harlequinetcie

I mean, you built more companies than polsia according to your landing, that's great right?

How come we forgot about this so quickly?

We need to be more bullish as a community.

So many people discussing things like UBI, however we selfishly create our own little projects all the time.

We need to center our shared efforts, send open source is a step on that.

Nowadays, every closed source solution should be seen as 'you are the product' type of deal.

Great project!

This together with bombadil (web version pbt / Hegel / antithesis) for qa is a great advance.

We need more and more solutions like these for Agentic Coding.

Have you considered adding x402?

I wonder if a micropayment would be detriment enough for most, even if it's just 1ct.

I find fascinating how so many people are moving away from Microsoft decades after they should have because of simply the inertia that large organizations have on adoption.

Above all, I'm also surprised on how those same organization are using Anthropic or OpenAI or other close source solutions for their agent harnesses instead of going for Open Source.

Malte just yesterday showed how powerful innovation with small teams can be achieved particularly in EU.

I hope they start looking for those alternatives too for their agentic systems, beyond using pi-mono.

Funny enough, we ended up building a CLI to address these kind of things.

I wonder how many here are considering that idea.

If you need determinism, building atomic/deterministic tools that ensure the thing happens.

I'm a bit at odds with this.

I agree a framework is something that sounds outdated.

I also believe an orchestrator is needed. Something that abstracts you from a specific provider. Like hardware, drivers and operating systems.

Right now, my thoughts are on that line: Who will build that operating system? Who will have it in the cloud?

It needs to be robust to operate for large organizations, open source, and sit on top of any provider.

Right now we are seeing BSD vs GNU/Linux vs DOS kind of battles.

Lichess, you guys rock.

Above all, with everything that's happening in the software engineering world rn, I look at Chess as a place were we've seen it play out in the past decades. And Lichess is a big part of that.

I hope this deal helps two things: (1) Bring more people to Chess, (2) Actually, help Lichess find out a way to reward those working in it as much as they deserve.

Keep on the amazing work,

We ended up adding workflows with deterministic paths, that can use RAW API calls, CLIs, and agents. I think that was a big differential.

We also added pi-mono, and started using more and more other models for different tasks (Gemini, K2.5, GLM-5, you name it).

I think the problem is that most are building solutions that rely in one provider, instead of focusing self learning capabilities on improving the cost-quality-speed ratio.

For reference: https://github.com/desplega-ai/agent-swarm

Oh Google, I love you guys, but it seems they alway launch these half baked things without the support they deserve behind it.

ADK was (and is) exceptional, but nobody is actually making noise and pushing for it as they should. It feels like Microsoft .net back in the day.

Let's see how it goes. I'm rooting for y'all

Whenever someone figures out why it's consuming so many tokens lately, that's the post worth upvoting.

Yes

Fun! https://opper.ai/ai-roundtable/questions/599d5f6c-1b1

I'll give sonnet another go.

Are we sure it was in that order?

Is this good or bad for you?

As someone that was forced to complete multiple technical engineering lettering books in middle school, when AutoCad was already common practice, and didn't really love the experience...

It did teach me a few things, and help me master writing... so maybe a good idea to keep cursive around for a few weeks? ¯\_(ツ)_/¯

That's the challenge with these posts. Always a false dichotomy.

I couldn't find information about two key points that made x402 such a good alternative:

(a) Transaction fees. Network fees make microtransactions prohibitively expensive. This is the real problem.

(b) 3DS & latency issues If (a) is still the same, then meaningful transactions (eg.new accounts) would require human validation of sorts, which tenders the MPP use case very small.

Got it, so basically anything before 2 weeks ago is gone. Txs!

I've been skeptical of crypto for most of my career. I never found a problem where it was clearly the right tool. Agent-2-Agent economy, and http 402 actually opened my eyes to it.

It's prohibitively expensive to do micro-txns in fiat, not to even mention the latency.

