CaseInnovation intelligence · US

    AI engine that finds and maps innovations for investors

    Short answer

    The client shows its customers who is creating what and where, and whether it fits their sustainability goals, with much less manual data processing for its analysts.

    Stack and topics

    Machine learningGoogle CloudBigQueryApache BeamGraph databaseData visualizationKeyLines
    ResultDelivered
    IndustryInnovation intelligence
    RegionUS
    Scale4 people

    Context

    The client and the problem

    Our client helps industry leaders and investors find, curate and track innovations being developed by startups, large companies, engineers and scientists around the world.

    The raw material was a mass of unstructured data: research papers and publications. Finding a solution, or the projects similar to it, took analysts a long time, and the client wanted an engine that connects innovations with the investors and corporations looking for them.

    What we built

    What did we build?

    We joined in 2019 to build the data side of the client's AI-based innovation engine and a custom way to see it.

    • Proof of concept. Research and a proof of concept came first, in 1.5 months, which let the client move quickly to the next stage of the project.
    • Serverless data collection. A serverless architecture on Google Cloud that gathers information on innovations from research papers and publications, with BigQuery and Apache Beam.
    • Graph database. Innovations, their creators, locations and goals stored as a graph, so links between them can be queried, not only listed.
    • Relationship visualization. A custom visualization of a large array of unstructured data, built with KeyLines and Three.js, so the client's customers see who is creating what, and where.
    • Match-making engine. Matching between innovations and the investors and corporations that look for them, including a check against sustainability goals.

    How we worked

    Team and process

    A team of four worked with the client: a solution architect, two machine learning engineers and a front-end engineer. The work ran for about a year and a half.

    Facts

    Project facts

    Project facts
    FactDetail
    ClientAn innovation intelligence company serving investors and corporations
    RegionUS
    Our roleProof of concept, data collection architecture, graph data model, visualization, match-making
    Team4: solution architect, 2 machine learning engineers, front-end engineer
    StackGoogle Cloud, BigQuery, Apache Beam, GraphDB, KeyLines, Three.js

    Result

    Where it stands

    The client gives its customers detailed information on who, where and what exactly is being created, and whether it fits their sustainability goals, with much less manual data processing for its own analysts.

    Next step

    Building something similar?

    A 30 minute call with Oleg. You leave with a clear view of scope, risks and the next step.

    Oleg Pun

    CEO, Vareger

    Oleg Pun, CEO, Vareger