AI Idea to Implementation: The 90-Day Model

When thinking about artificial intelligence, many companies get stuck in the same places. They may be AI curious, but struggle with use cases, staffing expertise, and building a business case. That’s why we believe in practical innovation through AI.

The KUNGFU team designed its Deep Innovation process with a progressive approach to AI. The goal is getting the first win on the board that delivers real ROI and develops the trust needed to support a long term relationship.

The Deep Innovation process can take a project from idea to implementation in 90 days, whether the client developed the idea or it came out of a KUNGFU Innovation Workshop.

Week 1: Innovation Workshop to Define Project

If a company doesn’t have a clear idea of the project they want developed, our Innovation Workshop is a fast and efficient method for generating and developing ideas and prioritizing them based on technical risk and business objectives. These workshops are typically three to five days and include a carefully selected group of stakeholders from the client along with AI experts and a professional facilitator from KUNGFU. It is not uncommon to also include an outside expert with relevant domain and technical expertise to help increase the opportunity for bold new ideas to emerge. Coming out of an Innovation Workshop, a company will have definition on one project, with stakeholder buy-in, as well as potential projects to build out in the future.

Weeks 2 to 5: Proof-of-Concept to De-Risk Project

The second step of the process is to proof of concept a machine learning model to prove feasibility and validate that the available data can support it. This is important because if a project is poorly defined, the data is dirty or unavailable, or the infrastructure can’t support the effort the overall project will fail. Developing a proof-of-concept and understanding the contours of its capabilities de-risks the project. It is common that a proof-of-concept can be developed in two to four weeks of experimentation, engineering, data wrangling and data cleaning.

Weeks 6 to 11: MVP Development

With the confidence of a successful prototype, the next step of the process is to build a minimum viable product (MVP). During the development of an MVP, the simplifications of the machine learning model used in the prototype are built out. And, the full data training set is also prepared.

Simultaneously, the interface is defined, refined and developed, whether it is a machine-to-machine interface or a human interface built in collaboration with the end users. Likewise, the infrastructure to support the solution is built out, whether it is cloud-based hosted servers or on-site servers. Because the project was selected and scoped to deliver a quick win, this phase is intended to take three to six weeks.

Week 12:Testing and Deployment

With the MVP completed, the next step is to perform final quality assurance testing and put it into production, with a goal of this phase taking a week. With a properly scoped project and fast and collaborative decision making, it’s possible to go from idea to deployed project in 90 days.

Final Thoughts

If you are considering jumping right into the AI pool, or just stuck petrified at the edge of the diving board, consider a quick-win strategy. Our Deep Innovation process helps our clients first I.D. the use cases and then build the business case. We offer the expertise so you can avoid the staffing challenges. Ideas to action happen fluidly with minimal downtime for handoff so a project can go from idea to implementation in 90 days. It’s a good place to start.