Trial Segments and the raison d'etre of Marmalade AI

Since Marmalade AI needs a mass of people with similar backgrounds for the MVP trial segments (if there’s no there there, why would you use it?), we think we need to bring as many people on board in as short a period of time as possible with basically no barriers to trying out the app.

At the same time, the Paul Graham (Y Combinator founder) mantra of “do things that don’t scale” means we need to be open to learning in ways we can’t even anticipate. Probably that means we need people who know and can own a segment.

Who needs smart connections? There’s been a lot of focus on the “buy side” (companies hiring and information sources selling to them) of this but not much on the “sell side” (people who are looking to do their best work). On the sell side, there’s smart connections to improve expertise, to get full-time jobs, and to freelance. We think freelance is the best area of focus initially.

So we’re looking for suggestions of specific freelance segments, and also people who know the segment and are enthusiastic about seeing how Marmalade AI could help make smart connections in that segment. The following came out of our user research in Q4 2019:

  • content writing, especially for grant proposals.
  • Python programming, especially for machine learning / data science.
  • UX design, especially for problem solving.
  • Japanese / English interpreting, especially for financial business.

We’re open to other ideas!

Seems like a big segment is software freelancing. As you seem to be doing, you are starting in a sub-segment (python/ML/DSc) which is a good strategy. There is a big enough pain point there. Talk to supply and demand side and find what they need.

Somehow, even with linkedIn, agencies still act as the intermediary. This may be because companies do not want to pay freelancers directly (or risk the ire of the IRS). If you can locate enough supply and demand to kickstart a market in a sub-segment, it would make sense to form a company that provides the ‘outsourcing’ service that an agency does (payroll, health ins, etc…). This should be relatively simple to do these days with the appropriate service providers at significant cost savings.

If you solve/improve the matching problem in AI, having an agency-like service to run matches (real freelance jobs) thru I think would be important. Better matches get made, freelancers get paid more, jobs get done with better quality. That’s a compelling story.

Thanks, Maurice. Especially interested in your observation that agencies are still there as intermediaries, even with Linkedin. It seems like it could even be a measure of success that an app could make agencies unnecessary.

Yes, I had thought before about how to disintermediate the agencies. It amazes me that they still exist. If your matching algorithm is of sufficient accuracy, and a global payroll/benefits company exists to dodge potential IRS bullets to the hiring company, then I would say you have a winner. One pain point that solves, among others, is that the agency can no longer arbitrarily set the markup to the company. Agencies are motivated to minimize the freelancer rate and maximize the company rate (anywhere from 10-30% and beyond, and neither side of the market knows). With your approach, and at scale, you have the opportunity to declare a fixed markup for all deals like 1-3% + a fixed % if payroll/benefits is required? That’s a big deal for freelancers. (both of these rates are usually paid by the hiring company).

As an example, is a large globally oriented agency. The are not transparent about their rates. Incidentally, also not clear if they process payroll, etc…

To compete with a toptal, your would need great matching, an innovative customer acquisition strategy, etc… Being transparent about your rates would be a significant differentiator for freelancers (they’ve been waiting for this for many years).