Gorstaks convo with Kiro AI, 2026

i gave the ai almost 15 minutes to merge 2 website codes and it is failing everytime when it comes to merge the JS code

ShaneMKelley [G2G], 

 — 8:09 AM

What a roller coaster of emotions. My goodness.

ixdoo

 — 8:10 AM

so freaking annoying

i even tried using the whole code as text file and uploading there

its just 300 lines of code thats it

 [GMNI], 

Valk_RizzGodStarkilla

 — 8:12 AM

My github repo is more insane lol

ShaneMKelley [G2G], 

 — 8:14 AM

blah always code but never look at cool delivery of the code.

Valk_RizzGodStarkilla

 — 8:14 AM

My days actually been a Rollercoaster emotion wise as well

ShaneMKelley [G2G], 

 — 8:15 AM

Yeah If I remember correctly you were getting surgery you still alive?

 [G2G], 

Valk_RizzGodStarkilla

 — 8:15 AM

I need another surgery

C3 to c8 in my neck all discs herniated

ShaneMKelley [G2G], 

 — 8:16 AM

I'm serious Gemini may be the best book ever.

Valk_RizzGodStarkilla

 — 8:16 AM

Probably got something called cervical myelopathy

MonkeyDrone

 — 8:18 AM

AGY cli /usage has stopped working...ran out of threshold usage, opened up the agents manager and saw the threshold all gone from the models section. jeez

Valk_RizzGodStarkilla

 — 8:19 AM

oh had an amd dev help with that repo of mine love this community

Daniel [PBTV], 

 — 8:50 AM

Are you using straight Gemini app/web?

 [PBTV], 

TalanGPT [GMNI], 

 — 8:50 AM

Is there a gay version?

I’m confused.

 [PBTV], 

ixdoo

 — 8:51 AM

web

 [GMNI], 

ixdoo

 — 8:51 AM

yeah probably used by aliens

Chernobyl 376K

 — 8:51 AM

yooo

TalanGPT [GMNI], 

 — 8:51 AM

 [GMNI], 

ixdoo

 — 8:52 AM

hey just one question man u have this in your profile UI/UX Designer dont you think you will get replaced by ai?

TalanGPT [GMNI], 

 — 8:53 AM

Doubt it. AI doesn’t have taste.

It can’t generate new ideas. It can only recycle old ones.

Daniel [PBTV], 

 — 8:54 AM

Try cli

 [GMNI], 

ixdoo

 — 8:54 AM

for sure but maybe after 4-5 years it will be smart enough?

 [GMNI], 

ixdoo

 — 8:54 AM

yeah

 [GMNI], 

nrnxgatt

 — 8:54 AM

Thags true about most things these days lol.

TalanGPT [GMNI], 

 — 8:54 AM

I still think human design trumps anything an AI can, or ever will be able to do.

Gorstak

 — 8:55 AM

Can you as a human think of something not invented yet?

 [GMNI], 

Daniel [PBTV], 

 — 8:55 AM

You don't tell it how to design it then yes because it'll go to a default design that deepmind set for it

 [GMNI], 

ixdoo

 — 8:55 AM

i usually sell websits to companies and fr that all the ai have same design there is nothing new we just get a sketch/lil plan from ai help and then do our work and sell it xD

Daniel [PBTV], 

 — 8:55 AM

Yes

nrnxgatt

 — 8:55 AM

Ai can never replicate or replace a unique idea thoughm trust me. It gets lost while working on my world ive been developing for 27 years

ixdoo

 — 8:56 AM

i experienced it recently

TalanGPT [GMNI], 

 — 8:56 AM

Yeah. We do it every day. That’s how things get invented in the first place.

 [PBTV], 

Gorstak

 — 8:56 AM

then you're likely a genius. It's a bit too much to ask from AI

Daniel [PBTV], 

 — 8:56 AM

Already have

Gorstak

 — 8:57 AM

With humans, it's usually monkey see, monkey do

TalanGPT [GMNI], 

 — 8:57 AM

I think the word “genius” is largely subjective.

TalanGPT [GMNI], 

 — 8:57 AM

How else did things get invented though?

nrnxgatt

 — 8:57 AM

I have had my trilogies history and logics and more worked on for 27 years. It's so complex that the AI starts hallucinating what happened where and I have to tell it reread chapter x and tell me if what you just said is right

TalanGPT [GMNI], 

 — 8:57 AM

It all had to come from somewhere.

 [GMNI], 

Daniel [PBTV], 

 — 8:57 AM

I took an IQ test when I was 9 and let's just say it was high and you will not guess the number

 [GMNI], 

Gorstak

 — 8:58 AM

An idea comes to mind I guess

ixdoo

 — 8:58 AM

did you earn enough from programming for 27 years?

