LLMs and LLM providers are massive black boxes. I get a lot of value from them and so I can put up with that to a certain extent, but these new "products"/features that Anthropic are shipping are very unappealing to me. Not because I can't see a use-case for them, but because I have 0 trust in them:
- No trust that they won't nerf the tool/model behind the feature
- No trust they won't sunset the feature (the graveyard of LLM-features is vast and growing quickly while they throw stuff at the wall to see what sticks)
- No trust in the company long-term. Both in them being around at all and them not rug-pulling. I don't want to build on their "platform". I'll use their harness and their models but I don't want more lock-in than that.
If Anthropic goes "bad" I want to pick up and move to another harness and/or model with minimal fuss. Buying in to things like this would make that much harder.
I'm not going to build my business or my development flows on things I can't replicate myself. Also, I imagine debugging any of this would be maddening. The value add is just not there IMHO.
EDIT: Put another way, LLM companies are trying to climb the ladder to be a platform, I have zero interest in that, I was a "dumb pipe", I want a commodity, I want a provider, not a platform. Claude Code is as far into the dragon's lair that I want to venture and I'm only okay with that because I know I can jump to OpenCode/Codex/etc if/when Anthropic "goes bad".
> Not because I can't see a use-case for them, but because I have 0 trust in them
> […]
> Put another way, LLM companies are trying to climb the ladder to be a platform, I have zero interest in that, I was a "dumb pipe", I want a commodity, I want a provider, not a platform.
That is my sentiment precisely, and a big reason why I’ve started moving away from Claude Code in the past few weeks when I realised how much of my workflow was becoming tied to their specific tools.
Claude Code’s "Memory" feature was the tipping point for me, with the model committing feedbacks and learnings to some local, provider-specific path, that won’t persist in the git repo itself.
That’s fine for user preferences, not for workflows, rules, etc.
And the latest ToS changes about not being allowed to even use another CLI made up my mind. At work we were experimenting with an autonomous debug agent using the Claude Code cli programmatically in ephemeral VMs. Now it just returns an error saying we can’t use subscriptions with third-party software… when there is no third-party software involved?
Think another way, these product features are easy to build in other harnesses too. And as the open source models and the other models which are much lower cost are getting better, there will be a time when it will be justified to have a harness that can work with many models and optimize your cost and efficiency.
> Claude Code’s "Memory" feature was the tipping point for me, with the model committing feedbacks and learnings to some local, provider-specific path, that won’t persist in the git repo itself.
It's a bit annoying, but as long as it's local and human (or LLM) readable, you can use your favourite agent to rework this stuff for itself.
They can’t allow third party software because the third parties save the outputs of Claude responses and distill them into new models to compete with Claude.
This echoes my thoughts exactly. I've tried to stay model-agnostic but the nudges and shoves from Anthropic continue to make that a challenge. No way I'm going that deep into their "cloud" services, unless it's a portable standard. I did MCP and skills because those were transferrable.
I also clearly see the lock-in/moat strategy playing out here, and I don't like it. It's classic SV tactics. I've been burned too many times to let it happen again if I can help it.
This is a similar sentiment I heard early on in the cloud adoption fever, many companies hedged by being “multi cloud” which ended up mostly being abandoned due to hostile patterns by cloud providers, and a lot of cost. Ultimately it didn’t really end up mattering and the most dire predictions of vendor lock in abuse didn’t really happen as feared (I know people will disagree with this, but specifically speaking about aws, the predictions vs what actually happened is a massive gap. note I have never and will never use azure, so I could be wrong on that particular one).
I see people making similar conclusions about various LLM providers. I suspect in the end it’ll shake out about the same way, the providers will become practically inoperable with each other either due to inconvenience, cost, or whatever. So I’ve not wasted much of my time thinking about it.
This sounds like someone complaining about how Windows is a black box while ignoring the existence of Linux/BSD.
I'm currently hosting, on very reasonable consumer grade hardware, an LLM that is on par performance wise what every anyone was paying for about a year ago. Including all the layers in between the model and the user.
Llama.cpp serves up Gemma-4-26B-A4B, Open WebUI handles the client details: system prompt, web search, image gen, file uploading etc. With Conduit and Tailscale providing the last layer so I can have a mobile experience as robust as anything I get from Anthropic, plus I know how all the pieces works and can upgrade, enhance, etc to my hearts delight. All this runs from a pretty standard MBP at > 70 tokens/sec.
If you want to better understand the agent side of things, look into Hermes agent and you can start understanding the internals of how all this stuff is done. You can run a very competitive coding agent using modest hardware and open models. In a similar note, image/video gen on local hardware has come a long way.
Just like Linux, you're going to exchanging time for this level of control, but it's something anyone who takes LLMs seriously and has the same concerns can easily get started with.
