AI agency or a ready sales system: five questions to ask before you sign

What are you actually looking for when you type "AI agency" or "automation agency" into Google? Almost nobody is looking for someone to build them a bot. They are looking for a way to make sure a customer gets an answer at ten at night, the enquiry from the ad does not get lost, and the appointment gets booked without someone chasing the customer for three days. The distinction sounds like semantics, but it decides what you get on the day after launch. An agency usually sells a project: discovery, build, handover. A ready sales system sells a result that runs every day. This article explains the difference without running anyone down, and gives you five questions to ask before you sign any proposal, ours included.

What you usually get when you work with an AI agency

The familiar model looks like this: a discovery meeting, a requirements document, a build on an automation platform or in code, a round of testing, a handover meeting with documentation, and a warranty period of a few weeks. It is a perfectly legitimate model, and it works well when the process is unique and stable, and when the business has someone who understands the system and can keep it running after the vendor is done. We wrote earlier about how to choose an AI consultant and what to check, and the central point there was the same one: the question is not who will build it, but who will stay.

The trouble starts because most small and mid-sized businesses in Israel do not have that person. The owner of a clinic, a law office or a moving company does not hire an automation specialist to maintain a bot. Then month two arrives. The price list changes, a service is added, the opening hours are different over the holidays, a WhatsApp message template is rejected, a form on the website gets a new field and an automation that depended on the old field name quietly stops working without anyone noticing. In a project, each of these is a new change request, a new quote, or a freelancer who built the system and moved on to the next client long ago.

Nobody here is a villain. It is simply the shape of the contract: a project is priced by the build, so it ends when the build ends. What a business needs is the operation, and the operation starts exactly there.

Project versus a running system: the difference starts the day after launch

In a running system, the vendor is paid monthly for the thing working, so the incentives line up on their own. If the agent stops answering, the vendor is the one who loses. If customers ask a question the agent cannot answer, there is someone to turn to, and that request is part of the service rather than a new project.

This is how it works with us, without the numbers that change from time to time:

• Fixing faults is free and unlimited. An agent that does not reply, an enquiry that was not saved, a calendar that does not sync - that is simply the product you are already paying for.
• Whatever you configure yourself in the interface is unlimited. Automations, follow-ups, fields, tables and views, and the number of times they run too. There are no trigger meters.
• A change to the agent itself is included in the subscription. Swapping the price list, rewording an answer, a new document in the knowledge base. The number of monthly changes included depends on the plan.
• What is genuinely new is quoted before work starts. A connection to an external system, an extra channel, a brand-new process. Never a surprise invoice.

Two more rules set your risk at the start. The setup fee is paid in one payment before work begins, and it is non-refundable. The monthly subscription starts on the day the agent goes live, not on the day you pay, so the first month you pay for is a month of a product that already works. The setup time is counted from the moment we have received all your materials. The full boundary between what is included and what is not is on the page what setup and support include, and we recommend reading it before any sales call, even if you end up not choosing us.

Five questions to ask before you sign

These questions work with any vendor: an agency, a freelancer, a software company, and us.

1. What is the end KPI, and how will we measure it in the first month?

"Implementing AI" is not a goal. A goal is something you can count: the time to the first reply, the share of enquiries that close with a complete card holding the name, phone, service and next step, the number of appointments booked out of enquiries, the share of out-of-hours enquiries that got an answer. If the proposal has no number you can check after a month, you are buying activity, not a result.

2. Which actions exactly will the AI take, and when?

"Answers questions" is not an action. Actions are: replies within seconds, asks the qualifying questions, books an appointment in the calendar, sends a reminder, sends a photo, a video or a document, sends a payment link, and hands a hot lead to a person with a summary of the conversation. Every action needs a defined moment, for example "when the customer asks about price, the agent gives a range and offers a consultation". It is just as important to ask what the agent does not do and when it hands over to a person. A vendor who answers "everything" has not thought about it yet.

3. Who maintains it after launch, and what does that cost?

Ask for the boundary in writing: what counts as a fault, what counts as a change, and what counts as development. Ask who answers when something breaks on a Friday afternoon. Ask whether automation runs are metered and billed, because on many platforms the bill grows precisely when the business does well.

4. Where does the data live, and who owns it?

Conversations, leads, phone numbers. Which account are they stored in? Can you export them when you leave? Are they used to train models? Under the Israeli Privacy Protection Law, when the agent answers on your channel you are the database owner and the vendor is a data processor, and an answer that does not fit that is a warning sign. With us, the data on a customer's channel belongs to the customer, it can be exported when the engagement ends, and we do not train models on identifiable data.