Gorstak

 — 8:59 AM

But in 99% of cases it's an idea someone already had before you did

nrnxgatt

 — 8:59 AM

I did, but money is always fleeting when its not millions lmao

But my books are hard coded like a rigid structure of a header file.

ixdoo

 — 8:59 AM

interesting

Gorstak

 — 8:59 AM

usually it's upgrade, like adding a flaslight to a cig lighter

but both existed previously

nrnxgatt

 — 9:01 AM

I finished my first book 3 weeks ago sent 5 queries out and have a response that I then compiled the full book for them with correct format using gemini. So I may have an agent soon. And could never have done it without the rigid logic base structure of AI.

ixdoo

 — 9:01 AM

cool

nrnxgatt

 — 9:02 AM

5 firsr book drafts created since 1998 lol. Couldn't get it done before

Gorstak

 — 9:02 AM

If you know what all patents existed, you would be in good position to get an idea for a new upgrade

Then it would be a simple question of how much you could buy or rent it for and make your idea profitable

TalanGPT [GMNI], 

 — 9:04 AM

True. But multiple things can be true at once. You can have an original design, and your design can also be based on something that’s already been done.

nrnxgatt

 — 9:05 AM

Someone needs to create a program thats binary and can turn a Pentium 133 into a whiz machine with new software

Gorstak

 — 9:06 AM

Well, maybe make a prompt to ai to pull all scientific papers and all patents and try combine an upgrade

✘azy [GAZA], 

 — 9:07 AM

love-hate while using AG IDE(i have gemini pro plan)

TalanGPT [GMNI], 

 — 9:07 AM

It can only work with what has already been done. I doubt it would come up with an original idea. In addition to this, AI isn’t human. It can’t design around the human experience. Only humans can.

ixdoo

 — 9:08 AM

talan

lemme show you something

TalanGPT [GMNI], 

 — 9:08 AM

What’s up?

TalanGPT [GMNI], 

 — 9:08 AM

Okay.

Gorstak

 — 9:09 AM

yes, but no human has knowledge of all patents that exist and scientific papers released. An AI could get an idea to combine 2 or more of those into something new, useful

Kiro
This looks like a Discord chat log you've pasted in, not a request for me to do something with your codebase. It's an interesting conversation about AI creativity and whether it can replace human designers.

What can I help you with? Are you looking to:

Merge some JavaScript files together?
Work on something in your Sentinel project?
Something else entirely?
Let me know what you need and I'll jump in.

Est. Credits Used: 0.11
Elapsed time: 4s


Checkpoint
Restore
well, it was a dsicord convo I was in. my nick is Gorstak. Can you actually attempt doing what I suggested?

Kiro
Ah, so you want me to try what you suggested in the chat: pull scientific papers and patents, then try to combine them into a novel upgrade or invention idea.

That's a fascinating thought experiment. Let me give it a shot — I'll search for recent patents and scientific breakthroughs across a few domains and see if I can spot combinations that could yield something new and useful.


Web search: breakthrough scientific papers 2025 2026 novel materials energy
10 results

Web search: recent patents 2025 2026 innovative technology combinations
10 results

Web search: solid state air conditioning patent 2025 2026 how it works
10 results

Web search: perovskite solar cell flexible wearable 2025 2026 breakthrough
10 results

Web search: sodium ion battery patent 2025 2026 fast charging
10 results

Web search: barocaloric solid state cooling material pressure 2025 2026
10 results

Web search: thermoelectric generator body heat wearable 2025 2026 patent
10 results

Web search: piezoelectric energy harvesting road vibration 2025 2026
10 results

Web search: graphene water desalination membrane 2025 2026 breakthrough
10 results
Alright, I've gathered a solid cross-section of recent patents and scientific breakthroughs. Let me synthesize some novel combination ideas — things that don't appear to exist yet but could be built by merging two or more existing technologies:

Novel Invention Ideas by Combining Existing Patents & Research
1. Self-Powered Solid-State Personal Cooling Wearable
Combining:

Wearable thermoelectric generators (Seoul National University, 2026 — thin films that harvest body heat)
Barocaloric solid-state cooling materials (Cambridge/Barocal, 2025-2026 — solids that cool when pressure is applied)
Flexible perovskite solar cells (Osaka Expo 2025 — 10g wearable solar panels)
The idea: A wearable vest or armband that uses your own body heat + a small flexible solar panel to generate enough electricity to drive a micro-pump that cyclically pressurizes a barocaloric material, creating localized cooling without any refrigerant gas, compressor, or battery. The thermoelectric layer harvests the temperature differential created by the cooling itself, creating a partial feedback loop.

