I have lately learnt of the term "co-innovation", really fancy terminology for incremental innovation. We can learnt from the automobile industry where parts manufacturers eventually become car manufacturers. Often, we can piggyback on established applications/processes as a means of growth. A good example is MLedger (again something I have learnt of recently) that wouldn't have been without MPesa.  In that respect, one doesn't need a pot of money for R&D.

There is also the Science, Technology and Innovation Act of 2012 that envisions (among other things) the creation of a Kenya National Innovation Agency and a National Research Fund. If we designated a % of our annual budget for R&D and disbursed this well through the NRF, we can be miles ahead.

Further, we could consider incentives (e.g. tax writeoffs, etc.) for companies' spending on R&D so as to encourage R&D.

On your taxonomy, I agree. I believe the local ICT sector has a similar taxonomy for classification of enterprises in the innovation value chain.