Craftplan handles recipes (versioned BOMs with cost rollups), inventory (lot traceability, demand forecasting, allergen tracking), orders, production batch planning, and purchasing. Built with Elixir, Ash Framework, Phoenix LiveView, and PostgreSQL.
Live demo: https://craftplan.fly.dev (test@test.com / Aa123123123123)
Go ahead - I'm ready to be down-voted again and again until folks realize it is inevitable, as is inevitable that many companies in the area of business software are going down down down.
I've observed a new trend, managers who are frequently in the wait list started to use AI to generate small local apps. They still rely on my input when it's complex, or when implementation could generate risks or need resilience and would ask for small code reviews when they are unsure of the generated code quality.
The result is win win, I have more time for high value projects the executives want to prioritize, and managers can innovate faster almost on their own.
I wonder if it will have a similar pattern of creating a mess as the app starts to get uptake and the SME can't scale their attention to be an app owner, as well as an SME at the same time.
Also makes me think that an llm-developed-app-friendly shared datastore would be a useful thing to have
While taking ownership of AI slop is not an option for me, I do want to avoid shadow IT.
Might be a moon shot .. could sharing a prompt template with git access to colleagues be a way to enforce it ?
Although if they were HTML/JS/CSS with no dependencies you could argue that might be quite bullet proof since browsers have an unmatched record of backward compatibility.
Also I believe the value is in the output, not the app itself. ROI of a 1 hour AI slop should™ be attained before it rots and they can just spun up a new one.
I might be hopeful and a bit selfish here, I expect my colleagues to own their toolbox.
Excel used to be, and probably still is, the primary competitor to enterprise-developed apps - a lot of businesses run on it. But, that was a locally deployed phenomenon, with an added ability to deploy it somewhere else by simply emailing the workbook to someone else.
In your organization, how do your managers turn their code into working software?
Some users with privileged access can run their app locally in cli or browser, else most softwares we work with can use custom modules in specific languages (html, css, js, dax, vb.net, perl, python, sql, etc.). Ownership and trust must be established, for example only the commercial manager has access to deploy modules to CRM. They usually are constrained to read access, unless they are informed engineering managers.
Ideally I would share pipelines to deploy static pages, or a predefined dynamic architecture. I'm wary the security risks are too great, I don't trust they would have enough time / interest to become autonomous in unconstrained environments so I didn't pursue the idea, maybe I could for static pages, or in isolated networks ..
I have been empowered to be the quick and dirty build guy for small local apps instead of fighting for engineering resources. Now our managers regularly hit me up on slack with small little items that if built, could increase productivity for their teams. I love it. I'm mostly building systems that work with google workspace (docs/sheets/forms/email). So its a lot of little appscripts, or single html file apps made available via google sites. Most of our operations staff dont have all the required underlying expertise to quickly pull this kind of thing off, and are not interested in strengthening those talents. It has allowed me to shine bright while also providing much needed relief to many teams.
i don't think its going to be a silver bullet, but it doesn't need to be. niche, well understood problems with simple tooling needs are the best ones to start with.
https://culturecompiled.com/p/things-are-getting-awkward-for...
ie are the efficiency gains of having something that's exactly tailored to you enough to create a competitive advantage.
It's back to the old idea - of software eating the world.
So for example in the UK - there is a relatively new 'energy' company called Octopus - it's grown and grown and finally overtaken the old established players.
In reality it's not an energy company - it's a software company - that used it's expertise in software to overtake it's energy supplier competitors - it was able to provide innovative products in the market because it controlled it's own software - rather than 'big vendor says no'.
I think it's telling that the founder originally left school at 16 to write computer games, before coming back to do a degree etc.
ie the question is - for any particular industry what's the benefit of custom software. Does a bakery having it's own give it enough of an advantage?
> Does a bakery having it's own give it enough of an advantage? 50 bakeries of the same franchise may benefit from it. But it does not need to be SAP or Dynamics365 or something along this line to work. People been doing business with text-mode AS360's for ages, and nobody complained. Coca Cola was using AS-something for the warehousing in 2004, while it was already discontinued for years.
ERPs and CRMs were highly configurable long before SaaS.
Not sure it is. Unless the Saas company is ripping you off (sure it can happen - but hopefully competition in the market would manage that over time ), then it won't be that much different from your own maintenance costs.