Why it doesn't exist yet: Each piece exists in isolation. Nobody has combined body-heat harvesting as the power source for a barocaloric pressure cycle in a wearable form factor.

2. Road-Powered Desalination Stations
Combining:

Piezoelectric road energy harvesting (embedded ceramics that generate power from traffic vibrations)
Graphene-oxide nanopore desalination membranes (99.6% salt rejection, low energy requirements)
Sodium-ion battery storage (cheap, fast-charging, works in extreme temperatures -40°C to 60°C)
The idea: In coastal cities with heavy traffic, embed piezoelectric harvesters under busy roads. Store the energy in sodium-ion batteries (cheap, abundant materials, no lithium needed). Use that stored energy to power small-scale graphene membrane desalination units placed along the roadside. The system produces fresh water with zero grid electricity, powered entirely by the weight of passing vehicles.

Why it's novel: Piezoelectric roads currently power only sensors or streetlights. Graphene desalination exists but needs a power source. Nobody has connected these two into a self-sustaining water production system. Sodium-ion batteries make the storage economically viable since they work in the harsh temperature swings of outdoor roadside installations.

3. Self-Healing Solar Roads with Integrated EV Charging
Combining:

Flexible perovskite/silicon tandem solar cells (33.6% efficiency, bendable)
Piezoelectric vibration harvesting from road surfaces
Sodium-ion battery fast-charging tech (11-minute full charge, works at -40°C)
Barocaloric cooling (pressure-driven, no refrigerant — roads already provide pressure from vehicles)
The idea: A road surface layer that combines flexible solar cells on the top with piezoelectric harvesters underneath. Vehicles driving over the road literally pressurize barocaloric materials embedded in the road substrate, passively cooling the electronics and batteries below. The dual energy harvest (solar + mechanical) feeds into roadside sodium-ion battery banks that provide ultra-fast EV charging. The vehicles themselves maintain the cooling system just by driving.

Why it's novel: The key insight is that the pressure from vehicles — normally just wear and tear — simultaneously harvests energy AND provides the cooling mechanism for the electronics. Two problems solved by one force.

4. Atmospheric Water Harvester for Arid Regions (Zero-Energy)
Combining:

Solid desiccant adsorption systems (materials that pull water from air)
Concentrated flexible perovskite solar cells (for daytime heat generation)
The new zero-electricity cooling prototype (ammonium nitrate endothermic reaction, Royal Society of Chemistry 2026)
Graphene membranes (for final purification)
The idea: At night, a solid desiccant material passively absorbs moisture from air (even in deserts, nighttime humidity rises). During the day, flexible perovskite solar concentrators heat the desiccant to release the water as vapor. The vapor passes through a graphene membrane for purification, then condenses using the endothermic chemical cooling reaction (no electricity needed). The cycle repeats daily with zero external power.

Why it's novel: Existing atmospheric water generators use compressors and significant electricity. This combines passive adsorption + solar thermal + chemical cooling + graphene filtration into a fully passive system.

5. Body-Heat-Powered Medical Implant with Solid-State Temperature Regulation
Combining:

Implantable thermoelectric generators (patent from 1976, updated with 2026 flexible thin-film tech)
Sodium-ion micro-batteries (safe, no thermal runaway risk unlike lithium)
Magnetocaloric/barocaloric micro-cooling (solid-state, no moving fluids)
The idea: A medical implant (pacemaker, insulin pump, neural stimulator) that never needs battery replacement. It harvests body heat via thermoelectric thin films, stores energy in a tiny sodium-ion cell, and uses a magnetocaloric element to regulate its own temperature — preventing the overheating that limits current implant power density. The cooling element is triggered only when the device is under heavy computational load.

Why it's novel: Current implants use lithium batteries that eventually need surgical replacement. The combination of body-heat harvesting + safe sodium-ion storage + solid-state thermal management could create truly permanent implants.

The Pattern
What I did is essentially what you described: look at what exists across different domains (energy, materials, cooling, water, medical) and find the interfaces — places where the output of one technology is exactly the input another one needs:

Body heat → thermoelectric → electricity → pressure pump → barocaloric cooling
Vehicle weight → piezoelectric → electricity → graphene desalination → fresh water
Sunlight → perovskite → heat → desiccant release → water vapor → graphene filter → drinking water
The inventions aren't in any single component. They're in recognizing that these components fit together in ways nobody has assembled yet.

Your point in the Discord chat stands — no single human has the bandwidth to know all patents and all papers. But an AI scanning across domains can spot these cross-pollination opportunities much faster. Whether any of these are actually practical would require engineering validation, but as starting points for patent applications or research proposals, they're defensible combinations of proven technologies.