System of Dialogue (SoD)


Within an enterprise documentation system, a system of dialogue (SoD) is one that facilitates unstructured, ad-hoc human communication, unlike systems of facts, whose data is structured and easily extracted and transformed.

Emblematic examples of systems of dialogue are:

Comparison with Enterprise Collaboration Tools

All systems of dialogue are enterprise collaboration tools, but not all enterprise collaboration tools are systems of dialogue. Systems of dialogue are about fluid, ad-hoc communication, without any explicit changes to the present or future state of the enterprise.

For example, while Atlassian Jira is an enterprise collaboration tool, it is not a system of dialogue because the tickets created on it embody a committed intention of change (and effective change once closed).

Comparison with Image Sources

The emergence of whiteboarding services like Miro and Mural create an overlap between systems of dialogue and image sources. While these are regular image sources—from a DocOps automation perspective—they may or may not be treated as systems of dialogue.

System of Dialogue (SoD) Image Source but not SoD
Example Miro Draw.io
Collaboration Multiple participants can work on the same diagram concurrenctly While the diagrams may not be locked down on a per-participant basis, concurrent collaboration usually works in lock step
Purpose Bring participants on the same page (result may be disposed off afterward) Describe a past, present, or future state of the enterprise
Value North Star Democratic participation Accuracy and clarity

In conclusion, a whiteboarding service, may not act as a system of dialogue if used by a small number of participants for the purpose of producing formal pictorial artifacts.

DocOps Automation

Barring the case of whiteboarding, most systems of dialogue involve text, audio, and/or video streams. As such, following the extraction of said assets, multiple transformations may be applied:

  • Text
    • Parsing and performing keyword lookups
    • LLM-based summarization
    • Translation
  • Audio
    • Speech-to-Text (STT) followed by text processing (above)
    • LLM-based summarization
    • AI/ML-based participant identification
    • Audio clip slicing and adding slice-specific metadata
  • Video
    • Extraction of audio stream for audio processing (above)
    • Video clip slicing and adding slice-specific metadata

Increasingly, many of these transformations are performed by the vendors themselves (for example, Microsoft Teams with Copilot offers meeting transcripts and summaries). However, the results may not be necessarily available through the product’s API.


© 2022-2024 Ernesto Garbarino | Contact me at ernesto@garba.org