SprintOverflow

You work on your project; the computer calculates your progress, and predicts the completion date.

1000

backers

$6000

of $100,000

100

days left

hourglass

SprintOverflow is an Agile Software Development tool. It differs from other tools available in the market in that it calculates your progress, and predicts the completion date for your project.

How? The system is based upon a Machine Learning strategy. Despite all software projects being unique in character, the historical progress made on a given project can be compared to other projects with similar experience, and thus a heuristic assessment can be made, statistically, as to the current progress of the project. It also provides an estimate as to when the project will be completed.

There are two fundamental concerns whilst doing a project. The first concern is: what needs to be done, and then assessing completion of tasks to achieve those goals. The second concern is working out how long things are going to take, and when will the overall desired result be achieved. Typically engineering has focussed on the first concern, and project management has focussed on the second. Desires, wishful thinking, business imperatives, and aspirations affect the second concern. Using an analytics driven approach, we remove this bias. As more projects are seen by the SprintOverflow tool, its predictions become better. This is a classic application of Big Data science.

effort and complexity

Contact

Contact faisal@perivalebluebell.com with comments, suggestions, or a request to join the Beta programme.

About the SprintOverflow Agile Software Development Tool

This project came about due to a personal interest in improving the Agile Software Development Method. I have had experience with an Enterprise-wide roll out of the Agile methodology whilst at Nokia; the largest deployment of Agile software engineering practices globally. Does Agile scale? What are the problems? What can be improved?

I realised that Agile was locking together two orthogonal problems, "what" and "when". I had a tangential interest in Machine Learning. As I learnt more about it, I realised that the "when" part of the problem could be assisted by the use of Machine Learning techniques.

I started a demo iOS app to try and firm up my ideas, but having started a Startup Engineering course realised that a preferrable approach is to first do a Responsive Web Application first, for maximum reach, and then to do a native App.

It is thanks to coursera.org, and Stanford University that I have been able to get this far. I aim to establish SprintOverflow in the market, and a create a new company, PerivaleBluebell, to deliver it.

Benefits of Paying Now

Paying now via the "Pay with Bitcoin" link will make this project actually happen. If you pay 10 BTC you will receive the latest version of the software, fully enabled, for life! If you pay 100 BTC, you and everyone in your company gets the latest version of the software, fully enabled, for life! Even if you can't pay much at all, your support counts because it is validation that there is a desire for the software. This helps drive the business forwards to achieving delivery of the product.

FAQ

In agile, you indirectly say how long a task is going to take by assigning story points. These are a balance of two ingredients: amount of work and amount of uncertainty. You add up the points and make sure it is near to what you achieved last sprint. Its the conversion from W(ork) and U(ncertainty) to P(oints) that we improve upon using mathematics, so the points achievement is rigourous. We don't attempt to calculate W and U from the story text, since even humans have difficulty with this.
The revenue model is similar to github. That is, public project work is free-to-use on the system. To create a private project, you need to pay. Only the person who creates the project pays. The team members don't need to pay. Typically a project leader, scrum master, will be creating projects. Open source projects, charitable foundations and academia would thus be free, and commercial use would be fee-paid. The fee would be in dollars or bitcoins. Perhaps three price tiers: individual (12 USD/year), small company (250 USD/year) and large corporation (12000 USD/year).
The cost to create the tool, perhaps 2 person years effort plus selling and operating overheads, is 100,000 USD. Bootstrapping the project is probably a homebrew effort over a year, so folks have something to look at. Two OS platforms are considered. Firstly a web-based mobile-optimised HTML5 app, and secondly a iOS native app. This doubles the front-end costs; the back-end is common code.