Yet I still see comments like this that seem to complete ignore the incredible work in the open model community that has been perpetually improving and is starting to really be competitive. If you relax the "local" requirement and just want more performance from an LLM backend you can replace the llama.cpp part with a call to Kimi 2.5 or Minimax 2.7 (which you could feasibly run at home, not kimi though). You can still control all the additional part of the experience but run models that are very competitive with current proprietary SoTA offering, 100% under your control still and a fraction of the price.
> I want to pick up and move to another harness and/or model with minimal fuss. Buying in to things like this would make that much harder.
Yes, I expect that is very much the point here. A bunch of product guys got on a whiteboard and said, okay the thing is in wide use but the main moat is that our competitors are even more distrusted in the market than we are; other than that it's completely undifferentiated and can be swapped out in a heartbeat for multiple other offerings. How do we do we persuade our investors we have a locked in customer base that won't just up-stakes in favour of other options or just running open source models themselves?
In my view, lock-in anxiety is a holdover from a previous era of tech platforms, and it doesn't really apply in an era where frontier agents can migrate you between vendors in hours. So I personally don't see any good worrying about this. On top of that, every major LLM provider is rapidly converging on the same feature set. They watch each other and clone what works. So the "platform" you're building on isn't really Anthropic's platform so much as it is the emerging shared surface area of what LLMs can do. By the time this Routines feature becomes a problem for you, other solutions will have emerged, and I'd be very surprised if you couldnt lift-and-shift very quickly.
The good news is that, apart from the models themselves, we don't need much from these companies:
- Use Opencode and other similar open-source solutions in place of their proprietary harnesses. This isn't very practical right now because of the heavily subsidized subscriptions that are hard to compete with. But subsidies will end soon, and with progress in inference, it should be very doable to work with open-source clients in the near future.
- Use Openrouter and similar to abstract the LLM itself. That makes AI companies interchangeable and removes a lot of any moat they might have.
I also don't see the value add here... "schedule" is just a cron. "GitHub Event" is probably a 20-minute integration, which Claude itself can write for you.
Maybe there's something I'm not seeing here, but I never want to outsource something so simple to a live service.
I think it behooves us to be selective right now. Frontier labs maybe great at developing models, but we shouldn't assume they know what they are doing from a product perspective. The current phase is throwing several ideas on the wall and see what sticks (see Sora). They don't know how these things will play out long term. There is no reason to believe Co-work/Routines/Skills will survive 5 years from now. So it might just be better to not invest too much in ecosystem upfront.
> I want a commodity, I want a provider, not a platform
That is exactly what the big LLM providers are trying to prevent. Them being only commodity providers might lead them to be easily replaced, and will likely lead to lower margins compared to "full feature" enterprise solutions. Switching LLM API provider is next to no work the moment a competitor is slightly cheaper/better.
Full solutions are more "sticky", harder to replace, and can be sold at higher prices.
> I'm not going to build my business or my development flows on things I can't replicate myself.
but you can replicate these yourself! i'm happy that ant/oai are experimenting to find pmf for "llm for dev-tools". After they figure out the proper stickyness, (or if they go away or nerf or raise prices, etc) you can always take the off-ramp and implement your own llm/agent using the existing open-source models. The cost of building dev-tools is near zero. it is not like codegen where you need the frontier performance.
I am still using the chat completion APIs exclusively. I tried the agent APIs and they're way too opinionated for me. I can see 100% of the tokens I am paying for with my current setup.
I have heard it said that tokens will become commodities. I like being able to switch between Open AI and Anthropics models, but I feel I'd manage if one of them disappeared. I'd probably even get by with Gemini. I don't want to lock in to any one provider any more than I want to lock in to my energy provider. I might pay 2x for a better model, but no more, and I can see that not being the case for much longer.
Every company is trying to become THE platform where all other tools connect to. Notion is integrating everything under the sun, as is Slack, big LLM providers have one-click MCP installation for all major services.
But... these are the "retail" tools that they sell to people organisations without the skills or knowhow to build a basic agentic loop by themselves. Complaining about these being bad and untrustworthy is like comparing a microwave dinner to something you cook yourself. Both will fill your belly equally. One requires zero skill from the user and the second one is 90% skill and 10% getting the right ingredients.
Creating a simple MVP *Claw with tool calling using a local model like gemma4 is literally a 15 minute thing. In 2-3 hours you can make it real pretty. If you base it on something like pi.dev, you can make it easily self-modifying and it can build its own safeguards.
That's all this "routines" thing is, it's just an agentic loop they launch in their cloud on a timer. Just like the scheduled tasks in Claude Cowork.
They have to become a platform because that is their only hope of locking in customers before the open models catch up enough to eat their lunch. Stuff like Gemma is already good enough to replace ChatGPT for the average consumer, and stuff like GLM 5.1 is not too far off from replacing Claude/Codex for the average developer.
We might be building something up your alley! I wanted an OSS platform that let me run any coding agent (or multiple agents) in a sandbox and control it either programmatically or via GUI / TUI.
