Sabitlenmiş Tweet
Zach Lowden
3.7K posts
Zach Lowden
@DevZacx
Infra Lead at Civitai +Building AI Agents for SMBs so we don't get left behind. https://t.co/FDH5E0ZWQV
Winnipeg, Manitoba Katılım Mart 2013
232 Takip Edilen138 Takipçiler
@0x_claudia Western Union will be the winner here, they are the incumbant (100+ years) and switching to stablecoins to earn yield from float (float is billions already) as well as cut out bank fees. Stablecoin starts rolling out this quarter so we'll know soon.
English
I'm doing this, using a versioned generic harness based on openclaw in k8s to address the hurdles you mentioned.
Working ok but only in a handful of local businesses so far.
Everyone uses the same UI, each biz manages the skills/data connections for their agents.
Wrote a bit about it here: x.com/DevZacx/status…
English
Whether it’s existing consulting firms, new ones that emerge, FDEs from agent vendors, or new internal agent engineering roles, the amount of work that is going to be created to implement agents in enterprises will exceed anything we imagine today.
The complexity of implementing agents in any existing organizations is very real. When I talk to large enterprises, as you move from a chat paradigm to agents that participate in meaningful workflows, there are a number of things they need to do.
First, you have to get agents to be able to talk to your data securely across your systems. In many cases, enterprises have decades of legacy infrastructure that contain the valuable context for AI agents. That’s going to take a ton of work to go modernize and move to systems that work well with agents.
Then, you need to ensure that you’ve implemented agents with the right access controls and entitlements, the right scopes to be safely used, and have ways of monitoring, logging, and securing the work that they do.
Next, you need to actually document the processes in the organization in a way that agents can utilize for doing the work. You also need to figure out what the new workflow looks like when agents and people are working together on a process, and who steps in where. Just replicating the old workflow will mute the gains. Oh and you likely need to create evals for your top new end-state processes.
Finally, you have to keep up with a rapidly changing set of best practices and architectural shifts happening in the agent space. While it’s fun for people to change their personal productivity tools on a dime, it’s 100X harder to do this in a business process. The speed of change is a blessing and a curse right now for anyone trying to keep a stable system design.
All of this means that individuals and companies that develop expertise on the above set of components (and more) are going to be needed to help organizations actually implement agents at scale. This is also the rationale for vertical AI agents right now that can go in deep on a business domain and help bring automation to it.
This is a huge opportunity right now whether you’re doing this internally or as an external business provider.
English
The winner in all of this AI token spend isn’t:
1. More revenue growth
2. Less OpEx
It’s:
1. Leisure time - go home early, work on a side gig, play video games etc etc etc
My suspicion is that people are using blower token budgets to do their job in less time. Not so more work in the same time.
This may change but humans are not motivated to do more work and see upside accrue to a few.
Take this to the bank as Wall Street starts demanding specific data about AI enhancements over these next 18 months.
English
I think OpenClaw will end up being an awesome but invisible-to-end-user layer in the AI app stack.
I love it as a generic agent harness building block and trust building on it long term because it works great, is OSS, and doesn't lock me into any specific provider.
That said, as far as it being a consumer-facing thing, I agree that side is probably dying/dead.
@levelsio@levelsio
Update after 2 months, I think Anthropic succesfully killed OpenClaw Back then I said: "The real power user for OpenClaw has become my gf who I put it in a group chat with me, her and OpenClaw, she's mostly stopped using ChatGPT etc and now only uses AI over Telegram with OpenClaw, it's much more user friendly for her in Telegram iOS than ChatGPT's own app" Since I switched it to OpenAI API my gf has since slowly stopped using it and switched back to the Claude app and now uses Projects she says So from my n=1, Anthropic successfully killed OpenClaw
English
Humanity's energy consumption is at the beginning of an acceleration cycle and Canada is uniquely positioned to benefit massively from this, LFG.
Natural Resources@NRCan
Today, we announced the development of a transformative Nuclear Energy Strategy for Canada. Learn more: ow.ly/gXvY50YRMLT #NuclearEnergy
English
Visa is positioning this as no big deal, but its a step towards inevitable disruption.
As consumers and merchants get comfortable with stablecoins they will start to wonder why they dont just put up a qr code and cut out the middle man to save the 3%
It doesnt sound like much but to big retailers like walmart, amazon, etc its literally billions of dollars per year. Plenty of margin to share with the customer to encourage them to switch off visa.
Once one of them does it, the first domino has fallen and the long tail of large and SMB merchants will follow suit.
Base@base
Base is now supported by Visa for stablecoin settlement Our mission is to make onchain the standard, and this partnership is a major step towards it It's time to make stablecoin payments a reality for billions
English
@martin_casado "using agents for long running things organizationally [..], I'm not sure we know how to do that"
We do know how, it's just not evenly distributed yet.
Models are plenty capable, the problem is not one of model intelligence, it's one of context management.
