 on Plus.

The free plan deserves a mention. It allows five users and one inbox, which is enough to run a single new client through a trial without paying anything, and Help Scout offers a 15-day free trial of Standard or Plus. Pro needs a minimum of ten users and a sales conversation, so it only matters if you grow past that.

Front: why the cheapest plan only fits one kind of client

Front looks the same at first glance, with a $25 seat on Starter, but Starter is sold for single-channel support: a shared inbox and ticketing for one channel type, with basic automation. Professional at $65 a seat adds email, SMS and social channels together, and goes up to 50 seats. Front's Copilot assistant is a $20 a seat add-on on Starter and Professional, and its Autopilot agent starts at $0.05 a conversation.

For a VA business that matters as soon as a client's customers write by more than one route. Imagine the same four client types:

• The independent bookshop only gets email, so Starter fits.
• The florist takes questions by email and Instagram messages, so it needs Professional.
• The photography studio's print shop adds SMS reminders for collection, so Professional again.
• The driving school handles booking requests by email and text, so Professional.

Three agents on Professional cost $195 a month, and $255 with Copilot for all three. That is clearly dearer than Help Scout at the same team size, so Front earns its place only when a client genuinely needs several channels in one inbox or wants Front's collaboration features. For a VA whose clients are mostly email and a contact form, Help Scout is the cheaper way to get the same shared-inbox behaviour.

Live chat on a client's site: Crisp and LiveChat's Text plans

Chat widgets answer different questions from email: stock, delivery cut-offs, whether the studio has a Saturday slot. They work only if someone answers within a minute or two during stated hours. If the client cannot promise that, an email-style inbox with an out-of-hours message beats a chat bubble that goes unanswered.

Crisp charges per workspace rather than per seat, which suits the client-owns-the-account set-up. The free workspace has two seats and a cap of 100 customer profiles. Mini at $45 gives four seats, 5,000 profiles and $5 of AI credits, which Crisp translates into around 90 automated conversations. Essentials at $95 gives ten seats, 50,000 profiles, two sub-inboxes and $25 of credits (around 450 conversations). Plus at $295 gives twenty or more seats, extra agents at $10, and $75 of credits.

A plausible spread across four clients: the florist stays on Free until the 100-profile cap bites (a few chats a week gets there inside a season), the independent bookshop sits on Mini at $45, the photography studio on Mini at $45, and the driving school, with several instructors checking messages, on Essentials at $95. That is $185 a month in total, paid by four different clients. None of it touches your own subscription costs.

LiveChat, which now sells new customers its Text plans, bills per user: Essential is $19 a user a month on annual billing ($25 monthly) and Growth is $79 ($99 monthly). Essential includes one AI Agent, 10 AI resolutions a month, and 60 days of conversation history. Growth includes ten AI Agents, 200 resolutions and unlimited history. Extra AI resolutions come in packs of 50 at $49.50. There is no minimum team size, and every person who answers chats needs a seat, so three VAs covering one client's chat cost $57 a month on Essential. Ask the vendor whether one account can serve several client sites before assuming it can.

Is AI reply drafting worth $20 a seat? The break-even sum

At Help Scout, the AI Drafts feature (a reply written from past conversations and help content) starts on Plus, which is $20 a seat above Standard. The honest test is whether it saves more minutes than it costs, and the sum is short.

Say a VA's time is worth $30 an hour to the business, which is 50 cents a minute. For a three-seat team moving from Standard to Plus at eight clients, the extra cost is $30 a month. Break-even is $30 divided by $0.50, which is 60 minutes saved across the whole team in a month. If the team answers 900 messages a month, that is four seconds per message. Drafts only need to be usable on a small share of messages to clear that bar.

The picture differs with two clients, where the premium is $60 and break-even is 120 minutes a month. Scenarios for the eight-client team, all illustrative:

The poor-fit row is real: if a client's messages are all one-off and personal, such as bespoke photography enquiries, drafts built from past conversations have little to copy. A client whose messages repeat, such as the bookshop's "where is my order" and "do you stock this title", is where drafts pay. The rules for AI inbox drafts and handover covers how to set the review step so a saved minute does not become a sent mistake.

A draft that looked right and was wrong

This is an illustrative case from the bookshop's inbox, to show what you check before sending. The customer wrote:

Hi, I ordered a signed copy last Tuesday and it still hasn't arrived.
Can you tell me where it is? Order number is BK-20418.
Thanks, [first name]

The AI draft, built from earlier replies in that inbox, came back as:

Hi [first name],

Sorry about the wait. Signed copies ship within 2 working days, so yours
should have been with you by now. I'll arrange a replacement today and
refund the postage.

Best wishes,
[your name]

Three problems. The 2-working-day line came from a different product (unsigned stock), whereas signed copies wait for the author's visit. The draft also promised a replacement and a postage refund that the client's written policy does not offer before ten working days. And it never looked up the order, so it cannot say where the parcel is. The VA's fix is to check the order, say the signed copy is waiting for the signing session with the real date, apologise, and leave out the promises. The saved minute became a three-minute edit, which is still fine, but a rushed send would have created a refund the shop never agreed to.

The lesson is a rule you can put in every client's saved procedure: refunds, replacements, delivery dates and discounts in a draft must match the client's written policy, or they come out before sending.

Letting AI answer alone: Envoy, Fin and Freddy per resolution

Drafts keep a person in the loop. The other pattern is an agent that answers customers directly and charges per resolved question. Help Scout's Envoy is $0.75 a resolution, with a three-month free trial and a monthly cap you can set. Intercom's Fin is $0.99 an outcome. Freshdesk's Freddy AI Agent gives 500 free sessions once per account, then $49 per 100. Crisp and LiveChat include AI credits or resolutions inside their plans.