In this regard, the release of open-weight Gemma models that can run on reasonable local hardware, and are not drastically worse than Anthropic flagships, is quite a punch. An M2 Mac Mini with 32GB is about 10 months worth of Claude Max subscription.
I found your response interesting, I've been working on a tool that is trying to tackle the problem I think you're describing. It's a CLI tool that sits between you and whatever agent you're using — your context lives in plain Markdown files on your machine, git-backed, portable across Claude Code, Codex, Cursor, whatever. You own it. Switch between providers and it comes with you. Happy to share more, we're only starting to share it now. Here's our site: https://www.fathym.com/
I'm glad I'm not the only one that feels this way. I've been creating a local first open source piece of software that lets me spin up different agent harnesses with different runtimes. I call it Major Tom because I wanted to be set free from the imprisonment of Claude Code after their DMCA aggression for their own leak and actions leading to lock down from open source adoption.
Don't put all your eggs in one basket has be true for me and my business for ages.
I could really use the open source community to help make this a reality so I'll release this soon hopefully to positive reception from others who want a similar path forward.
Anthropic wants a moat, but that ship has sailed. Now all I keep reading about is: token burn, downtime and... Wait for it, another new product!
Anthropic thinks they are pulling one over on the enterprise, and maybe they are with annual lock-in akin to Microsoft. But I really hope enterprise buyers are not this gullible, after all these years. At least with Microsoft the product used to be tangible. Now it's... Well, non-deterministic and it's clear providers will gimp models at will.
I had a Pro Max account only for a short period of time and during that short stint Anthropic changed their tune on how I could use that product, I hit limits on a Max account within hours with one CC agent, and experienced multiple outages! But don't worry, Anthropic gave me $200 in credits for OpenClaw. Give me a break.
The current state of LLM providers is the cloud amplified 100x over and in all the worst ways. I had hopes for Anthropic to be the least shitty but it's very clear they've embraced enshittification through and through.
Now I'm spending time looking at how to minimize agent and LLM use with deterministic automation being the foundation with LLM use only where need be and implemented in simple and cost controllable ways.
I think AI labs are realizing that they no longer have any competitive advantage other than being the incumbents. Plus hardware improvements might render their models irrelevant for most tasks.
I've had so many websites break and die because Google or Amazon sunsetted something.
For example I had a graphing calculator website that had 250K monthly active users (mostly school students, I think) and it just vanished one day because Amazon sunsetted EC2 clasic and I didn't have time to deal with that. Hopefully those students found something else to do their homework with that day.
I agree with your analysis. Platforms are some of the most profitable business models because they come with vendor lock-in, but they are always shittier on the long run compared to commodities. Platforms are a way for companies to capture part of the market and decrease competition by increasing the cost of changing vendors.
Also, remember the code quality in the accidental Claude Code source publishing? Expect that for all their features. Thinking about having to debug automations hidden by their SaaS gives me the shudders.
It all went downhill from the moment they changed Reading *.* to reading (*) files.
I can’t use Claude Code at all anymore, not even for simple tasks. The output genuinely disgusts me. Like a friend who constantly stabs you in the back.
My favorite AI feature at the moment is the JetBrains predict next edit. It‘s so fast that I don’t lose attention and I’m still fully under control.
The framing is off. AI is a tool that can operate as a human. GOV is how the humans are organized. AI can basically scale GOV. That’s the paradigm shift. Provenance is durable. AI is just the first opportunity we have had to make it scaleable.
I fully endorse building a custom stack (1) because you will learn a lot (2) for full control and not having Big Ai define our UX/DX for this technology. Let's learn from history this time around?
Without getting too pedantic for no reason… I think it’s important to not call this an LLM.
This isn’t an LLM. It’s a product powered by an LLM. You don’t get access to the model you get access to the product.
An LLM can’t do a web search, an LLM can’t convert Excel files into something and then into PDF. Products do that.
I think it’s a mistake to say I don’t trust this engine to get me here, rather than it is to say I don’t trust this car. Because for the most part, the engine, despite giving you a different performance all the time is roughly doing the same thing over and over.
The product is the curious entity you have no control over.
I'm a little confused on the ToS here. From what I gathered, running claude -p on cron is fine, but putting it in my Telegram bot is a ToS violation (unless I use per-token billing) because it's a 3rd party harness, right? (claude -p being a trivial workaround for the "no 3rd party stuff on the subscription" rule)
This Routines feature notably works with the subscription, and it also has API callbacks. So if my Telegram bot calls that API... do I get my Anthropic account nuked or not?
Unrelated, but Claude was performing so tragically last few days, maybe week(s), but days mostly, that I had to reluctantly switch. Reluctantly because I enjoy it. Even the most basic stuff, like most python scripts it has to rerun because of some syntax error.
