The reality is, this is a HUGE project to do with no GIS / image gen experience without Gen AI.
1,128 karma · joined October 31, 2012
I'm currently working on dev tools for agentic engineering and software-first simulated hardware design. Writing and researching as a means of growing.
Projects:
- Intraview (intraview.ai) give Agents the ability to co-create code tours in VS Code to speed up onboarding and alignment on code-bases
- Unfudged (unfudged.io) is source control without the commits so Agents (and you) can recover from mistakes.
- Writing / research (cyrusradfar.com) is where I do market research, guesstimating the future and keep searching the glorious unknown canvas of knowledge.
Live in Portland, OR. Previously lived/worked in both DC, and SF.Find all my projects, writing and contact methods at https://cyrusradfar.com
The reality is, this is a HUGE project to do with no GIS / image gen experience without Gen AI.
One is a VSCode extension and has thousands of downloads across different flavors of the IDE -- won't plug it here to spare the downvotes ;)
Been a developer professionally for nearly 20 years. It is 100% replacing most of the things I used to code.
I spend most of my time while it's working testing what it's built to decide on what's next. I also spend way more time on DX of my own setup, improving orchestration, figuring out best practice guidance for the Agent(s), and building reusable tools for my Agents (MCP).
But I DON'T think the standard should start by piping the prompt directly into claude/model cli. I say this as someone who has seen, first hand, an exfiltration attack locally and almost fell for it after 20 years as a developer.
Even if initially the install.md is safe, install prompt scripts and the things they download aren't packaged and static. They're all surfaces to exploit. The sub-components can be changed between any install, this is true unless we image versions and cache the "safe" imaged version and approve it.
What would be safer to me is a hub that you give a single install script that creates "images", .e.g. DMG for a Mac, .exe for Windows, etc., for platforms. That may actually be an installer app that the User or Agent opens then finishes locally for configuration. Then you point your Agent to that hub.
Nevertheless, then I would just recall XKCD and say, why not just package it with NPM, PyPi, brew, etc.
I've built products that solve my problems and have released one, Intraview.ai -- it's functional, solves a real problem for me and my customers.
That said, as a business goes, it's not a sensation but it's gone from idea to customers using it in less than 6 months. Is it a VC hit, no -- am I happy with where it is and how fast I'm learning -- absolutely!
I think it's said in other words elsewhere but the "picked" state needs to be clearer that it's picked up. What may be helpful is to actually add a drop shadow/scale/border and ideally there should be a hole where it was that makes it clear that you are holding it.
Not trying to diminish, but this pattern has other uses already. Anyone who builds accessibly for folks who use keyboard and no mouse.
Tab/arrows to highlight, Enter to select, Arrows to move, Enter to drop.
I'll dump a few thoughts as they come for the creators, feel free to riff with me on the thread if that'll be of value.
My perspective, as a User, is I'm interested in rooting out bias and where it's coming from. Moreover, the influence networks are fascinating as well.
I think, for example, understanding which publications "picked up" a story vs didn't is very very viral use case as you could imagine people using you as a backdrop to a social post about editorial bias. That said, I think you need to pick who you serve because the folks who will be interested in this aren't the average person as they're not super news focused.
One way to learn may be looking at the types of meta-stories posted about the analysis on media and see how you could support those types of ongoing analysis. Scoring, honestly, is an another really interesting idea. What are publications "for" or "against" based on how they do editorial, and how they bias their headlines, and ledes.
The useful framing is not “where can we bolt on AI” but “what does the system look like if AI is a first-class component.” That requires mapping the workflow, identifying the decision points, and separating deterministic steps from judgment calls.
Most teams try to apply AI inside existing org boundaries.
That assumes the current structure is optimal. The better approach is to model the business as a set of subsystems, pick the one with the highest operational cost or latency, and simulate what happens if that subsystem becomes an order of magnitude more efficient. The rest of the architecture tends to reconfigure from that starting point.
For example, in insurance (just an illustration, not a claim about any specific firm), underwriting, sales, and support dominate cost. If underwriting throughput improves by an order of magnitude, the downstream constraints shift: pricing cycles compress, risk models refresh faster, and the human-in-the-loop boundary moves. That’s the level where AI changes the system shape and acts beyond the local workflow.
This lens seems more productive than incremental insertion into existing silos.
