On January 9th, 18F released an Agile Delivery Services RFI, and challenged respondents to show not tell. This approach removed the constraints of page counts, font sizes, and compliance matrices; a refreshing break. We allowed ourselves to just do. To design, to work cooperatively, to test, revise, build, expand, refine, and deliver.

The RFI was the best stress test we could have asked for: real-life users on hand to define, inform, and validate, proving to ourselves that we had the skills, knowledge, and will to respond with a best-of-class solution. We also learned a lot about what we didn’t know.[^1]

So we practiced. Following the same basic principles, we designed a way to prepare for the RFP, to advance our readiness and make a place where all the crazy and beautiful ideas we hatch and nurture become the normal way to work.

We sourced our own challenges based on different datasets on http://www.data.gov, giving teams a day to analyze the problem, understand audience needs, define a solution, qualify a viable product, and build. Debriefs and retrospectives followed. We built a core set of components that we found we would need in a majority of challenges; and identified overlapping skillsets on the team, allowing for teamed programming.

As a result, we found our right mix … defining what the team would need to succeed in a 3-day challenge and qualifying a tech stack that best supports rapid development of working code. We focused on APIs, knowing that we wanted to design for content versus a specific medium or tool. We built out the automated aspects of a continuous delivery / continuous integration environment, and surveyed across our commercial and federal dev teams to learn more about their favorite open source frameworks and understand the advantages of each.

diagram from Andy: IDE >> GitHub >> CodeShip >> Heroku >> AWS[^2]

Two dry runs in, we have a refined process that rapidly qualifies a user-focused MVP and sequential sprints; established practices that define how far we can take server calls and allow front-end developers, UX, and the back-end team develop and iterate in parallel; created API wrapping layers that facilitate use of intermediary services such as ElasticSearch to consolidate multiple data sources and streamlined data discovery, and increased test coverage.

With each test, we’ve started with a clean concept and relied up on a continuous integration environment as well as tools, like CodeShip and Heroku, to quickly draft and deliver a working product. Tackling this challenge compelled us to adapt. It renewed our focus and infused a new energy and purpose. We are ready, and can’t wait to see what the 3-day challenge brings.