The new reality of coding took away one of the best things for me - that the computer always just does what it is told to do. If the results are wrong it means I'm wrong, I made a bug and I can debug it. Here.. I'm not a hater, it's a powerful tool, but.. it's different.
Given the alleged recent extreme reduction in Claude Code usage limits (https://news.ycombinator.com/item?id=47739260), how do these more autonomous tools work within that constraint? Are they effectively only usable with a 20x Max plan?
EDIT: This comment is apparently [dead] and idk why.
You'd think that if they were compute-limited ... Trying to get people to use it less ... The rational thing to do would be to not ship features that will use more compute automatedly? Or does this use extra usage?
I've been using it for a while (it was just called "Scheduled", so I assume this is an attempt to rebrand it?)
It was a bit buggy, but it seems to work better now. Some use cases that worked for me:
1. Go over a slack channel used for feedback for an internal tool, triage, open issues, fix obvious ones, reply with the PR link. Some devs liked it, some freaked out. I kept it.
2. Surprisingly non code related - give me a daily rundown (GitHub activity, slack messages, emails) - tried it with non Claude Code scheduled tasks (CoWork) not as good, as it seems the GitHub connector only works in Claude Code. Really good correlation between threads that start on slack, related to email (outlook), or even my personal gmail.
I can share the markdowns if anyone is interested, but it's pretty basic.
Are they going to mirror every tool software engineers were used to for decades, but in a mangled/proprietary form?
I think to become really efficient they'll have to invent new programming language to eliminate all the ambiguity and non-determinism. Call it "prompt language", with ai-subroutines, ai-labels and ai-goto.
This is the beginning of AI clouds in my estimation. Cloud services provide needed lock-in and support the push to provide higher level services over the top of models. It just makes sense, they'll never recoup the costs on just inference.
Anthropic is burning their good will faster than the tokens we use these days. It is hard to be excited about these new features when the core product has been neutered into oblivion.
The reason someone would use this vs. third-party alternatives is still the fact that the $200/mo subscription is markedly cheaper than per-token API billing.
Not sure how this works out in the long term when switching costs are virtually zero.
I’m moving away from Claude for anything complicated. It’s got such nice DX but I can’t take the confident flaky results. Finding Codex on the high plan more thorough, and for any complicated project that’s what I need.
Still using Claude for UX (playgrounds) and language. OpenAI has always been a little more cerebral and stern, which doesn’t suit those areas. When it tries to be friendly it comes off as someone my age trying to be a 20-something.
I used the claude-code-action GitHub Action to review PRs before, but it is pretty buggy e.g. PRs from forked repositories do not work, and I had to fix it myself. This should work better with Claude Code Routines. claude-code-action only works with the API and is therefore pretty expensive compared to the subscription.
I think LLM reviews on PRs are helpful and will reduce the load on maintainers. I am working on OpenWrt and was approved for the Claude Code Max Open Source Program today. The cap of 15 automatic Claude Code Routines runs per day is a bit low. We get 5 to 20 new PRs per day and I would like to run it on all of them. I would also like to re-run it when authors make changes, in that case it should be sufficient to just check if the problems were addressed.
Is it possible to get more runs per day, or to carry over unused ones from the last 7 days? Maybe 30 on Sonnet and 15 on Opus?
When I was editing a routine, the window closed and showed an error message twice. Looks like there are still some bugs.
Everything is big race! Each company is trying to do as much as possible, to provide as many tools as possible, to catch the wave and beat the concurrency. I remember how Antropic and OpenAI made releases in just 10-15 minutes of difference, trying to compete and gain momentum.
And because they use AI heavily, they produce new product every week. So fast, that I have no time to check, does it worth or not.
This one looks interesting. I have some custom commands that I execute manually weekly, for monitoring, audits, summary, reports.
It it can send reports on email, or generate something that I can read in the morning with my coffee, or after I finish with it ;) it might be a good tool.
The question is, do I really want to so much productive? I am already much better in performance with AI, compared with the 'old school' way...
It seemed OpenClaw is just Pi with Cron and hooks, and it seems like this is just Claude Code with Cron and hooks. Based on the superiority of Pi, I would not expect this to attract any one from OpenClaw, but it will increase token usage in Claude Code.
It’s interesting to watch Ant try to ship every value-add product feature they can while they still have the SOTA model for agentic. When an open weights equivalent to Opus 4.5’s agentic capabilities comes out, I expect massive shifts of workloads away from Claude.
Don’t get me wrong, I think their business model is still solid and they will be able to sell every token they can generate for the next couple years. They just won’t be critical path for AI diffusion anymore, which will be good for all sides.
