Manager

Context Window

I have heard that once AI / agent / llm tools have been given enough access to the whole project and the bug, feature tickets. It does recursively try to invent or fix things.
Even open up PRs to update the dependency versioning to make sure you're on latest to avoid security exploits on previous versions.

It also opens a job security issues for so called scrum master project integrators product owners
If the new AI agents are able to communicate within themselves, with a specific API / contract between each other.
eg. Speak only in these terms, unicode formats, architecture diagram, mermaid, git PRs, issues, timezones (funny! no more indian EOD emails or Please do the needful)

ref indian managers communication skillz
thoughts/management#Indian IT Words

This would be the classic, no need to pay health insurance, servicing fees, less lawsuits, hot deploy / invoke / deinit flow.
No need to have knowledge transfers when onboarding or off boarding people.

Your knowledge silos & bs dependency creation for the sake of creating "value" to your leadership team or stakeholders or shareholders are all prey to this house of cards.

You were there when this architecture limitation decision was taken 7 years ago, to be helpful in a meeting that are discussing things of blockers, wont be as impressive. When while talking in the meeting, AI agents or whatever can be ran in the background deciphering what we are speaking to understand few outstanding questions which gets automatically generated for template meeting summaries. Imagine if AI agent can automatically perform an company wide document search to your confluence, wikis, github spec / rfs and link the original meeting recap text document with another for your review document answering things and retrospecting itself to make an estimate calculations of systems being affected after upgrading, while ingesting already logged analytics of your product's telemetry.

Eg.
We are in a meeting to discuss if we can upgrade our video platform delivery resolution / framerate from 720p HD x 30fps to 2160p 4K UHD x 60fps.
We asked various different senior / lead engineers / architects for specific platforms like android / apple / roku / web / set top boxes.

AI agent summarizes the meeting with 5 outstanding tasks and questions.
Then it starts cloning all the respective repos, platform public docs, company internal spec docs, company's data dog / analytics platform / telemetry.

It deduces that 70% of our clients hardware are capable of playing 4K @ 60fps, our pipeline video engine already uses open source libraries which can do 4K low latency processing
cdns already have specific load balancer and 4K ~100 - 150 Mbps throughput, average internet speed nationwide is 300Mbps, so pushing for infrastructure upgrades is feasible and can make the long term bet worth while for the company and their clients.

It will check outstanding issues for a specific libraries supporting HDR, Spatial audio, DRM protection etc.,

Will all that mapped out, it can further create tickets with sub tasks and if you still gave the go ahead, it can start vibe coding or tbh real defensive coding.

While making thin wrapper around few components in isolation & once that works in prototype. It can ask to be reviewed or if you want to go one more step.

It will start doing the integration in its own sandbox environment, making tagged releases and deploying on web portals, java backend, android, apple test flight etc.,

No more need of pinging managers who are the team owners for creating development signing certificates and provisioning profiles to be added to a new iPhone 17 pro test device.

No more resistance from slack / teams reply of 20 - 2 days SLE, when AI agent is available 24 x 7 in RAM. No need to do pre warm ups no more spinning disks, fans.

End eg.

Sadly, this was my perceived workflow from AI or technology 2 / 3 years back & at that time I felt it wasn't possible especially how everything was being outputted. But in 2026, this now doesn't feel like a rx.pipe'd dream anymore.
This seems achievable in the next 3 - 5 years instead of decade.
Unless humanity decides to not pursue these ventures for the sake of keeping the ferris wheel / guinea pig with some purpose to spin the wheel, this is inevitable is what i feel like.
The question still stays, what will the upper echelon of people want?
Do they want slave labor doing menial tasks like gig workers and no middle class or they want to eliminate bigger portion of resistance so that there would be equity and universal basic income coupled with good EU laws (forced socialism) eventually.
Or we would have a better future of resources being so efficient at making / creation that there won't be people to actually utilize them (consumption US economy).
Wonder if AI agents can just brute force R&D to a point that every industry would just have more margins and the people with physical and material stake in it (Stock holders) only benefits from this future, unless the disparity is so wide, that people start protesting and place temporary bans on deployment of AI agents for profit based ventures.
It can only be utilized for good, solving medical, food crisis etc.,

I don't have a hope for this and even though this post won't be for politics, we have all seen how USA does when it comes to capitalism. If the current timeline has any references of the future it holds, I feel very sadden by the things happening in this time.
Maybe Ultron from Iron man was right, everything needs to be wiped clean and start again, since AI agents could also reach to that conclusion given how humanity has behaved every time, they cling onto superior power.
Absolute power corrupts absolutely.
Terminator typed dystopian or Cyberpunk | blade runner | Akira