We open-sourced optimize_anything, an API that optimizes any text artifact. You provide a starting artifact (or just describe what you want) and an evaluator it handles the search.

import gepa.optimize_anything as oa result = oa.optimize_anything( seed_candidate="<your artifact>", evaluator=evaluate, # returns score + diagnostics
)

It extends GEPA (our state of the art prompt optimizer) to code, agent architectures, scheduling policies, and more. Two key ideas:
(1) diagnostic feedback (stack traces, rendered images, profiler output) is a first-class API concept the LLM proposer reads to make targeted fixes, and
(2) Pareto-efficient search across metrics preserves specialized strengths instead of

averaging them away.

Immi helps UK visa holders track their ILR absences and key milestones in one organised system - without spreadsheets or manual reminders.

I'm a UK visa holder working towards ILR - like many of you.

My calendar is full of deadline reminders: "Visa Renewal in 3 Months", "ILR Eligible Date" - also like many of you.

Hey everyone,

I'm Andy, founder and builder of Byterover.

Had a thought, and I must put it out here:
Over the past year, building with LLMs, I ve noticed that the big change is not necessarily that AI gives us better answers. Change your perspective and think like this: we, humans, are starting to ask better questions. And here's why:
When search engines dominated, we asked:
What is X?
How do I do Y?
Best tools for Z?
With LLMs, the questions are different:
Help me think through this decision.
Challenge my assumptions.
What am I not seeing?
Act as a product strategist and critique this.
Clarify my messy thinking.
That s not getting the information, that s cognitive augmentation.
I feel we re moving from:
Finding knowledge to extending reasoning
And this raises bigger questions for builders:
1. Are we designing AI tools that optimize for answers
or for better thinking?
2. Is the real advantage having a stronger model...
or asking better questions?
3. In a world where everyone has access to GPT-level intelligence,
does leverage come from prompting skill, system design, or taste?
4. What new human skills become valuable when execution becomes cheap?
It feels like we re entering a phase where:
Curiosity > credentials
Clarity > speed
Systems thinking > single outputs
If you re building in AI right now:
What types of questions are your users asking that they couldn t (or wouldn t) ask before?
And are you building for answers or for thinking?
Would love to hear what patterns you're seeing.
______________________
www.zackapp.space
Download Zack from the App Store and soon from Google Play: https://apps.apple.com/us/app/za...

Break the spiral!

Try now: https://optimism-engine.vercel.a...

You don't want to walk all the way back to your office just to write down "buy milk" or "email the legal team on Monday." You just tell your assistant, and they handle the paperwork for you.

That s why I m building ThinkTapGo.

The most common question I get is: "What s the difference between opening my task app and just using WhatsApp?" The difference is cognitive friction. Opening a heavy app, finding the right list, and typing takes time and kills your flow. You re already in WhatsApp 90% of your day anyway.

How ThinkTapGo works:

TaskBuddy AI Discipline Alarm for Students & Hustlers

Set tasks

Get daily AI motivation

Build streaks

Honestly, we just wanted a simpler way to use Alpaca.

Alpaca is an incredible backend, but most of the apps sitting on top of it are either built for high-frequency bots or they re packed with enough 'gamification' to feel like a casino.

Launching a context aware suggestion chrome extension tomorrow where I am trying to change how people search and I am looking for contributors.

If you are interested feel free to comment thanks.