The trigger matrix here is actually the most interesting part. Schedule plus API plus GitHub event on the same routine unlocks some nice patterns, and the /fire endpoint returning a session URL means you can wire this into alerting tools or a CD pipeline from almost anywhere. The part that is not really covered in the docs is what state a routine is supposed to recover from if a previous run died halfway through a repository change. The protection around claude/-prefixed branches helps you not clobber main, but it does not tell the next run what the previous run actually finished. I run scheduled jobs against multi step pipelines on my own infra and the failure mode that bites is not the crash, it is the run that returned success while a downstream side effect quietly broke. The /fire response returns a session URL and a session ID, which tells you the routine started, but how is a routine expected to notice when the downstream thing it kicked off (a CD pipeline, an alert follow-up, a library port PR like the one in the examples quietly fell over five minutes after the session ended?
Claude and Open AI seems to be trying not to be 'Just a model', but this is intrinsically problematic because model can be degraded and prices only goes up once they lock-in customers.
It is increasingly important for anyone who are responsible of managing 'AI workflows' to keep the sovereignty about how you use AI models.
This is why I'm super excited in building the local-first workflow orchestration software called "Dagu", that allows us to own your harness on your own. It's not only more cost-effective, but outcome is better as well because you have 100% full control. I think it's only matter of time that people notice they need to own their workflow orchestration on their own not relying on Anthropic, OpenAI, or Google.
This is one of the best features of OpenClaw - makes sense to swipe it into Claude Code directly. I wonder if Anthropic wants to just make claude a full stand-in replacement for openclaw, or just chip away at what they think the best features are, now that oAI has acquired.
Yeah the only issue is that it is a major vendor lock-in + hazy perspective, so it is essentially unfit for any "serious" processes (read: mission critical / where you have to invest time and effort). I guess you could simply copy-paste the prompt from there in case of emergency. That is, if they allow you to.
Other than that it is a really neat utility and saves some time. However, I feel like there can be OpenClaw moment for this as well. Somebody simply needs to create an open-source version that supports all of the agents and models.
This wild, one of the pieces I was lacking for a very openclaw-esque future. Now I think I have all the mcp tools I need (github, linear, slack, gmail, querybear), all the skills I need, and now can run these on a loop.
I don't get the use case for these... Their primary customers are enterprises. Are most enterprises happy with running daily tasks on a third party cloud outside of their ecosystem? I think not.
Why not just do event based triggers e.g. register (web)hooks instead of schedules time based triggers. Have a mechanism to listen to an event and then run some flow - analyze, plan, execute, feedback
Did this sort of a thing in my own macos app which can have routines with a cron, custom configs and chains of prompts. There is also more like custom VMs and models to be used for different tasks. Interesting to see larger providers trying to do the same.
But their own failure is the fact that there is a limited way to configure it with other models, think 3d modelling and integrating 3d apps on a VM to work with. I believe an OSS solution is needed here, which is not too hard to do either.
Having used the cowork version of this: scheduled automations. I have very little confidence in this from Anthropic. 90% of the time the automation never even runs.
Am I crazy in thinking an LLM doing any kind of serious workload is risky as hell?
Like, say it works today, but tomorrow they update the model and instead of emailing you an update it emails your api keys to all your contacts? Or if it works 999 times out of 1000 but then commits code to master that makes all your products free?
Idk man… call me Adama, but i do not trust long-running networked ai one bit
I have a small team of 4 engineers, each of us is on the personal max subscription plan and prefer to stay this way to save cost.
Does anyone know how I can overcome the challenge with setting up Routines or Scheduled Tasks with Anthropic infra in a collaborate manner: ie: all teammates can contribute to these nightly job of cleaning up the docs, cleaning up vibe coding slops.
The docs list the GitHub events that can be used as triggers. This is included in the list:
Push Commits are pushed to a branch
But when I try to create a routine, the only GitHub events available in the drop down related to pull requests and releases. Nothing available related to pushes/commits or issues. Am I holding it wrong?
One gripe I have with Claude Code is that the CLI, Desktop app, and apparently the Webapp have a Venn Diagram of features. Plugins (sets of skills and more) are supported in Code CLI, maybe in Cowork (custom fail to import) but not Code Desktop. Now this?
The report that they are 90% Ai code generated seems more likely the more I attempt to use their products.
Not saying it doesn’t look useful, but it’s something that keeps you from ever switching off Claude.
Next year, if Claude raises rates after getting bought by Google… what then?
And what happens when Claude goes down and misses events that were supposed to trigger Routines? I’m not at the point where I trust them to have business-dependable uptime.
I don't get all the hype around these kinds of things, it's not something that crazy that when it gets released you solve some kind of problem you had. Those are things that can easily be replicated and that many of us already built for ourselves months ago, before those were published and became mainstream
I wish they'd release more stuff that didn't rely on me routing all my data through their cloud to work. Obviously the LLM is cloud based but I don't want any more lock-in than that. Plus not everyone has their repositories in GitHub.
Two things about My experience, first you only have one at a time per suscription, if you need implement two at the same time i could not able to. The second is that you can do that with a well configured cron.