Going from 'long, progressively less-useful chatgpt threads' approach that is currently being used, to an agent that reliably manages a part of my business takes thoughtful harness and skillfile design, followed by plugging in the appropriate model.
A well designed harness decouples both the model AND the chat threads, intellectual capital accrues in the skillfiles backed by a traditional filesystem+DB, usable by entire org rather than in an ephemeral, siloed chat thread
From @a16z :
youtube.com/watch?v=dvVbA9…
YouTube
English
This is happening right now, as a devops eng. producing 10x the code now, and being responsible for deploying it + the 10x amount of code created by team members I can confirm it is indeed playing out in every product I see.
Instead of long code review cycles and a few releases per week, releases happen every few hours, sometimes with 10k+ LOC changes.
This step-up in velocity and complexity means I the human must move up a layer, and instead of it being 'agents manage the infra, we dont need devops', it's 'devops is more critical than ever to now drive the agent swarm managing the rapidly scaling amount of infra and complexity. (human attn. moving to the 'last mile', and it becoming more complex and critical than ever)
This shift has already re-written the dev and devops roles and while it hasn't yet diffused to other professions, it's likely they will experience the same shift from 'human doing 100%' to 'human doing the last mile (incl. driving the agents)' with that last mile becoming increasingly complex, specialized, and critical to businesses.
Aaron Levie@levie
English
@chamath I keep seeing you say this, I am a dev with 15yrs exp and AI has multiplied my earnings.
I take on more contracts than before, at a higher rate of concurrency, and doing them all solo vs before I had a team.
Done w/ a few claude subs (~1k/mo) unlocking (~20k/mo) revenue
English
But very little lift in productivity thus far if true.
unusual_whales@unusual_whales
Over half of the planet’s internet traffic is now made up of AI bots, per Lumen Technologies
English
Manitoba is banning AI for youth. I live there.
While Premier Kinew's instinct is right, the "AI Chatbot Bad" stance is too broad, and will disadvantage Manitoba students if implemented.
The current phrasing lumps together the truly parasitic chatbots (snap, characterai, etc) with ones that deliver
real educational value (ai-powered tutors, learning tools, etc.).
Policy should protect AND empower our kids, this proposal in it's current form will result in kids being educated in a bubble, and graduate into a world and workforce powered by technology they have zero knowledge and exposure to, putting them at a steep disadvantage vs peers and setting up the next generation for failure.
Wab Kinew@WabKinew
Our kids are not for sale. We’re moving forward with a first-in-Canada plan to ban social media and AI chatbots for youth - because these platforms are built to capture attention, not protect it. Childhood should be about learning, growing, and being present. Not endless scrolling designed for profit. We owe the next generation something simple: the freedom to be kids. A lot done. A lot more to do.
English
@awilkinson Kubernetes pod for each with their files, skills, and workspace. Then they are isolated and can scale to multiple instances. Connect to a groupchat with an orchestrator agent and you've got a team.
English
I want to hire someone to build me an team of remote Openclaw employees. Who can help?
🤑 Paid gig.
TLDR: I want - support, engineering team, designer, marketer, etc.
English
@levie As a dev building agents, this was true until DSPy was created. DSPy optimizers eliminate the need for manual prompt engineering.
English
Prompting doesn’t show any signs of letting up as a useful skill. The variability in the quality of what we can get Agents to do for us based on the prompt is super wide. The smarter someone is, the more they can do with good instructions. AI will be no different.
Harj Taggar@harjtaggar
Prompting AI agents to consistently do what you want is becoming the most important skill for founders to learn and build their companies around. We share some of the more advanced techniques we've learned from founders, and building AI agents ourselves at YC, in the latest @LightconePod episode.
English
@shaneparrish @harleyf I'll build it in 48hrs.
- Pull bill text
- Use NLP to surface all important details, cost, implications, etc
- Publish summary on here via api call with a link back to a site where people can see the source/citations in original bill text
English
@RichardRyan By descending grant value: Ethereum, Polygon, Arbitrum, Avalanche, BSC, Optimism, Fantom, Moonbeam, Celo, Gnosis. Dependent on factors such as type of project, traction, active users, etc. If you can share more details via DM I can compile exact list of what you'd qualify for
English
Who are some of the best EVM compatible chains offering grants? I will be launching with a few companies over next couple months. The products are chain agnostic so I want the best collaborative opportunity and or offset of initial investments.
English
@RichardRyan you can check a lot of boxes with a lander + signup with a wallet interaction, that will put you ahead of 90% of others in the eyes of grant programs because you can use your audience to kickstart traction. I'll build this for free in a week and we can quickly see if it will work
English
Glad you got something out of it! Thanks for the kind words 😄👍
Javier Feliu 👋 🇵🇦@JavierFeliuA
This is a fantastic post by @DevZacx at @codewallblog on how to improve your Dockerfiles. It is well explained, has clear code samples, and links to additional references. #100DaysOfCode #Docker #DevOps Writing An Optimized Dockerfile | CodeWall buff.ly/2Rosy8e
English