5. What happens when I change a price?

This is the question that reveals the most. Prices change, services are added, the holidays come every year. In a project the answer is usually a ticket, a quote and a wait. In a running system the answer should be one sentence: you send a request, and it is part of the monthly changes. And note the other side too: an agent that has not been given the new price list will keep quoting the old one, so the vendor should tell you exactly how and when updates are made.

Comparison: a typical agency project versus a running sales system

The last row matters, because it prevents the wrong conclusion. A ready system does not fit everything, and a project is not always unnecessary. The only question is what foundation you build on.

When you really need a project, and how we do it

Some businesses do not fit in a box, and it is better to say so early. Several branches or brands, each with its own sales and booking process. An internal system the agent has to write into directly. A call centre or sales team where conversations have to be routed by your own rules. A very high volume of enquiries. Data requirements such as isolated storage, restricted permissions and action logging. In these cases you need a project, and we do it.

Our process has four stages: a discovery call, architecture, a statement of work, and a fixed project price. When the scope evolves as the work goes on, we can also work by the hour. The essential difference is that the project is built on top of a system that is already running: the agents, the CRM, the calendars and the channels already exist and are maintained, and the custom development adds what is missing. When the project is delivered, the custom part lives inside a product someone is responsible for every day, not in the folder of someone who has finished. All the details are on the custom AI solutions development page, and if you are torn between a Low-Code platform and full development, the comparison is in the article when Low-Code holds you back and when you need a custom solution.

And honesty works in the other direction too. If none of the conditions above describes you, a ready plan will be cheaper and faster than any project. The differences between the plans are in capabilities: Start is a website widget or Telegram, with no WhatsApp and no appointment booking. WhatsApp, appointment booking and media sending start on Standard. Instagram and Facebook, in-chat payments, several agents and several calendars are on Business. Prices and quotas are updated from time to time, so they live only in the full plan comparison.

What WhaleBiz does today, and what is still vision

WhaleBiz is not an agency. We are an AI startup, a technology company that wrote and runs its own sales system, and that also does the project work an agency does, built on that product. Someone who searched for an "automation agency" and found us will get what they were looking for, with a product running underneath.

What the system does today: AI agents on WhatsApp, the website, Instagram, Telegram and Facebook that answer in Hebrew, Russian and English, qualify enquiries, book appointments in the calendar, send photos, videos and documents, send a payment link, and record every conversation in the built-in WhaleBiz CRM. When a person is needed, the agent hands the conversation to your team with the full context, and the team takes it from there. Around the core we build connections to the client's systems, custom projects, websites built as a sales tool with the agents inside, and campaign management on Google and Meta. Organic search promotion is something we do not sell.

What is still vision: a system that runs more links of the sales chain by itself, such as negotiation or logistics. We are working in that direction, but today it is not a product, and we will not sell it to you as if it already exists. That is exactly why the questions above include "what does the agent not do".

Frequently Asked Questions

What is the difference between an AI agency and a company that sells a ready system?

An agency usually sells a project: discovery, build and handover, and responsibility after launch passes to the client or to a separate retainer. A company that sells a running system is paid monthly for the system working, so maintenance, fault fixing and routine changes are part of the service. Both models are legitimate. The practical difference shows up in month two, when something has to change.

Is WhaleBiz an automation agency?

Not in the usual sense. WhaleBiz is an AI startup that wrote and runs its own sales system, with AI agents and a built-in CRM. We also do project work, such as connections to other systems, custom development and websites, but always on top of that system, so what is built keeps being maintained after delivery.

Can I start on a ready plan and add custom development later?

Yes, and that is the route we recommend for most businesses. You start on the plan that covers the channels and actions you need today, see what happens with real enquiries, and only then decide whether you need a connection to an internal system or a unique process. Development like that is quoted and approved before work starts.

What happens to my data if I stop working with a vendor?

That depends on the vendor, so it is worth asking before you sign. With us, when the agent answers on your channel, you are the database owner and we are the data processor. When the engagement ends you can export the data, and we do not train models on identifiable data. The full details are in our privacy policy.

How much does a custom project cost?

The price is set after a discovery call and the architecture design, and it is written as a fixed project price in the statement of work, or as hourly work when the scope evolves along the way. Before you get to a project, it is worth checking whether one of the ready plans already covers what you need, because a ready plan is almost always cheaper and faster.