I always think if that's the business case for custom software ( a few quid license cost savings ) then you probably shouldn't be doing it as there is almost always a better ROI case for transformation through custom software.
So back to the bakery case. Is the benefit savings on license costs, or the fact that you can give much better estimates to customers, better de-risk supply chain issues, hire less people to operate, and improve morale via reducing busy work?
All these sort of things have to be more valuable than a few quid on licensing.
Many workflows are about B2B transactions. PO -> Sales Order -> Invoicing workflows. ASNs, etc.
So a lot of workflows are not driven by companies but by the standard operating framework of B2B
ie not sure issues with the smart meters themselves is the fault of Octopus - as the meter standards are set centrally so they can still work if you switch supplier?
Back to another old adage,
"people who are really serious about software should make their own hardware"
Mind you, I couldn't help noticing that the meters themselves are owned by a leasing company... (there is a plate on each one explaining this).
Sure this is awesome now and maybe he shipped it in a week using AI or something, but he now owns a critical part of his wife's business. 5 years from now he is gonna be working 50/hrs a week and not want to deal with this project he barely remembers even doing, whenever an SSL cert goes bad or the CC he was paying the server bills with expires or actual bugs happen he is on the line for it.
It is lame to let family/friends pay $20/mo for something you could build in a few weeks, but they will own the product forever, I don't want to.
[1]: https://finance.yahoo.com/sectors/technology/software-applic...
But in this context, is Uber[9% weight, down ~4% YTD] a transportation company that roles it's own software for competitive advantage? I think other's in the composition are similar. The takeaway is maybe that the tech landscape is changing or LLMs have spooked investors and they're running without direction. But that doesn't necessarily speak to bespoke software uptake (already) cutting into profits(?) Uber would be fine in that case?
Wait, what?
The big ERP vendors aren't under any threat, the small ones are.
No business is going to switch from a system that has armies of low-paid consultants to in house AI developed system that has effectively zero consultants who can come in and perform the deployment with tested integrations to their accounting system, their 200 suppliers, their customer systems and their 3rd party auditing systems.
But, small businesses who were not going with a 12m contract for 5 consultants, and who dont have any need for integrations to suppliers, customers and 3rd party systems can do their own systems.
It sounds like you are very far removed from ERP and business systems in general.
All magnetic coding is going to do is further entrench existing large systems because new systems, whether AI generated or not, will be too numerous for any one of them to gain traction.
It was developed by a single guy in the IT department and she liked it.
About 5 years ago the company was acquired, and they had to move to their COTS 'enterprise' system (Maconomy).
All staff from the old company had to do a week long (!) training course in how to use this and she hates it.
In future I think there will be more things like 'Time' (though presumably not MS Access based!)
That's my assertion - those things like 'Time' can be developed by an AI primarily because there is no requirement of an existence of a community from which to hire.
It's an example of a small ERP system - no consultants, no changes, no community, etc.
Large systems (Sage, SAP, Syspro, etc) are purchased based on the existing pool of contractors that can be hired.
Right now, if you had a competing SAP/Syspro system freshly developed, that had all the integrations that a customer needs, how on earth will they deploy it if they cannot hire people to deploy it?
It's certainly not "SAP 10 million dollar deployments". we see implementation rarely run into 6 figures for SMB distributors and manufacturing firms. That's less than most of their yearly budget for buying new fleet vehicles or equipment
The most heard gripe was the concurrent access to the database file but I think that was solved by backing the forms by accessing anything over odbc.
It looked terrible but also was highly functional.
It was for school, and I recently found the write up and was surprised how well the system worked.
Ever since I've marvelled at how easy it was to build something highly functional that could incorporate complex business logic, and wished there was a more modern equivalent.
I've not seen anything as easy to use as the Access visual query builder and drag-n-drop report builder thing.
Are you sure about that? Because thats exactly what Klarna is doing/has done.
https://threadreaderapp.com/thread/1957789124930286065.html?...
Agents are limited. With you so far.
> This week we get a demo of a vibe coded frontend that is more beautiful and easy to use than any ticket management system I have seen
Again, totally matches my expectations. Agents totally make pretty stuff that looks like working software.
I just haven't drunk enough management wine to connect the dots and figure out these facts support a jira replacement.