I was! I spent several days spinning in place after I thought it could help me clean up my code quality with biome. Afterwards it destroyed the whole app and I needed to figure out how it worked -- that need, inspired me to prototype and extension for vccode I'm actually still building :)
In contrast, frameworks from learning theorists such as Vygotsky, Piaget, Bloom, Gagné, Maslow, Bruner and Kolb provide more explicit actionable guidance for parents/teachers (e.g., scaffolding learning in the “zone of proximal development”, designing spiral curricula, applying experiential learning cycles).
My perspective guides me to prefer the pragmatic actionable frameworks. That help someone guiding children (or students) set norms, limits and scaffold growth in daily practice.
I do like the conversation that's cropping up here though from this article. A lot of lovely self-reflection.
Switched to purchasing and renting when there's something we want to watch that isn't available and we're finding it to force us to be more conscious of what we're watching.
We're considering ditching Spotify and music streaming to return to buying albums so our children can start to be more thoughtful listeners. After falling down a rabbit hole of some insider music vlogs, I recognized how much streaming is harming independent music.
Given this tech is new, the experience of how we relate to their mistakes is something I think a bit about.
Am I alone here, are others finding themselves more forgiving of "their preferred" model provider?
Can imagine more and more forms being built in as the community goes.
kudos!
There's so much you can do as an extension and the Microsoft team behind it seems to be rational and listening to the extension community about their needs.
I do think Microsoft needs to find a way to host a conversation that discusses why folks must fork and how they can provide APIs to make the tools more configurable.
One big problem for extension programmers in the world with AI chatbots, is the extension is sandboxed and there isn't cross extension communications or a message bus. If there was that standard and a way to send messages, then my extension could push to a AI chat bot. This is a UX shortcoming I'm seeing.
There are also limitations in the ability to overlay UI/UX on top of the editors, e.g. to add comments or annotations that are persistent.
That said, these aren't reasons to fork in my view. I think you a get a good enough experience.
Anywho, I'm new to the eco-system so maybe I'm an optimistic idiot, happy to be wrong.
Frankly, I was assuming there were folks doing this and I'd get educated on that.
Also, really good info on Canada and kudos on compiling that data. I followed your article to this one: https://jacobfilipp.com/the-story-of-canadas-bread-prices-in... which was a good read.
Thanks for your work.
In this case, the reaction is already visible: more interest in decentralized systems, peer-to-peer coordination, and local computing instead of cloud-centric pipelines. Many developers have wanted this for years.
AI companies are spending heavily on centralized infrastructure, but the trend does not exclude the rise of strong local models. The pace of progress suggests that within a few years, consumer hardware and local models will meet most common needs, including product development.
Plenty of people are already moving in that direction.
Qwen models run well locally, and while I still use Claude Code day-to-day, the gap is narrowing. I'm waiting on the NVIDIA AI hardware to come down from $3500 USD
Maybe I'm too old, but after I read the post I thought -- oh this is an "AI-first Quicksilver" -- who remembers that plugin for Mac? I don't think they stayed relevant enough
- Why didn't venues want more of the revenue?
- Why couldn't artists simply set their terms for their tickets? I don't have the answers and it may depend on
There are hands at play I'm not seeing/understanding as this legislation was pushed for by major artists.Really enjoyed how you traced your mental model through the journey of solving the problem.
We also know prices can go up on the raw ticket and we know that consumers can bear that given where they were before this legislation. That additional money will go to Venues, Artists and Labels who will, ideally, create more and better shows for the all of us.
If you imagine today that a ticket is $45 (speaking US) and customers pay $450, eventually for it. That $405 doesn't go to the Artist or the Venue, it goes to a middle layer which adds little to no value for the eco-system beyond supporting a transaction. Moreover, when you spend all the budget on the ticket, you're less likely to buy merch and food at the venue.
Something most on this site could easily build with the help of stripe, a few database tables, and an API to the venue's ticketing management system.
What did you find to be the most surprising challenge in this project?
1. search conversations for PII they know of for the given user and redact
2. use their own models to scrub the conversations of non-verbatim mentions
In the end, the NYT isn't asking for the identity of the posters. If that was the case, I'd 100% be onboard to scream bloody murder with 'em.Sources?
- https://en.wikipedia.org/wiki/Fasting
- https://zerolongevity.com/blog/the-history-of-fasting/My preference would likely be to make the quickest buck and move on as I can/will be disrupted by someone who wants to charge less. It's a race to the bottom so try to get out at some local minimum.