413 comments
- No trust that they won't nerf the tool/model behind the feature
- No trust they won't sunset the feature (the graveyard of LLM-features is vast and growing quickly while they throw stuff at the wall to see what sticks)
- No trust in the company long-term. Both in them being around at all and them not rug-pulling. I don't want to build on their "platform". I'll use their harness and their models but I don't want more lock-in than that.
If Anthropic goes "bad" I want to pick up and move to another harness and/or model with minimal fuss. Buying in to things like this would make that much harder.
I'm not going to build my business or my development flows on things I can't replicate myself. Also, I imagine debugging any of this would be maddening. The value add is just not there IMHO.
EDIT: Put another way, LLM companies are trying to climb the ladder to be a platform, I have zero interest in that, I was a "dumb pipe", I want a commodity, I want a provider, not a platform. Claude Code is as far into the dragon's lair that I want to venture and I'm only okay with that because I know I can jump to OpenCode/Codex/etc if/when Anthropic "goes bad".
> Not because I can't see a use-case for them, but because I have 0 trust in them
> […]
> Put another way, LLM companies are trying to climb the ladder to be a platform, I have zero interest in that, I was a "dumb pipe", I want a commodity, I want a provider, not a platform.
That is my sentiment precisely, and a big reason why I’ve started moving away from Claude Code in the past few weeks when I realised how much of my workflow was becoming tied to their specific tools.
Claude Code’s "Memory" feature was the tipping point for me, with the model committing feedbacks and learnings to some local, provider-specific path, that won’t persist in the git repo itself.
That’s fine for user preferences, not for workflows, rules, etc.
And the latest ToS changes about not being allowed to even use another CLI made up my mind. At work we were experimenting with an autonomous debug agent using the Claude Code cli programmatically in ephemeral VMs. Now it just returns an error saying we can’t use subscriptions with third-party software… when there is no third-party software involved?
Anyway, so long Claude.
> Claude Code’s "Memory" feature was the tipping point for me
My standing orders are the default MEMORY.md must be a stub directing Claude to another MEMORY.md file in the local folder, project, etc.
All memories remain with their respective projects over syncs, moves, devices, etc. The stub must state all this clearly, and nothing else.
This has worked very well.
If you give the model/memory a name, that name can be persistent and independent over "backend" model swaps.
> Claude Code’s "Memory" feature was the tipping point for me, with the model committing feedbacks and learnings to some local, provider-specific path, that won’t persist in the git repo itself.
It's a bit annoying, but as long as it's local and human (or LLM) readable, you can use your favourite agent to rework this stuff for itself.
Both give you optionality because they support N models.
I also clearly see the lock-in/moat strategy playing out here, and I don't like it. It's classic SV tactics. I've been burned too many times to let it happen again if I can help it.
I see people making similar conclusions about various LLM providers. I suspect in the end it’ll shake out about the same way, the providers will become practically inoperable with each other either due to inconvenience, cost, or whatever. So I’ve not wasted much of my time thinking about it.
> - No trust that they won't nerf the tool/model behind the feature
To the contrary, they've proven again and again and again they'll absolutely do that the first chance they get.
I'm currently hosting, on very reasonable consumer grade hardware, an LLM that is on par performance wise what every anyone was paying for about a year ago. Including all the layers in between the model and the user.
Llama.cpp serves up Gemma-4-26B-A4B, Open WebUI handles the client details: system prompt, web search, image gen, file uploading etc. With Conduit and Tailscale providing the last layer so I can have a mobile experience as robust as anything I get from Anthropic, plus I know how all the pieces works and can upgrade, enhance, etc to my hearts delight. All this runs from a pretty standard MBP at > 70 tokens/sec.
If you want to better understand the agent side of things, look into Hermes agent and you can start understanding the internals of how all this stuff is done. You can run a very competitive coding agent using modest hardware and open models. In a similar note, image/video gen on local hardware has come a long way.
Just like Linux, you're going to exchanging time for this level of control, but it's something anyone who takes LLMs seriously and has the same concerns can easily get started with.
Yet I still see comments like this that seem to complete ignore the incredible work in the open model community that has been perpetually improving and is starting to really be competitive. If you relax the "local" requirement and just want more performance from an LLM backend you can replace the llama.cpp part with a call to Kimi 2.5 or Minimax 2.7 (which you could feasibly run at home, not kimi though). You can still control all the additional part of the experience but run models that are very competitive with current proprietary SoTA offering, 100% under your control still and a fraction of the price.
> I want to pick up and move to another harness and/or model with minimal fuss. Buying in to things like this would make that much harder.
Yes, I expect that is very much the point here. A bunch of product guys got on a whiteboard and said, okay the thing is in wide use but the main moat is that our competitors are even more distrusted in the market than we are; other than that it's completely undifferentiated and can be swapped out in a heartbeat for multiple other offerings. How do we do we persuade our investors we have a locked in customer base that won't just up-stakes in favour of other options or just running open source models themselves?
