Raw LLM Responses

Impact of code AI on developers... Up until now development team consisted of PMs, BAs, Architects, Designers, coders, testers, deployment specialists etc FROM HERE ON ... PM role - will be affected as LLMs can produce project plans and schedules and work will be achieved through deligation of authority and via collaboration tools and agile processes. Business Analyst role - will come under pressure and some companies may do away with it. IMO they should still be there as business are not good at writing business requirements and it is not their forte. Architects role - will exists for new bespoke large enterprise developments. Also due to the evolving technological landscape they may get more involved in product evaluation and in building a playbook, standards etc for using LLMs. May become less important for changes to existing enterprise apps. Designers and Coding roles - may not require as many developers as in the past...most likely will be downsized. The role will change in that they will need to skill themselves to use productivity tools like Claude code, codex etc. Challenges : Now they will be required to specify the product requirements to the coding tools in a succinct manner to get the optimal result. Here in lies the challenge....how to ensure product requirements are given that align closely to the business requirements in order to ensure the code generated by tools meet the business rqmts in minimal cycles in order to contain the number of tokens used given that will be a big cost constraint $$$$$. So the role will now become more a code reviewer than a full fledged coder and the KEY SKILL will be how well the product requirements can be specified to the LLMs. To think that all will hunky-dory from day 1 will be like living in a fantasy world. The coding agents are still in their infancy...SO EXPECT A LOT OF REVIEWS AND CORRECTIONS....IE GOING BACK AND FORTH TO LLMS OR DOING CODE CHANGES MANUALLY TO FIX THE PROBLEMS. Testers role - This has already been impacted by automated testing tools and it will be impacted further. Standalone apps may not require testers. Enterprise apps that integrate many applications / systems will still need testers for integration testing.. to set up the data which is not to be underestimated as they have to be correlated across many apps and systems and need to be in different states or cycles for long lived transactions..all this requires domain knowledge. Conclusion : Legacy bespoke apps and systems, packaged software will have minimal impact as business will not rewrite a WORKING SYSTEM DELIVERING PREDICTABLE RESULTS just for the sake of getting thrill from using coding agents. Enterprises will use coding agents for new bespoke developments especially POCs. Start-ups will benefit the most given they do not carry the baggage of legacy apps and systems. Ofcourse the number of people required in a project will reduce due to the productivity gains from coding agents assuming the code generated require minimal or no manual change for it to work as anticipated. In summary as with any other technological change some IT jobs will be lost but new ones will be created.😊 Last point ...AI-AI integration and other standards are yet to be cemented. Also many of the architectural / design concerns like 2 way commits, minitoring, redundancy etc will need to be catered for at an enterprise level. AND LLM IS A BLACK BOX AND WILL ANTHROPIC, OPENAI, GEMINI ETC TAKE OWNERSHIP WHEN LLM STARTS HALLUCINATING !! SO THIS IS JUST EARLY DAYS...SOME TRUTH AND A LOT OF LIES ABOUT AI !!😊