Addestino defines scope, approach and RFP for AI-based public tender process
A public body responsible for auditing dominantly relied on manual steps for its core business, resulting in an inefficient and hard-to-monitor process. The customer wished to rethink their approach from both a technical and conceptual level, leveraging the possibilities of data mining and artificial intelligence. The focus was to create a system to maximally automate the auditing output based on data inputs, without resorting to simply automating the current process. Addestino was asked to advise on the scope, approach and to create a high-level target architecture.
There were two key challenges in this project. The first challenge was translating the high-level vision to a tangible solution that can be grasped any party that wishes to respond to the RFP. The second challenge lies in the necessary technology maturity levels in natural language processing and automatic risk assessment, which surpass the current state of the art. This last point implies that developing a solution has a high risk of failure and a solid plan of action is needed to minimize this risk.
Addestino combines the key success factors that are required for this project: deep technological expertise on artificial intelligence and data mining, methodologies to translate vision to concrete functionalities and the insight to structure complex projects in a roadmap while keeping the eyes on the goal. Additionally, Addestino has a track record of assisting companies and public bodies with RFPs.
After defining the precise scope of the project, Addestino constructed use cases together with the client to capture functionalities of the target system, including which data must be loaded, which information must be extracted from the raw data and what should an auditor be able to see in each step of the project. This input was used to list the required technologies and their current state of the art. After this, a market consultation was set up to determine the technical feasibility for each of the requirements. This includes not only what is possible today, but also the time horizon on which certain functionalities will become possible, like the interpretation of free text. Combining all this information, Addestino created a high-level target architecture that outlines which building blocks are required to realize the project goal. For each of the building blocks, different “generations” in terms of maturity were defined and combined in a phased timeline indicating how the target solution can be built up.
Impact & results
Because of the clear way in which Addestino was able to translate the vision into a concrete action plan, Addestino was asked to further support the technical parts of the RFP, which was launched briefly after completion of the preparation track. By presenting the building blocks of the final solution and working back from there, the original vision of the project is guarded and simply automating the current auditing process is not an option. This ensures that interested party have the right mindset when responding to the RFP.