It also gets me wondering, if Atlassian leaned more heavily into AI (More vibe coding, more agents, more layoffs) would they have been able to keep the Klarna contract?
That link does not say that they are switching away from a system that requires armies of consultants to implement.
AFAICT, they are switching away from Jira (Atlassian/confluence products). Those are not ERP systems.
Once again, I must point out that the these sorts of assertions reveal that the person making the assertion has never been involved in an ERP rollout, neither a big one nor a small one.
And, again, I reiterate, the only threat is to small players in the market, who don't have a community to hire from. Because to become a big player, you need to gain traction as a small player, and if every small ERP system can be replaced with an AI generated system, non single one is ever going to gain traction (Why pay $10/user/month for a basic system when you can have AI generate that for a once of fee and some employee time?)
That said, plenty of banks still run on mainframes and use COBOL.
https://www.salesforceben.com/klarna-salesforce-workday-part...
am I really? it sounds so many people in big ERP service providers are oblivious of the tide rising that will wash them away, because you know what - most of these companies are super pricey, super slow, very messy and tend to fail large-scale projects that cost millions? I've seen this happen personally, and I have personally, as a sole player implemented ERPs with custom inhouse software.
Trust me brother, I know ERPs very well and seen hundreds of high-profile fakers that have zero knowledge of E/R, Business Architecture, and integrations, that still believe they can get away with nonsense.
Hope you're not one of them.
I've been part of enough ERP contracts to know that customers evaluate their options based on how easy it is to hire consultants on the open market.
I did do quite a lot of custom software prior to AI-slop, and that market is completely destroyed with vibe-coding agents. The ERP one looks like it is simply going to further entrench the big players, because new ERPs evolve from some custom application, and once that pipeline is gone, your choices are going to be "build it yourself with no ability to hire for it" or "Go with one of the existing behemoths".
My argument is that AI will remove the pipeline that leads to incumbents seeing more competition.
In the not not so distant future (5 years? 10?) that consultant market will be of people with very good process knowledge and very good prompting skills. Which might or might not be in a good part the same consultant market we have today.
My work uses these services and it’s interesting to see the divide between companies who have documented APIs (some also with “marketplaces”), and others who have many thousands of dollars partnerships license requirements to get API access.
Or companies who are very restrictive about what they will let you do with their API, presumably in an effort to control the ecosystem around their products—whether it’s because they fear being open would devalue their product, or they have a strong notion of their market position and won’t admit any integration which upsets their governance.
No. I’m in financial sector, accounting.
For personal use, those making their own software will still be a minority for a while, but at my job we are seeing potential to save $1M-$10M a year by rolling a custom tool vs paying for a commercial one. The saving here come from the tool doing a better job, not the license we pay.
Companies have always been able to hire software engineers as well. At least that's my impression in the UK. Is this different in other parts? Not enough engineers left after big tech has hired them all?
So I shouldn't download and use this right? I can't verify if it's potentially malicious or not
Now imagine my not-so-complext ERP or internal system - can be developed with little or no effort. Why would I give Benioff my dollars rather than spend it on in-house assets, that also increase the valuation of my company? I find very little reason to do so in 2026.
An ERP has an established workflow that follows GAAP principles.
Hundreds of thousands of customers have cut their teeth on that workflow and improvements are metered out.
The last thing you want is to have to do PCI compliance or 1099 reporting, tax calculations for every jurisdiction. IFRS, Inventory valuation methods, SOX controls, revenue recognition rules, etc.
Not to mention if you get audited saying "Oh yeah we vibe-reconcilled all those statements".
Anything that touches the ERP? sure.
If you re-design ERPs for total AI? Maybe actual ledgers (no - not tables in MSSQL), imdepotency, rollbacks, maybe. still a bad idea.
Don't roll your own crypto. Don't roll your own ERP. Roll everything else around it.
Perhaps you would agree that in 2026 it is fairly easy it is to actually agentically-dev a very decent ERP given one has the blueprints such as value stream diagrams, caps, BPMNs, domain models, seq. diagrams, state diagrams and... basically a complete SRS bundle. I doubt this person even needs to be super technical to deploy it.
Does it require that you use large (or small) SAAS? I guess not.
It requires one understands business architecture and know-how related to the application of such mental tools.
