[HN Gopher] From Records to Agents: The Overlooked Revolution in...
       ___________________________________________________________________
        
       From Records to Agents: The Overlooked Revolution in Enterprise
       Software
        
       Author : sebg
       Score  : 16 points
       Date   : 2025-02-26 18:52 UTC (4 hours ago)
        
 (HTM) web link (sperand.io)
 (TXT) w3m dump (sperand.io)
        
       | jcgrillo wrote:
       | pure word salad.. i want these drugs where can i get some?
        
         | DrillShopper wrote:
         | Much like with drugs the enterprise software folks from time to
         | time have to remember that getting high on their own supply is
         | a bad idea.
        
       | beardedwizard wrote:
       | Why do so many see LLM as the inflection that makes this all
       | possible? So much of the proposed automation can be built in
       | totally deterministic ways for cheaper. Not everything benefits
       | from being forced to look like a non deterministic language
       | problem.
       | 
       | You might argue the biggest barrier to enterprise software
       | innovation is the enterprise mindset vs a more engineering first
       | mindset.
       | 
       | Rippling comes to mind as an example of bringing engineering to
       | enterprise.
        
         | soulofmischief wrote:
         | I can go back to the NES or Atari and probably develop what
         | would have been groundbreaking games and genres well before
         | their time. But it took several small advancements and
         | inflection points for those conceptual mechanics and hardware
         | tricks to now be obvious in hindsight, and that's why no one
         | did them earlier than they were done.
        
           | beardedwizard wrote:
           | I agree there are a bunch of small inflections that make me
           | confident such a solution is possible today, but llms just
           | aren't one of them. Enterprise software is a prisoner of its
           | own old mindset and low resource allocation.
        
         | SrslyJosh wrote:
         | Yeah, this is where I stopped reading:
         | 
         | > LLMs now offer the intelligence layer that makes it all
         | viable
         | 
         | LLMs are the opposite of what you want in enterprise software:
         | unpredictable black boxes that go awry seemingly at random. I
         | also can't see how they're at all necessary if you've got a
         | state machine describing a process: the state machine tells you
         | what needs to happen next.
        
         | jcgrillo wrote:
         | But don't forget the Actor Model and Durable Execution! That'll
         | surely keep the LLM on track.
        
       | soco wrote:
       | There's a lot of science-fiction in this "article". Almost none
       | of the premises they listed for building the enterprise LLM
       | future are actually there. No data silos, then data sharing, zero
       | copy, encapsulated processes... yes I could _imagine_ all this.
       | But that 's all we have for now, imagination.
        
       | ab_testing wrote:
       | Why does it feel like the author has never worked on enterprise
       | software. The enterprise ERP systems are clumsy because they are
       | built on a system of checks and balances . He talks about
       | invoices finding their own approves when that is exactly what
       | business rules are coded for - to handle complexity and notify in
       | case of no approves found. Enterprise software is unsexy because
       | it drives much of GAAP accounting and businesses do not like
       | surprises in that domain.
        
         | gwbas1c wrote:
         | > Imagine: the invoice isn't just a record in your ERP anymore.
         | It's a machine that wants to get approved. It wants to get
         | paid.
         | 
         | Stopped reading at that point. Realized this is just fluff from
         | someone letting their imagination run away.
        
       ___________________________________________________________________
       (page generated 2025-02-26 23:01 UTC)