Moving Beyond Prompts to Loops and Graphs in AI Systems

X postMahaxResource

Citing Andrew Ng, Mahax highlights the shift from single-shot prompting to agentic loops and graphs. An accompanying guide outlines the architecture of AI agents, self-evaluating feedback loops, and graph-based execution pipelines using Claude.

Key points

  1. The post cites Andrew Ng predicting that prompting will be replaced by iterative loops and graphs.
  2. An AI agent differs from standard chat through tool access, memory persistence, and goal-oriented execution loops.
  3. Agentic workflows span four levels: standard chat, tool-augmented queries, multi-step execution, and autonomous workflows.
  4. Loops stop only when a verifiable condition or test is met, rather than after a single output generation.
  5. Reliable loops require explicit failure tests, context retention, and strict retry limits to prevent infinite execution.
  6. Graphs do not discard loops; they act as networks running multiple loop nodes with explicit dependencies in parallel.

Why it matters

The shift marks a transition in AI engineering from conversational prompt tuning to structured programmatic architectures that ensure quality control and autonomous task completion.

Analysis

Mahax explains that building effective agentic systems is an architectural challenge rather than a model upgrade. Success relies on structuring tools, memory, and validation checks around the LLM. While multi-pass loops consume more tokens, iterative self-testing significantly improves output reliability over single-prompt workflows.

Original post

Mahax

@Mahaximus_ · 6 August 2026

Google Brain founder, Andrew Ng: "Prompting will die in 6 months. Loops and Graphs are what's replacing it." In 2 hours, he shows exactly what the best engineers already build instead, and how to start building it yourself. The missing piece most people skip: how to connect t.co/G2HZP4l89U t.co/2nl0rWLfHC

4.6k likes83 replies11.8k bookmarks1m views

View on X
More posts in the thread (2)
  1. @ForthUtility loops don't go away, they just become the nodes inside a graph, a graph is still running loops, it's just running several of them at the same time with explicit dependencies between them one builds on the other rather than replacing it
  2. @joakim_a there's something to that, more iterations does mean more tokens, but the practical result is real a loop that checks its own output and retries actually produces better work than a single pass
Selected replies (5)
  • @CsTominaga @Mahaximus_ There is something wonderfully modern about watching people announce, with grave faces and revolutionary terminology, ideas I was already working on, filing, publishing, and building in 2013 and 2014. Agent-based computing. Swarms of interacting agents. Tokenised access.
  • @joakim_a @Mahaximus_ Isn't this whole loops thing just a marketing gimmick from frontier labs to make users rake up tokens?
  • @BigBlocksBoss @CsTominaga @Mahaximus_ I can testify to that years ago Dr.Wright has been talking about machine's paying each other .What can I say technocrat's are afraid of genuine Bitcoin for that reason.
  • @ForthUtility @Mahaximus_ Is it true graph “engineering” is already overtaking loops?
  • @Ieanianz @spacetimeshare @Mahaximus_ the human does not leave, the job moves. you stop writing the instruction and start defining what counts as done, and most loops fail because nobody ever wrote that part down

Mentioned links

Summary generated by AI (Gemini 3.8 Flash, 27 September 2026). It may contain errors – the original source is authoritative.

Related entries

Similar in content, found by the AI search.

  • X post:nicco

    Beyond Prompting: Agentic Workflows as Execution Graphs

    Citing a Stanford lecture by Andrew Ng, author nicco highlights why zero-shot prompt engineering is reaching its limits and being replaced by autonomous execution graphs, supported by a guide on moving away from linear agent chains.

    275bookmarks25.1kviews

    AI & LLMs· Opinion

  • X post:Mahax

    Graph Engineering: Why Agent Workflows Depend on Graphs

    Mahax cites Andrej Karpathy's claim that prompting is disappearing in favour of graphs, and provides a practical guide on graph engineering with Claude to optimise AI agent workflows.

    2.4kbookmarks240.8kviews

    AI & LLMs· Opinion

  • X post:nicco

    Graph Engineering Over Prompting: Why Agent Workflows Require Graph Architectures

    A post by nicco highlights remarks by Andrej Karpathy asserting that prompting is an ephemeral interface and that graph architectures represent the true architectural layer for agent workflows. It pairs this with a practical guide by seeco on replacing sequential agent chains with explicit graphs.

    529bookmarks58.5kviews

    AI & LLMs· Resource

  • X post:Kirill

    From Loop Engineering to Graph Engineering: Deterministic AI Architectures

    Kirill describes an architectural shift in agent development from fragile single-agent loops to Graph Engineering. He argues that system reliability depends not on larger models or more agents, but on deterministic structures using nodes, edges, state verifiers, and knowledge graphs.

    27.3kbookmarks2.7mviews

    AI & LLMs· Opinion

Back to the knowledge base

Let's talk about your project

Get in touch

Locations

  • Mattersburg
    Johann Nepomuk Bergerstraße 7/2/14
    7210 Mattersburg, Austria
  • Vienna
    Ungargasse 64-66/3/404
    1030 Wien, Austria

Parts of this content were created with the assistance of AI.