I Asked Will&Be to Help Run a Small Business. We Need to Talk About the Website.

This AI workspace wants to bring your website, marketing, invoices, and customer support under one roof. Our hands-on test found a promising idea and a few loose floorboards.

You have a business idea. A good one.

Maybe you review websites for local businesses. Maybe you sell handmade furniture. Maybe you’ve finally decided that the thing everyone asks you to do for free should have a price tag.

Then comes the other job: running the business.

You need a website. Something to post on LinkedIn. A way to track customers. Invoices that add up. Answers to the same questions people keep emailing you.

Suddenly, your exciting new venture involves twelve browser tabs and a deeply personal relationship with a spreadsheet.

Will&Be wants to help with that. It presents itself as an "AI co-founder” for solo business owners, combining finance, marketing, legal-document tools, customer support, and more in one workspace.

An ambitious pitch. But can it actually help you get something useful done?

We created an account, connected Gemini, and tried it with a fictional business. The results ranged from "okay, that’s useful” to "why is my website displaying the instructions I gave you?”

Let’s unpack it.

The problem: a business idea isn’t a working business

For our test, we imagined Sample Studio, a small Berlin business offering a 90-minute website audit for €149.

The service was simple: review a customer’s website and suggest improvements. No mysterious growth formulas. No promises to turn someone’s plumbing company into the next Apple.

We wanted a small set of practical things:

• A one-page website explaining the service.
• A LinkedIn campaign introducing it.
• A helpful reply to a potential customer.
• A way to review that reply before it went anywhere.

This is where an all-in-one tool could earn its place. Individually, none of these tasks sounds enormous. Together, they eat the afternoon.

The appeal is having your business information travel between tasks. You explain what you sell once, then use that information in your website, marketing, and customer replies.

Think of a restaurant: the waiter, chef, and cashier should all be working from the same menu. You shouldn’t have to explain the soup three times.

What is Will&Be?

Will&Be is a business workspace with AI features built into different sections.

Its main dashboard, called the Cockpit, links to finance, marketing, branding, CRM, support, operations, strategy, analytics, and approvals.

A CRM is simply a place to keep customer information and track relationships. The approval area is a waiting room for proposed actions, such as sending a support reply.

The tool has a particular focus on solo founders in German-speaking markets. Its onboarding includes German business topics, and its finance and legal sections advertise features intended for that audience.

That focus gives the product a clearer purpose. A small business in Berlin has different administrative needs from a creator selling digital downloads in another country.

However, the presence of a legal-document generator doesn’t establish that its output meets your particular obligations. We inspected the interface; we did not validate its legal or tax compliance.

Where Gemini comes in

Will&Be asks you to connect your own AI provider key. Options shown during testing included Gemini, OpenAI, Claude, and Grok.

An API key is a digital access pass. It allows another application to request work from your AI provider account.

Here’s the simple version:

You give Will&Be a task → Will&Be asks the AI provider for help → the result appears in the workspace.

Imagine Will&Be as a restaurant and Gemini as a supplier. The restaurant organizes the experience, but it still depends on the supplier delivering what it needs.

That dependency became rather noticeable during our test.

First stop: build the website

The website workflow starts sensibly.

You enter a name, describe what you want, add optional style instructions, and choose your pages. You can also choose AI images, stock photos, or no images.

We kept things modest: one page, no images, fewer than 250 words.

Our brief asked for a clear explanation of the €149 audit, three deliverables, and a button inviting visitors to request a sample. For the design, we asked for a white background, dark text, and a teal accent.

The builder returned a working preview.

So far, so encouraging.

Then we read it.

The main paragraph began with our instruction:

"Create a compact single-page English draft…”

The tool had placed the brief itself into the website.

It’s the digital equivalent of ordering a birthday cake and receiving one decorated with "please write Happy Birthday in blue icing.”

The preview also used a dark background and orange button rather than the requested white-and-teal look. The requested three deliverables weren’t properly presented, and the button said "Book a conversation” rather than inviting readers to request a sample audit.

There was a page. There was a layout. There was even a respectable-looking button.

But the business message—the reason for creating the page—was missing.

Could a follow-up instruction fix it?

We tried a more precise edit, supplying the exact headline and paragraph we wanted.

That attempt returned a message saying the selected model was busy. It suggested switching to Gemini, even though Gemini was already selected.

That’s a confusing recovery instruction for anyone. For a nontechnical user, it’s especially unhelpful: you’ve reached the point where the software needs to guide you, and it sends you back toward the option you’re already using.

The website feature has an appealing structure, including an editable preview. In our test, though, it didn’t deliver a usable finished page.

Next: tell people about the business

Marketing offers a choice between guided manual setup and describing your campaign to AI.

We selected LinkedIn and asked for one short draft introducing Sample Studio. We included the price, audience, and purpose, and explicitly told it not to invent testimonials, statistics, or guarantees.

This is a useful test for any AI marketing tool. A polished paragraph isn’t much help if it quietly adds "trusted by 10,000 businesses” to a company we invented five minutes ago.

Unfortunately, we didn’t get as far as judging the copy.

The Gemini 2.5 Flash option returned an error saying that model was no longer available to new users.

We switched to the default Gemini Flash option. That returned a temporary high-demand error. One retry failed too.

Those are different problems.

