Aleph’s rapid concept prototyping service develops an initial demonstrator which explores the feasibility of a prospective tool based on a data science technology of interest to a client.
This work provides a prototype tool based on a conceptual blueprint. It begins by defining with the client which elements of the conceptual blueprint will be demonstrated within the prototype tool. This depends on the purpose of the demonstration, which might be to explore feasibility, envision capability or gain organisational support. It might, therefore, entail prototyping a relatively sophisticated specific module of a prospective tool, or a set of rudimentary elements which demonstrate a tool’s entire functionality.
Aleph’s approach to prototyping emphasises practical implementation, incorporating approaches identified in any earlier conceptual design stages, and takes a trial-and-error approach, with design decisions and their rationale documented at all stages. Customers receive the prototype tool, accompanied by a technical report providing evidence-based insights which might cover the feasibility of developing a fully operational solution, estimates of the potential performance gains the tool might support, order-of-magnitude costs for its subsequent development, and suggested measures to reduce the technical risk associated with development.
Case studies
01
Working on behalf of a government agency, Aleph developed and tested a prototype autonomous cyber defence agent designed to take into account the operational context the system was working within and to be effective with relatively small datasets.
To demonstrate and test this novel approach to cyber defence decision-making, we developed a simulated network and real-world use case, trialled multiple versions of the agent under different conditions, collected and analysed the data from these trials and provided empirical evidence for the effectiveness of the approach. The prototype and accompanying test results demonstrated the feasibility of this approach and identified the most effective configurations of the agent.
02
We worked with geopolitical analysts and forecasters to build a series of demonstrators which showed how a wide array of machine learning techniques could be applied to different tasks performed in their role.
This work involved the rapid development and testing of discrete instantiations of emerging generative AI techniques which were operated via a basic interface. It demonstrated that these approaches could be applied to tasks such as ontology and knowledge graph building, and analytical report writing. It also gained valuable structured feedback from end users about how they would envisage using such tools once they were available and directly informed subsequent development work.
03
Aleph applied a novel approach to context aware information retrieval to build a prototype search tool for its government client.
This tool showed technically how it was possible to use information about an individual’s role and the tasks they were carrying out at the time to increase the relevance of search returns. Developing the prototype helped to quantify the search performance gains that might be possible with such an approach, informing departmental decisions about whether to invest in the full-scale development of a search tool based on this technology.