Traditionally it requires someone (a person) implementing these, translating them into code. Well this is precisely what LLM agentic systems do - translation. And they do it much cheaper with much shorter dev.cycles. And tailored also.
If what you predict is true then the sheer amount of software is going to explode. Good for people who actually understand how it all works!
So still, for ransom business much cheaper and better to buy software from SaaS vendor.
In this case it was better ONLY because the client is the wife of the developer.
And even now, if he sells this to other businesses - it will be MUCH cheaper to buy his subscription than homebrew the same version of it - as if it starts selling it he will be adding features and support which requires time (which is money).
No, this is not true. There are so many non-technical users of Microsoft Access that run their won businesses without hiring anyone. A friend of mine had a business with an yearly turnover shy of $3M (which is small, alright) and it was running wired spreadsheets and google forms. 20 people. He never ever bought any software, and existed for more than 10 years, until his wife (yes his waifu) decided to divorce and bring the company down.
Business Architecture is not so much about writing the software, sorry, we as IT professionals would love to think it is, but this is a super weak bias.
On the other hand - we did recently pitche some brand new RAG for TBs of complex schematics for a company in the Netherlands, and guess what - they didn't like the enterprise rates that the middleman offered, not because they did not like the demo (which they loved absolutely), and not because it was late or incomplete (it took less than a month). Had I approached this company directly, with normal rates, I would be deploying it in production already. It's very telling, and not good news for large SAAS vendors.
>> was running wired spreadsheets and google forms
So he was using someone's software.
Probably he grew his business to 3M for this exact reason - by focusing on business and not playing tech company by homebrewing software. Which only proves my point.
I still had to think hard about how to make this simple and easy for someone who does not have deep knowledge of this manufacturing domain.
*Lessons learned:*
1. Data structure is almost everything, then comes business logic
2. You must have deep domain knowledge of what you are building
3. Iterate fast on the views built on top of the data structure
All mobile fixed had been resolved and deployed ;)
"Show me your flowchart and conceal your tables, and I shall continue to be mystified. Show me your tables, and I won't usually need your flowchart; it'll be obvious." -- Fred Brooks
If we're assuming that:
I find it less time consuming to just write the code, understand it 99% (since I wrote it), and debug the rest, than it is to try to describe it to the AI, understand a fraction of what it spits out, and spend more time understanding it and fixing it.
If you can just write clean code just do that. Also, you will improve your skills even more the more you do that (shocker). So the next time you have to do that it will be even easier. This is called learning a skill.
Sorry for the rough tone, reading that back haha. But still posting because I'm just passionate about it, it's nothing personal though.
Elixir + Ash is an interesting choice for this domain. LiveView particularly shines for internal tools like this where you want the interactivity without managing a separate frontend build. Curious how the AI code generation worked with Ash specifically - the declarative nature seems like it could either help a lot (clear patterns) or confuse models that expect more explicit code.
The BOM with cost rollups is the feature that would have saved me hours in a previous project. Most small batch producers I know either overprice everything out of caution or underestimate costs because tracking ingredient pricing through recipes is tedious in spreadsheets.
But I do think the days of just having to learn, tolerate, and accept any commercial software stack for your business because it's too complex to build yourself are over. What vendors remain will have to absolutely meet users with their unique requirements and budget.
I don't think it's useful to anyone - not white label, not open source - but still funny :)
(Le filtre pain/biscuits/gluten est pété sous safari)
My only nit, as a legacy internet goober is, use example.com for these throw away addresses; it's reserved for that purpose.
Some small things: When trying to edit a product (almond cookies) on the phone, I cannot scroll the pop-up so cannot go to all fields or the save button. When calculating the total calories it prints kg as the unit instead of cal. On the overview page of materials for a product it only shows “grams” per ingredient without the actual number.
So often, these tools lack usability because they’re built generically any use case. Here, it was designed for your wife’s bakery.
The process to build a site/app like this will only get easier and more defect-free over the coming months.
I'm an Elixir newbie and wondering if I should start with learning Ash or stick with Liveview until I know more. Any thoughts on what Ash solved for you over Phoenix Liveview?
It's simple, well documented, and uses appealing technologies.
I'm sure your wife's business will take off.
I’m an elixir noob
Seems like the live demo website is about to die though.