> - No trust that they won't nerf the tool/model behind the feature
I actually trust that they will.
Too bad we've now managed to turn programming into the same annoying guesswork.
- Use Opencode and other similar open-source solutions in place of their proprietary harnesses. This isn't very practical right now because of the heavily subsidized subscriptions that are hard to compete with. But subsidies will end soon, and with progress in inference, it should be very doable to work with open-source clients in the near future.
- Use Openrouter and similar to abstract the LLM itself. That makes AI companies interchangeable and removes a lot of any moat they might have.
Maybe there's something I'm not seeing here, but I never want to outsource something so simple to a live service.
> I want a commodity, I want a provider, not a platform
That is exactly what the big LLM providers are trying to prevent. Them being only commodity providers might lead them to be easily replaced, and will likely lead to lower margins compared to "full feature" enterprise solutions. Switching LLM API provider is next to no work the moment a competitor is slightly cheaper/better.
Full solutions are more "sticky", harder to replace, and can be sold at higher prices.
> I'm not going to build my business or my development flows on things I can't replicate myself.
but you can replicate these yourself! i'm happy that ant/oai are experimenting to find pmf for "llm for dev-tools". After they figure out the proper stickyness, (or if they go away or nerf or raise prices, etc) you can always take the off-ramp and implement your own llm/agent using the existing open-source models. The cost of building dev-tools is near zero. it is not like codegen where you need the frontier performance.
But... these are the "retail" tools that they sell to people organisations without the skills or knowhow to build a basic agentic loop by themselves. Complaining about these being bad and untrustworthy is like comparing a microwave dinner to something you cook yourself. Both will fill your belly equally. One requires zero skill from the user and the second one is 90% skill and 10% getting the right ingredients.
Creating a simple MVP *Claw with tool calling using a local model like gemma4 is literally a 15 minute thing. In 2-3 hours you can make it real pretty. If you base it on something like pi.dev, you can make it easily self-modifying and it can build its own safeguards.
That's all this "routines" thing is, it's just an agentic loop they launch in their cloud on a timer. Just like the scheduled tasks in Claude Cowork.
Website is https://amika.dev
And part of our code is OSS (https://github.com/gofixpoint/amika) but we're working on open sourcing more of it: https://docs.google.com/document/d/1vevSJsSCWT_reuD7JwAuGCX5...
We've been signing up private beta users, and also looking for feedback on the OSS plans.
Don't put all your eggs in one basket has be true for me and my business for ages.
I could really use the open source community to help make this a reality so I'll release this soon hopefully to positive reception from others who want a similar path forward.
Anthropic wants a moat, but that ship has sailed. Now all I keep reading about is: token burn, downtime and... Wait for it, another new product!
Anthropic thinks they are pulling one over on the enterprise, and maybe they are with annual lock-in akin to Microsoft. But I really hope enterprise buyers are not this gullible, after all these years. At least with Microsoft the product used to be tangible. Now it's... Well, non-deterministic and it's clear providers will gimp models at will.
I had a Pro Max account only for a short period of time and during that short stint Anthropic changed their tune on how I could use that product, I hit limits on a Max account within hours with one CC agent, and experienced multiple outages! But don't worry, Anthropic gave me $200 in credits for OpenClaw. Give me a break.
The current state of LLM providers is the cloud amplified 100x over and in all the worst ways. I had hopes for Anthropic to be the least shitty but it's very clear they've embraced enshittification through and through.
Now I'm spending time looking at how to minimize agent and LLM use with deterministic automation being the foundation with LLM use only where need be and implemented in simple and cost controllable ways.
If you can define good enough for you and local llms can meet that you'll get:
- no vendor lock-in (control)
- price
- stability (you decide when to hot swap with newer models)
- speed (control)
- full observability and predictability.
- Privacy / Data Locality (Depending on implementation of infrastructure).
- [1] https://alexhans.github.io/posts/series/evals/measure-first-...
> No trust they won't sunset the feature
I've had so many websites break and die because Google or Amazon sunsetted something.
For example I had a graphing calculator website that had 250K monthly active users (mostly school students, I think) and it just vanished one day because Amazon sunsetted EC2 clasic and I didn't have time to deal with that. Hopefully those students found something else to do their homework with that day.
I can’t use Claude Code at all anymore, not even for simple tasks. The output genuinely disgusts me. Like a friend who constantly stabs you in the back.
My favorite AI feature at the moment is the JetBrains predict next edit. It‘s so fast that I don’t lose attention and I’m still fully under control.
Building business on top of SaaS products, iPaaS integrations, and serverless middleware.
This isn’t an LLM. It’s a product powered by an LLM. You don’t get access to the model you get access to the product.
An LLM can’t do a web search, an LLM can’t convert Excel files into something and then into PDF. Products do that.
