Begin by creating a portfolio of your work and highlighting it on your LinkedIn profile. If you don't have a portfolio of work to point to, you need to get on that ASAP. Take on interesting paid projects via elance, guru, etc. or via a local placement agency, bidding aggressively if necessary to build a portfolio of work that will become the single most important reference point of your capabilities.
Determine a technology or platform focus. It's much easier to get work (and charge more) if you've got a reputation vs. being a jack of all trades developer. Yes there's tremendous value -- once you land a project -- to a broad set of experiences, but in the sales process it creates confusion. Pick what you love and can talk passionately about. If you're on the fence between 3 technologies, pick the ones that corporations view as most in demand.
Understand your target individual (the person who signs your contract). If you can understand and communicate directly with CEOs and CFOs in terms of business outcomes you will make 2-3x more money than if you are selling technology expertise to a CTO / head of IT. Technical managers are accustomed to looking for talent they generally view programmers as higher or lower skilled replaceable components. With a CEO, you can become a trusted partner - that's often not the case with an IT manager.
Once you've done all that, begin networking. You can find projects online, you find long-term clients via relationships.
Setting up a software services business comes at a hefty expense of time, energy, effort, patience, and perseverance. It's advisable to have a solutions around your set of services. If you don't have any such solution credentials then building one should be your top priority.
As far as new projects is concerned, it's good to seek projects through freelancing websites as long as you don't make that your habit. Because, you are out there to establish a business venture and not end up being a freelancer. Also, it's easier to bag projects as freelancer, due to obvious cost advantage, but the scenario is exactly opposite for agencies. So, make sure you invest enough time to think and plan around your business model, growth strategy, and log-short term planning.
Looking for anything specific? Feel free to reach out. Would love to share my more than a decade of experience with startup and entrepreneurs.
Find a niche. Establish your expertise / authority in that niche by writing about it, giving talks on it, consider things like a drip-email campaign.
Without a niche, you're competing with every other software development firm out there, and it's hard to differentiate. With a niche, you can establish your authority.
The niche can be a platform, particularly a new platform where it's easier to establish authority, or it can be solving a particular class of problem (e.g. customer on-boarding, A/B testing, migrating subversion to git, etc.).