Bit of a backstory here....

Years and years ago, I built a startup called Kohab and I put everything I had into that. I didn't have another job and I only had small income from Kohab. We went through Royal Commission. We had to pivot, and then it went to COVID, and it had to close.

That experience taught me something crucial: don't go all in on one project.

With these new products that we're building, I wanted to have several projects that we could build and then focus on them as they needed, focus on either the marketing side or the product side.

Some projects have quick take up. Some projects just take longer to build and test. Some projects have a easier decision making process for adoption, others longer...

For me, I am not a patient person, I am much happier when things are go go go. So, multiple projects at once works for me. Might not work for others, but it does for me.

The Reality of Building AI-Driven Social Content with Mber

Let's start with Mber. One of the things we've been building into Mber is the ability to create social content, from the other content it creates. Remember mber is the one where it interviews you, then turns that into content for your website with built in publishing....

Twitter and LinkedIn are the two that we're trying to get right first.

It's funny, though. Fine tuning AI to do social posts based off content is being a little bit trickier than what I thought it would be.

"You wanna be able to get the human voice in there, and that's the part that's a little bit tricky because you've got less content to work with.

The challenge with social media content is that it needs to feel authentic and conversational. When you're working with AI, you're essentially trying to capture the nuance of human communication in 280 characters or less. Not all social posts need images either - a lot of tweets don't need them. We're working through these challenges, and hopefully, in the next week, we'll have that sorted so that it can be launched.

Arbeo Evolution: From Forms to AI-Powered Interviews

With Arbeo this week, we released the application form agents - if you're looking to hire someone, you can put in your job, and you can say that you want the application form to either be a form or the application form to be an AI chat. The good thing about that is that the agent is designed to get more information than what's available in the uploaded CV. It can ask really detailed questions about what happened in this position, what were the roadblocks you hit, which is really, really good. It's a good way to extract a lot of information.

In my tests, AI is more detailed than I thought here because I typed in lots of spelling mistakes and errors. And the matching came back and said, actually, this person might not be considered for a designer role because their attention to detail was low. It was unexpected, but it turned out that those types of things are really quite good when it comes to AI to be matching. lol

The Human Touch in AI Hiring

Now, the other question we "could" do, should we allow for a voice interview almost as a part of the first application process? Technically, sure we can do it, but is it the right thing to do? Will it provide the correct value for the job applicant? For the person putting in the application, they're probably not ready for an interview with an agent, and it's a lot less personal.

"I suppose when you're hiring someone, you're hiring people. So a person should interview a person, but we might add it in as a configurable option."

Being too impersonal is always a risk when you're talking about someone wanting a job, and the job is more than just something you do. It's something that people love to do, hopefully. And it's something that hopefully provides lots of rewards. If it is too impersonal, then you might turn people off.

What we've got with the agent is personality that's part of the context that we've got, and we'll refine the context as well to make sure it's not overly intrusive. The detailed questions, though, it's so far, it's doing quite well. It nails into a position, nails into a recent project, nails into a recent roadblock. Talks around how do you communicate with your team members depending on if it's remote or on-site.

Handl Billing: The Complexity of Building for Agencies

Handl is still being worked on.

The problem with building Handl is that it's a bigger product than any of these other ones. It's also a little bit more complex in terms of us getting it right, because it's got to do with agency invoices and finances and budgets. A lot of areas that just wont accept any mistakes.

The product is simple to use though, but under the covers its got alot going on.

I need to do a lot more testing on it to make sure that the new things we're adding in are working as expected and there's no glitches.

There's also two sides of the story when it comes to the Handl product.

You've got the agency side, of course, and they've got different levels of people in that side.

And you've got the client, the customer portal, the client portal.

Each agency is going to want the clients to have different access.

• Some clients will get access to everything
• Some will have access to limited project updates
• Some just access to invoices
• Some will want to use the client agent that we have built, that wil help answer all the questions in terms of progress and invoices and terms.

Handl - Solving the Cash Flow Problem

In the next couple of weeks, we'll be able to release a pretty solid version of Handl to our early adopters. In that, we've got some really cool tools for being able to view the margins on invoices as you're sending them, Agency agent to talk to to help with most things around project billing, auto invoice chasing and alerts to ensure project hours are kept on track.

"There is nothing worse than sending an invoice and having to chase up for payments. Agencies don't run a huge margin to have to wait four, six, eight weeks for invoice payments, even two to three weeks."

Because of this, we are also gonna be releasing the auto pay feature for invoices. So when an invoice goes out, you can request or even mandate that the invoice is automatically paid as soon as the invoice has been raised.

At the same time, you obviously wanna respect the client's ability to review an invoice.

Most invoices are gonna be correct if you got a good relationship with your clients. BUT occasionally, there might be a few hours wrong here or something.

So we're gonna allow auto pay on an invoice up to a certain percentage. For instance, Maybe it's auto paid the client's credit card for 50% as soon as the invoice is raised.

And the other 50% comes out automatically after seven days, which gives clients enough time to review and request any changes to the invoice.

‍

‍

‍

Frequently Asked Questions

Why is Dazlab building multiple SaaS products at the same time instead of focusing on one?

After experiencing the closure of my previous startup Kohab, which went through Royal Commission, had to pivot, and then faced COVID, I learned not to put all eggs in one basket. By building several products simultaneously - Arbeo, Mber, and Handl - we can focus on different elements as they need it. Some projects have quick take-up while others take longer for users to find and adopt them.

What challenges are you facing with AI-generated social media content in Mber?

Fine-tuning AI to create social posts is trickier than expected. The main challenges are: making AI sound like your authentic voice rather than generic, working with limited content context (social posts are short), and determining when and what visuals to include. The AI needs to write more like someone sending a text message rather than writing an article.

How does RVO's AI chat application form differ from traditional job application forms?

RVO's AI chat can extract more information than what's available in an uploaded CV. It asks detailed questions about specific positions, roadblocks candidates faced, and communication styles. Interestingly, it even picks up on details like spelling errors and attention to detail, which can be valuable for matching candidates to appropriate roles.

What's the autopay feature in Handl and how does it balance agency cash flow with client needs?

Handl's autopay feature allows invoices to be automatically paid when raised, with configurable percentages. For example, 50% can be charged immediately and the remaining 50% after seven days, giving clients time to review. This addresses the major pain point of agencies waiting weeks for payment while respecting clients' need to verify invoices before full payment.

When will these products be available and which is launching first?

Mber should have its social content generation features sorted within the next week. Arbeo already released its AI chat application feature this week. Handl is the most complex product and will be released to early adopters in the next couple of weeks, as it requires more testing due to dealing with agency finances, invoices, and budgets.