First Bitcoin in 2008, and later Ethereum in 2014, held a powerful promise: online decentralized governance, without servers or central controllers, not just for finance applications like crypto-currencies but for any organization. The so called Decentralized Autonomous Organizations (DAOs) were expected to fulfill such a promise, enabling people to organize online relying on blockchain-based systems and smart contracts automatizing part of their governance. In 2016, three DAO software frameworks —Aragon, DAOstack and Colony— emerged aiming to facilitate development and experimentation in this field. To which extent do they facilitate DAO development today? This paper performs an analytical comparison of these three frameworks, focusing on their current functionalities for building DAOs. We find Aragon to be the most complete in several aspects. In order to provide more details on the challenges on building DAOs with current frameworks, we present a case study using the Aragon framework. Through this case study, we have piloted DAO development using this framework, and thus we may highlight the benefits, limitations and problems that developers face when adopting it. Our findings show that, even if Aragon does provide superior capabilities to other frameworks, it is still highly challenging to build a DAO with the current tools. Today, problems include issues on software engineering, instability, localization, documentation, lack of formalization and standards, and interoperability. Complementarily, this paper aims to provide some guidance to those developers aiming to face the challenges in developing a DAO, and to those aiming to fix the major weak points that make DAOs the organizations of a still distant future.