ChatAgentLyAISOLUTIONS™
Frequently asked questions

The questions people ask before they sign up

Short answers, in plain terms. If you are already set up and want the detail behind any of this, the documentation covers every feature in full.

Getting started

What it takes to go from signing up to answering a real customer question.

Do I need a developer to set up ChatAgentLy?

No. Setting up ChatAgentLy means uploading documents you already have, copying one script tag, and pasting it into your site before the closing </body> tag. There is nothing to build, train or configure first, and most businesses are answering live questions within an hour. The quick start walks through it in four steps.

Do I need a website to use ChatAgentLy?

No. If you have no site, ChatAgentLy gives you a shareable link and a downloadable QR code that open a full-page version of the same assistant, with the same answers and the same citations. A restaurant can put the QR code on a table card.

How long does setup take?

Most businesses are live within an hour, and the bulk of that is deciding which documents to upload rather than anything technical. Adding a source and embedding it takes minutes; pasting the script tag takes seconds; and appearance changes apply to every page immediately, with nothing to re-embed.

What it can read

The formats and sources an assistant can answer from.

What file types can I upload to ChatAgentLy?

PDF, Word (.docx), CSV, plain text and Markdown. Images work too: JPG, PNG, WebP and GIF have their text read out automatically, so a photographed menu or a scanned policy becomes a source without anyone retyping it. HEIC, the iPhone default, is refused at upload rather than accepted and failed later, because the model cannot read it.

Can ChatAgentLy answer from my live database?

Not yet. Live database connections are in development and are not available to customers today. When they arrive, Postgres, MySQL, SQL Server and Oracle will connect over a read-only connection string, and nothing will be copied out: a question will trigger a sandboxed read-only query against your real data at the moment it is asked, so figures like stock levels and prices stay current rather than being a snapshot. Until then the assistant answers from files, web pages, written questions and answers, and pasted text.

Can it read my website instead of files?

Yes. Give ChatAgentLy a single page or your whole site. A site-wide scan finds pages through your sitemap where you have one and by following links where you do not, and it respects robots.txt. Each page it finds becomes its own citable source.

What if a scan cannot read my site?

It tells you which of four things happened, because they need different answers. A site behind bot protection answers every request with a security check, so uploading the content directly is the way through. A site behind a sign-in redirects us to a login page, and being signed in yourself does not help, because the scan runs on our servers rather than in your browser. A robots.txt containing Disallow: / asks automated readers to stay away from the whole site, which is common on staging and preview deployments, and we respect it. And some sites simply publish no page index. In every case you can still add individual addresses by pasting them, because that is you telling us which pages to read rather than us going looking.

How many sources can I add?

Your plan sets how many, and the plan comparison lists the number for each one. Three things are worth knowing about how they are counted. Every page found by a site-wide scan is its own source, so scanning a large site uses more of the allowance than adding a single page. A source that failed to process does not count against it. A source you have disabled does still count, because it is stored and ready to switch back on. The billing page in your dashboard shows how much of your allowance is in use, and moving up a plan raises the limit immediately. Questions per month and storage work the same way.

How long does it take before a source can be answered from?

Adding a source does not hold you up: it is read, split and indexed in the background while you carry on, and each source shows as Processing and then Ready on the Sources page. A single page or file is the quick case. A whole-site scan takes as long as the number of pages it discovers, so a large site is longer than a small one, and each page becomes its own source as it completes rather than everything appearing at the end. Anything that fails says why on the source itself rather than disappearing.

Answer quality

What the assistant does with your content, and what it will not do.

Which AI model does ChatAgentLy use?

Answers are generated by large language models supplied by third parties, as set out in the Terms of Service, and your content is processed by those providers only to produce the answer. Which model is in use is an operational choice that changes as better ones appear, so it is not published here. What does not change is where answers come from: only the content you added, with a citation on each claim, and the model is never asked to answer from its own general knowledge. See the Privacy Policy for how content reaches a sub-processor.

How does ChatAgentLy stay accurate?

Every answer is built only from the content you added, and every claim carries a citation back to the source it came from, so a reader can check it. The wording shown when nothing matches is yours to set. Questions the assistant could not answer are collected and grouped by topic, so you can see what to add next.

Can I control how the assistant answers?

Yes. You can set how strictly it sticks to your source material, how much it retrieves before answering, whether it remembers earlier questions in the same conversation, and the exact wording it uses when it has no answer. Every one of these has a sensible default, so changing nothing is a valid choice. Which of them you can adjust depends on your plan.

What languages does ChatAgentLy support?

On plans that include it, the assistant replies in the language the visitor's own browser is set to, without you configuring anything or translating your documents. Your source material can stay in one language.

Orders and your POS

Where an order goes once a customer places it, and what you have to set up.

Can I see what I took in this month, or find one order?

Yes. The Orders screen opens on the current month, with arrows and month and year dropdowns to move around, This month and Last month in one press, and a date range. A search finds an order by name, phone or email. The totals shown are for the whole period you selected, counted in the database rather than from the rows on screen, and they leave out cancelled orders.

Will my kitchen have to re-type orders?

Not if your POS is connected. Square connects directly: press Connect Square on the Payments screen and each order appears in Square's own Orders queue and kitchen display, so your existing printer routing handles it. Toast, Clover, Lightspeed and most others are reached through a connector service, Deliverect or Checkmate, and the order lands in the POS the same way.

If you connect nothing, orders still arrive as an email ticket and on the Orders screen, which many kitchens run on a tablet. Worth knowing: that route does not put the order in your POS, so your end of day totals will not include it.