I think it’s a mistake to say I don’t trust this engine to get me here, rather than it is to say I don’t trust this car. Because for the most part, the engine, despite giving you a different performance all the time is roughly doing the same thing over and over.
The product is the curious entity you have no control over.
claude -pon cron is fine, but putting it in my Telegram bot is a ToS violation (unless I use per-token billing) because it's a 3rd party harness, right? (claude -pbeing a trivial workaround for the "no 3rd party stuff on the subscription" rule)This Routines feature notably works with the subscription, and it also has API callbacks. So if my Telegram bot calls that API... do I get my Anthropic account nuked or not?
The new reality of coding took away one of the best things for me - that the computer always just does what it is told to do. If the results are wrong it means I'm wrong, I made a bug and I can debug it. Here.. I'm not a hater, it's a powerful tool, but.. it's different.
EDIT: This comment is apparently [dead] and idk why.
We ought to come up with a term for this new discipline, eg "software engineering" or "programming"
It was a bit buggy, but it seems to work better now. Some use cases that worked for me:
1. Go over a slack channel used for feedback for an internal tool, triage, open issues, fix obvious ones, reply with the PR link. Some devs liked it, some freaked out. I kept it.
2. Surprisingly non code related - give me a daily rundown (GitHub activity, slack messages, emails) - tried it with non Claude Code scheduled tasks (CoWork) not as good, as it seems the GitHub connector only works in Claude Code. Really good correlation between threads that start on slack, related to email (outlook), or even my personal gmail.
I can share the markdowns if anyone is interested, but it's pretty basic.
Very useful, (when it works).
I think to become really efficient they'll have to invent new programming language to eliminate all the ambiguity and non-determinism. Call it "prompt language", with ai-subroutines, ai-labels and ai-goto.
They support much of the same triggers and come with many additional security controls out of the box
The reason someone would use this vs. third-party alternatives is still the fact that the $200/mo subscription is markedly cheaper than per-token API billing.
Not sure how this works out in the long term when switching costs are virtually zero.
Still using Claude for UX (playgrounds) and language. OpenAI has always been a little more cerebral and stern, which doesn’t suit those areas. When it tries to be friendly it comes off as someone my age trying to be a 20-something.
I think LLM reviews on PRs are helpful and will reduce the load on maintainers. I am working on OpenWrt and was approved for the Claude Code Max Open Source Program today. The cap of 15 automatic Claude Code Routines runs per day is a bit low. We get 5 to 20 new PRs per day and I would like to run it on all of them. I would also like to re-run it when authors make changes, in that case it should be sufficient to just check if the problems were addressed.
Is it possible to get more runs per day, or to carry over unused ones from the last 7 days? Maybe 30 on Sonnet and 15 on Opus?
When I was editing a routine, the window closed and showed an error message twice. Looks like there are still some bugs.
And because they use AI heavily, they produce new product every week. So fast, that I have no time to check, does it worth or not.
This one looks interesting. I have some custom commands that I execute manually weekly, for monitoring, audits, summary, reports. It it can send reports on email, or generate something that I can read in the morning with my coffee, or after I finish with it ;) it might be a good tool.
The question is, do I really want to so much productive? I am already much better in performance with AI, compared with the 'old school' way...
Everything is just getting to much for me.
n8n: https://n8n.io/
Don’t get me wrong, I think their business model is still solid and they will be able to sell every token they can generate for the next couple years. They just won’t be critical path for AI diffusion anymore, which will be good for all sides.
If the Lovable clone is real that’s going to piss off many model consumers out there.
Is Sierra next?
Other than that it is a really neat utility and saves some time. However, I feel like there can be OpenClaw moment for this as well. Somebody simply needs to create an open-source version that supports all of the agents and models.
Am I needed anymore?
So who are they building these for?
> react to GitHub events from Anthropic-managed cloud infrastructure
Oh cool! vendor lock-in.
But their own failure is the fact that there is a limited way to configure it with other models, think 3d modelling and integrating 3d apps on a VM to work with. I believe an OSS solution is needed here, which is not too hard to do either.
It’s fine if it’s a stop gap. But, it’s too inconsistent to ever be reliable.
This PR was created by the Claude Code Routine:
https://github.com/srid/claude-dump/pull/5
The original prompt: https://i.imgur.com/mWmkw5e.png
Like, say it works today, but tomorrow they update the model and instead of emailing you an update it emails your api keys to all your contacts? Or if it works 999 times out of 1000 but then commits code to master that makes all your products free?
Idk man… call me Adama, but i do not trust long-running networked ai one bit
The report that they are 90% Ai code generated seems more likely the more I attempt to use their products.
Not saying it doesn’t look useful, but it’s something that keeps you from ever switching off Claude.
Next year, if Claude raises rates after getting bought by Google… what then?
And what happens when Claude goes down and misses events that were supposed to trigger Routines? I’m not at the point where I trust them to have business-dependable uptime.