Good to know<\/strong> The goal of the Conversational style is to augment developer problem-solving productivity, such as when writing specifications or reviewing the generated code, by collaborating with the AI Agent to benefit from its abilities, while maintaining control and awareness of the progress of the problem solution.<\/p>\n\n\n\n Good to know<\/strong> These collaboration limits burden the developer, causing cognitive fatigue. Maximising productivity when engineering an AI-Augmented Development workflow requires balancing upfront autonomous delegation with strategic, conversational pairing. <\/p>\n\n\n\n \n\t
Effective delegation can reduce the need for developer intervention during task execution to close to zero most of the time, but it cannot eliminate it entirely.
This is because, in an evolving codebase, the need for difficult decisions regarding design, architecture, and infrastructure emerges gradually in ways that cannot always be anticipated.
These decisions often require an overall product vision, technical strategy, and a holistic understanding of the technology ecosystem and codebase that only a person can possess.
Specification gaps, unanticipated edge cases, and missing authorisations may also require developer intervention.<\/p>\n\n\n\nConversational style<\/h2>\n\n\n\n
\n
The conversational style is constrained by the AI agent\u2019s limited ability to collaborate naturally with humans.
The same issue occurs during handovers in the directive style, even if these are much less frequent.
Key friction points include:
– Opacity<\/strong>: the agent’s internal process is hidden, making it difficult to follow its “thinking”
– Lack of natural friction<\/strong>: the agent treats and presents equally simple and hard parts, making them indistinguishable
– Muddled presentation<\/strong>: the agent presents its “thinking” and solution with a hard-to-follow structure, over-explaining the simple points and glossing over important and difficult points<\/p>\n\n\n\n
<\/p>\n\n\n\nConclusion<\/h2>\n\n\n\n
A key part of this requires knowledge and understanding of different interaction styles between the AI agent and the developer, to the point of deciding correctly when to use one or the other in the workflow, based on your context, needs and wants.<\/p>\n\n\n\nRead Also<\/h2>\n\n\n\n
\n
\n\n\n\n![\"\"](\"https:\/\/www.smharter.com\/blog\/wp-content\/uploads\/ai-develop.png\")
<\/figure>\n<\/div><\/div>\n\n\n\nShip Production-grade Software Faster with AI. No AI slop.<\/h3>\n
\n\tDrive Value Faster with AI. No AI bloat.<\/h3>\n\t
\n\tSee how we can help across the full AI adoption lifecycle.\n\t
\n\t
\n\t<\/p>\n\n\t