I use Toast. What do I have to do?

Toast allows order injection only through approved partners, so nobody reaches it with a switch. You reach it through a connector your Toast already trusts: Deliverect or Checkmate.

If you already pay for one, link ChatAgentLy as a channel in their console and paste the one value they give you into the Payments screen. If you use neither, the email ticket and the Orders screen work from day one, and our webhook can point at any connector you sign up for later. Tell us which you use and we will point you at the shortest path.

Do I need Stripe to take orders?

Only if you want to be paid before pickup. Pay now charges the card as the customer orders, through your own Stripe account, so the kitchen only ever sees paid orders and no-shows stop costing you food. Pay at pickup takes no card at all and needs no Stripe, exactly like a phone order.

You choose on the Payments screen and can change your mind whenever you like.

If I connect more than one destination, will orders arrive twice?

Each destination gets its own copy on purpose, because most restaurants want the kitchen ticket and the POS.

The one combination to avoid is two routes into the same POS: connecting Square directly while also running a connector that already feeds Square would print the order twice. Pick one route per POS, and use the others for people rather than systems.

What does ChatAgentLy charge on an order?

Whatever your agreement says, and your Payments screen lists it in full. A fee can be a percentage of the order, a flat amount per order, both, or nothing at all. A flat monthly amount is also possible, billed with your subscription rather than taken from an order.

The part worth knowing is who pays it. On one arrangement the fee is added to your customer's bill as its own line and you receive the food money whole; on the other you absorb it and your customers never see a fee. Your plan sets the default and we can agree something different for you.

How do I stop taking orders when the kitchen is slammed?

Open Online Ordering, Orders and press Pause online ordering. Visitors can still read the menu, and nothing can be ordered until you press Accept orders again.

Hide menu goes further and removes ordering from the widget entirely, for a seasonal close or a menu being rebuilt. Both take effect immediately.

Booking a time

How a visitor books an appointment with you, and what you set up first.

How does a visitor book an appointment?

A Book a time tab sits beside Question in the same chat, on your site, on your shareable link and in the Playground. The visitor picks a department, then a day from a month calendar that only offers days with something free, then a time, and leaves a name and the contact details you require. The time is checked again at the moment of booking, so two people cannot take the same slot.

What do I have to set up before anyone can book?

Under Scheduling, add at least one department and set the hours it takes bookings on, per weekday. A department with no hours offers no times, and the screen says so. Optionally set how long each appointment is, how many people you can see at once, how much notice you need, how far ahead you take bookings in days or months, and where to email or text each booking. The documentation walks through it.

Does it work with my calendar?

Every booking emails a standard calendar invitation to the department and to the visitor, which Google Calendar, Outlook, Apple Calendar and every other calendar accept without an app or a connection. If you move a booking, an updated invitation goes out that replaces the one already in the calendar rather than adding a second event. It does not yet read your calendar to block times you have booked elsewhere.

Can I move or cancel a booking, and leave notes for my team?

Yes. Each booking on the Scheduling screen can be marked done, no show or cancelled, moved to another open time with the visitor told automatically, and annotated with internal notes that are stamped with who wrote them and when. Notes are for your team and never shown to the visitor. The list can be searched by name, phone or email and filtered to one department.

Which details must a visitor give to book or to order?

That is your decision, set separately for bookings and for orders. A name is always required. Phone and email can each be required or optional, with one of the two always required so there is a way to reach the person. A kitchen usually wants a phone number; a clinic usually wants an email for the calendar invitation.

Plans and billing

What things cost, and how changing your mind works.

What does ChatAgentLy cost?

Pricing, the allowances on each plan and which features each one includes are all on the plan comparison, which reads from the live catalog. They are not repeated here on purpose: a page that copies prices is wrong from the next change onward, and quietly.

How many people can I add to my account?

Your plan sets the number of team members, and plan comparison lists it for each one. Pending invitations count against the limit, and the Team page in your dashboard shows how many seats are in use. Moving up a plan raises the limit immediately.

Is there a free plan or a trial?

There is a free plan, and paid plans may carry a free trial. Whether a given plan has one, and for how long, is shown on the plan comparison. During a trial you get that plan's full feature set at reduced usage volumes, and your card is not charged until it ends. The checkout page tells you the date of the first charge and the amount before you start, and we email you again before that date arrives, so a trial never turns into a charge you were not expecting. Each account gets one trial, so cancelling and resubscribing does not start another.

Can I cancel at any time?

Yes, from the Billing page in your dashboard, with no phone call and no notice period. Cancelling takes effect at the end of the period you have already paid for rather than immediately, and the account then moves to the free plan rather than closing.

What happens if I go over my limits?

Sources, storage and team seats are hard limits: you are told when you try to add one past the limit, and nothing already there is removed. The monthly question allowance simply stops answering new questions until it resets, so a busy month never produces a surprise charge.

Data and privacy

Where your content lives and who can reach it.

Is my data private?

Yes. Uploaded files go to private storage and are never given a public URL. Each organization's content is separated at every layer, so no other customer can reach it, and your documents are not used to train anything. Database connections, when they arrive, will be read-only with their credentials encrypted.

What happens to my data if I cancel?

The content the assistant answers from is removed. Deleting the account is the full exit: it cancels the subscription, removes your sources, database connections and API keys, and removes everyone from the account.

Who on my team can change things?

An Admin can change anything, including who else is on the team. A Member sees every screen and changes nothing, which is what lets you give somebody the numbers without giving them control over the answers your customers receive. Billing and account deletion are restricted further still, to the organization owner.

Something not answered here?

Contact us