A provider being temporarily busy doesn’t prove that Will&Be always fails. But offering a model that rejects new users—and providing no successful recovery in our test—is still a problem with the experience.

For you, the practical outcome is the same: the campaign is unfinished, and you’re back in the browser-tab jungle.

The support test finally gave us something useful

Customer support was the brighter spot.

We created a fictional message from someone asking what the €149 audit included—and whether it would guarantee double their sales.

Without any supporting information, the draft was a generic acknowledgement:

"Thank you for your message. We are looking into this…”

Polite? Yes. Helpful? About as much as an elevator announcing that it is, indeed, an elevator.

Then we added an FAQ explaining the offer:

• A written assessment of homepage clarity.
• A mobile usability checklist.
• Three prioritized improvements.
• No guarantee of sales or revenue.

The next draft included those details and the no-guarantee statement. It also displayed the FAQ it used.

That’s a small but important success.

A knowledge base is essentially a company handbook. Giving the assistant that handbook means it has something concrete to consult when answering questions.

The useful moment here wasn’t dazzling creativity. It was getting the right business information into the reply.

The response mostly copied the FAQ, including our wording identifying it as a fictional test. So it still needed editing, and we couldn’t establish how much AI generation happened behind the scenes. But the information was relevant, and the source was visible.

For routine customer questions, that’s a practical foundation.

The approval queue is one of its better ideas

We sent the draft into Will&Be’s internal approval queue.

It appeared as a pending support reply, with the recipient, subject, message, and options to approve or cancel. We left it pending.

Think of this as an outbox with a supervisor’s desk in front of it. A draft can reach the desk, but it still needs a decision before proceeding.

This matters because the difference between "AI wrote something” and "AI sent something to a customer” is substantial.

We verified that the draft reached the queue. We did not test final delivery or establish that every possible action follows the same controls.

Still, it’s a useful part of the product’s design.

What about invoices, branding, and the rest?

The finance section includes invoices, quotes, expenses, and an export option for a tax adviser.

In an earlier form test, we entered two items at €100 each with 19% VAT. The interface correctly showed €200 net, €38 VAT, and €238 total.

That verifies a basic calculation. It doesn’t verify a complete accounting system—but basic calculations are a good place to start.

Branding was less convincing. The colour-suggestion feature returned neutral defaults. A one-option logo test briefly showed that it was generating concepts, then returned to the starting form without a visible logo or explanation.

Other sections felt thin during our walkthrough. After we created a task in Operations, we couldn’t find visible controls to add another, edit it, or mark it complete. Strategy described goals and key results, but its creation form offered a title, notes, and categories such as positioning and competitors.

There’s a broad menu here. The depth varies.

The design: an attractive dashboard needs useful next steps

Will&Be’s dark workspace has a consistent visual identity. Navigation is familiar, primary buttons are easy to spot, and the sections generally feel like parts of the same product.

That consistency helps when you’re moving between several jobs.

However, the dashboard repeats empty counts, some text has low contrast, and several pages offer limited guidance once you hit a problem. We also saw a substantial contrast issue on the public landing page during our initial visit.

For nontechnical users, clear error messages matter just as much as rounded corners.

"Something went wrong” leaves you stranded. "Your selected model is unavailable; choose this supported option” gives you a next step.

The best business software reduces the number of decisions you need to make. Will&Be still creates some extra ones.

Pricing: check both parts of the bill

We created an account and accessed the workspace without encountering a payment step during our test. That does not establish that the service is permanently free.

We couldn’t verify a public subscription price or a complete list of plan limits.

There’s also the separate AI provider account to consider. Because Will&Be uses your API key, AI usage may carry charges from that provider, depending on your account and the features used.

Think of it as renting an apartment where electricity may be billed separately. You need both numbers to understand the cost.

Before relying on it, check the Will&Be plan, provider charges, and any hosting costs for a website. The app didn’t expose verified charges for our test, so we can’t give a measured cost per task.

Privacy and terms: what we could establish

Will&Be’s settings state that API keys are encrypted on its server and are not shown again. That is the company’s statement; we didn’t independently audit its security.

Connecting a key gives the service a way to make AI requests through your provider account. Prompts and relevant business content may also need to reach that provider for processing.

We couldn’t verify accessible privacy and terms documents explaining data retention, deletion, training use, or the full conditions of service. Those remain unanswered questions rather than evidence of a particular practice.

Our tests used fictional business information and a sample customer address. For anyone considering real use, the sensible starting point is similarly ordinary sample content until the data-handling terms are clear.

So, would we use it?

Will&Be is pursuing a worthwhile idea: help a solo business owner move from "I have an offer” to "I can explain it, sell it, and support customers.”

That’s more useful than another empty chat box asking what you’d like to discuss today.

The support FAQ workflow and approval queue showed why this approach could work. When business information is shared between tools, you spend less time copying it around and more time using it.

But our tests also exposed a gap between the product’s promise and its current results. Campaign generation failed. The website displayed instructions as copy. Editing hit an error. Logo generation produced no visible result.

Our verdict: promising enough to watch, but we wouldn’t yet recommend it as the main workspace for running your business.

A future retest could change that assessment, especially if model compatibility and error handling improve. For now, the most valuable upgrade would be making a few everyday workflows work reliably from beginning to end.

Because when you’re running a business alone, you don’t need your AI co-founder to have every job title.

You need it to finish the website.