Find heart
It"s a risky time to work on things you don"t deeply care about. A wierd pattern emerges in such cases.
Agent deciding the vision and as-a-result the tasks, You forward them to the team. Team(who probably dont even even have the project set on their local) gives that task to their agents who run it in as sandbox and send the "completed task with beutiful prose" back to you. Now since you didnt care to pick that vision or the task yourself, you wouldnt have the appetite/zeal to review the solution either. Forget about the solution/code review, not even the final outcome. So you give it to your agent to review it, send the review back to resource if you have patience or just ask your agent to fix it in place. Merge it.
Rinse Repeat the same process.
Who"s benifitting from this soulless work: The LLM API providers cashing on tokens from both the sides and producing mediocre work and technical debt along the way while giving you the feel of #momentum and #progress.
We get more and more done but no satisfaction whatsoever. Agents give you a this false sense of grandeur. That you are some #builder or #architect of sorts. Like they are jarvis and you are Ironman when in reality people are being reduced to mere APIs to agents talking and doling out work to each other.
Only things that come to our rescue: heart, hardness and deliberate slow down.
-
More you have your heart into some problem, more empathy you have for people facing that problem, more is your attention to detail to elevate that problem, more is your curiosity towards the feedback you get from people using that thing. More is the learning, and hence the satisfaction.
-
Hard problems have already become easier with the advent of Agentic coding. If you are not working on the hardest most ambitious version of the problem, you stand a little chance. It"s easy to get the hard stuff done by Agents and giving yourself pat on the back, harder to solve near impossible problems while delegating hard to Agents. Better still, have some bussiness rely on you to solve that near impossible problem and let it put some fire in your belly.
-
It takes a few prompts to add a new feature into the project. When you mindlessly keep on adding it without understanding it, parts of your brain atrophy and looses the capability to operate and combine ideas at granular level. Sometimes it helps to use guided-learning instead of shipping code right away. To chew on something for a week rather than shipping a feature in an afternoon and forget once it"s shipped.