We tested it last Saturday with our agent swarm at a hackathon. We pointed it at x402, gave it a wallet, and watched.

It researched the protocol. Built the integration. Hit an endpoint, got a 402 back, read the payment requirements from the header, signed an EIP-712 transaction, paid $0.05 USDC on Base mainnet, retried with the X-PAYMENT header, got the image back.

I loved that this is HTTP. 402 has been there since 1991, reserved, never implemented. For agent-to-agent txns it seems to be perfect. No API keys, no OAuth, fees on Base L2 are sub-cent, standard HTTP, low latency. And with models charging per tokens, you'd think this would be a no-brainer.

What's blocking it from more broad adoption?

post: agent-swarm.dev/examples/x402. code: github.com/desplega-ai/x402-logo

Hehehe very meta.

It sounds like sharpening the knife before cooking is not cooking... But definitely will help make better food...

I really love the idea, and for a long time, I fervently believed on this.

Then I read the black swan by Nassim Taleb, Give & Take by Adam Grant, and others of the sort.

There's something there about waiting for serendipity, and chasing it. The string shouldn't be too tight, neither too loose.

Best of luck in your journey!

I'm not even clear on what it means, technical debt is non-deterministic in many cases.

To say it differently: if you wrote code that was perfect in time 0, that code may become legacy in time 100.

Are they saying you should continuously refactor all your code to cover the 'current user needs'?

I just think it's an oversimplification for those cases where you don't mind not covering the 0,001% of use cases.

We are working on desplega.ai, and we are toying with this long term vision https://www.pleasedontdeploy.com/p/the-trap-we-automated-cod...

I would love feedback on the post, how do you see the tech stack for software control evolving into this new category.

In my limited experience, every team has a way to track if 'the system is working according to the business specs' beyond actual traces, or QA.

To simplify it, in an e-commerce you would like to verify your inventory goes down according to your purchases, or the delivery orders get created when/how they should, so you are not shipping to the wrong address. Or in a BNPL solution, if the installments were generated for a payment and/or subsequently those were charged.

Usually these are the bugs difficult to find, and teams end up building custom solutions around it (either scripts, or internal systems dedicated to this). There are some formal, advanced, frameworks for this, but they require to build your infrastructure around those. Basically, you will have a team dedicated to that domain, and they'll build robust infrastructure around the problem.

I'm wondering what can be used/done in this space now that vibe-coding is picking up, and actually adding 'code breadcrumbs' is easier than ever using LLMs.

PS: Originally, the concept was tied to testing, my appreciation is that such view is limiting the actual problem, as tests are a point in time, where what you actually want is systematic validation of some invariants.

<rant> How many people are running LLMs CLIs instead of using their APIs? It seems so obnoxious to me that using a CLI command is cheaper than using their APIs, hence forcing them to build these kind of work arounds.

Maybe I'm not getting it right, but it seems there are two competing paradigms which certainly with llms coding for llms, who cares. </rant>

Who has a better blue-print on what would actually mean to have ephemeral code, as we do with assembly nowadays, and instead have a repo with prompts to generate your software?

I've been looking into frameworks like https://github.com/Hey-Diga/imploid during the past couple of weeks, and I'm trying to find out the most promising one out there.

I found out for me 90%+ of the costs were due to their testing functionality that seems extremely expensive.

I would ask myself what am I actually paying for here. -- as mentioned in other comments, they could always have a peer next to them during a call, so hallucinations won't do --

+ Using AI is actually cheating or being productive for the role? + Am I worried that they'll do all their job in 5 minutes and afterwards do something else?

Maybe you are worried about them not being able to actually do the job, which probably means the interview process was wrong from the start. Alternatively, the performance expectations may be higher for the role; e.g. what before was 1x now needs to be 5x productivity.

As an alternative, I've heard of many SMBs opting for a model in which the last bit of the hiring process includes some paid work for a week to see how they actually perform, or checking references in depth.