The sum is the same shape as the drafts one, with a catch about definitions. Help Scout counts a resolution only when the customer gets an AI answer and does not escalate, search the help centre, ask another question or indicate they need more help. Other vendors count differently, and some treat silence as success, so read each definition before you compare prices.

Take the florist, who gets 260 repeat questions a month (opening hours, delivery areas, last order for Mother's Day). Suppose the agent resolves 60%, which is 156 questions. At $0.75 each that is $117. The VA time those questions would have used, at 1.5 minutes each, is 234 minutes, or 3.9 hours, which at $30 an hour is also $117. The cost is a tie. Per-resolution AI pays for itself only when it beats about 90 seconds of human time, or when it answers messages that would otherwise wait until morning, which is a sales argument and not a cost one. Set the monthly cap on day one, and treat the first month as a test on one client.

Keeping client A's data out of client B's drafts

If a drafting feature learns from past conversations, you need to know where its memory stops. Does it read only the inbox it sits in, or every inbox in the account? Vendors describe this differently and the answer can change with an update, so test it yourself. The confidentiality checklist for VAs using AI has the wider policy, and this is the practical check:

1. In client A's inbox, send yourself a test message from a spare address containing an invented rule, such as "Our Zebra-Lantern returns window is 45 days."
2. Reply to it as an agent, repeating the rule in your answer, so it exists in A's conversation history.
3. In client B's inbox, send a similar question ("What is your returns window?") and ask for an AI draft.
4. Read the draft. If it mentions the Zebra-Lantern rule or 45 days, drafting is reading across inboxes.
5. Repeat the next day, because some tools index history on a delay.

If it leaks, you have two safe options: move each client to its own account (the second set-up), or switch drafting off for clients in the same inbox pool. Either costs less than explaining to a bookshop why its refund policy appeared in a florist's reply. Note the date of the test in your records.

A scoring grid for your own shortlist

Scores beat opinions when you are comparing three tools late in the evening. Weight what matters to a VA business, score each tool 1 to 5 from your own trial, and multiply. This example is filled in for the eight-client team, with illustrative scores:

Crisp finishes four points ahead of Help Scout, and all of its lead comes from two rows: keeping clients separate and exporting on a client's last day. Those rows are the account-ownership question from the start, not a feature gap. If most clients will pay for and own their tools, Crisp wins. If you run everything under your own account, Help Scout's lower total cost and better drafts carry it, and a one-point change in either tool's separation score would swap the order. Front only takes the lead when a client needs several channels in one inbox.

A two-week trial that settles it

Help Scout offers 15 days on Standard or Plus, LiveChat 14 days on Essential or Growth without a card, and Crisp 14 days with full access to every feature. That is enough to run one real client and one dummy client side by side. A schedule that works:

• Days 1 to 3: set up one inbox per client, with saved replies, signature and a short help article for the two most common questions.
• Days 4 to 8: run your real messages through it. Log, for each draft, whether you sent it as it was, edited it or discarded it.
• Day 9: run the canary test above.
• Days 10 to 12: test the exit: export one inbox's history and see how readable the file is.
• Days 13 and 14: total the minutes saved from the log, compare with the break-even figure for your seat count, and decide.

Keep the log honest. A draft you edited for three minutes did not save a minute, and the log is what tells you whether to pay for the AI tier or switch it off. For the same reason, compare vendors on what happens to your sums with the free allowances gone: the limits on free AI helpdesks for small teams shows where free plans stop. If your clients sell mostly through email and a contact form, the approach to AI triage for shared inboxes may remove the need for chat entirely, and the comparison of chatbots, live chat and help desks will help you explain the choice to a client in plain terms.

Questions VAs ask before picking a support tool

Should the client or the VA pay for the helpdesk?

Default to the client paying when the tool bills per workspace, such as Crisp, because they own the data and the contract. Pay yourself when you run several clients through one shared inbox, then pass the cost through in the retainer at cost plus a small margin. Write down who owns the account and the exports in your service agreement.

Is a free plan enough for one small client?

Often for the first few months. Help Scout's free plan allows five users and one inbox, and Crisp's free workspace gives two seats and 100 customer profiles. The limits that bite are AI features, which are mostly paid, and Crisp's profile cap, which a busy shop passes quickly.

Do I need live chat if clients mostly get email?

No. A shared inbox with saved replies handles email-led support more cheaply. Add live chat only for a client whose customers ask short questions while browsing the site, such as delivery cut-offs or stock, and where someone can answer within a minute or two during set hours.

How do I stop an AI draft promising something the client does not offer?

Treat every draft as unsent until a person has read it, keep each client's policies in its own help content, and add a checklist line: refunds, delivery dates and discounts must match the client's written policy. Run the canary test in this tutorial to confirm drafts never borrow another client's rules.

Further reads

• AI Tools Used by Virtual Assistants: A Stack for Five Clients — The wider tool stack a VA runs across five clients.
• Virtual Assistant Pricing When AI Does the Drafting: Rates and Packages — Turn the tool cost into retainer and package prices.
• AI Virtual Assistant Services: How a VA Business Can Package Them — Package AI-assisted support as a service you can sell.
• ChatGPT Prompts for Virtual Assistants: Inbox, Scheduling, Reports — Prompts for inbox work when drafts are not built in.
• How Virtual Assistants Use AI to Manage More Clients — How AI lets one VA carry more client accounts.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: the pricing pages of Help Scout, Front, Crisp, LiveChat (Text plans), Freshdesk and Intercom, read on 1 November 2026.