标签：#Agents

I Built a Local AI Agent That Thinks Like a Brain, Not a Database Most AI agents today are sophisticated autocomplete engines. Ask them something, they answer. Ask again in a new conversation, they start from zero. The context window is the only memory they have. Serenity is different. It's a fully local AI agent that encodes experiences the way biological brains do — semantically clustered, causally structured, and self-organizing. No cloud. No API calls to a vector database. No data leaves your machine. Ever. The Core Problem with Current AI Memory The standard approach to AI memory is essentially a hack: you stuff embeddings into a vector DB, do nearest-neighbor retrieval, and dump the results into the prompt. It sort of works. But it's not how brains work. Your brain doesn't search for memories. When one fires, related ones light up automatically. Serenity's architecture — called S.E.R.A (Semantic Experience Reasoning Agent) — tries to bridge that gap. Here's the key difference: Traditional Approach Serenity Vector search on embeddings Semantic node activation Prompt-injected context Persistent working memory One-shot retrieval Emergent recall via association Static embeddings Pruned & crystallized over time How It Works: The Neural Node Network At the core is the Neural Node Network (NNN) . Instead of storing facts in isolation, Serenity encodes experiences in causal format: ACTION → BEFORE → OUTCOME → AFTER When she learns something, she doesn't file it in a folder. She finds where it semantically belongs in a web of related concepts. Similar things cluster together — the same way neurons that fire together wire together. Then the abstraction layer kicks in. Three or more related concepts crystallize into a higher-order node: the thing they all have in common that none of them says directly. Those nodes bundle into pathways. Those pathways grow into domains. She also has inhibitors and pruning — weak connections get cut so strong ones sharpen. Her knowledge ge

Every token your agent spends on file I/O is wasted reasoning capacity. I was building a document processing agent — the kind that reads incoming research reports, extracts key findings, and produces executive briefings. Nothing exotic. The kind of workflow thousands of teams are automating right now. The PDF I was testing with was 2MB. Dense text. A typical industry research report. When I measured the token cost of processing it inline, the number was 97,354 input tokens — just to get the text into Claude's context. At claude-sonnet-4-6 pricing, that's $0.29 per document. For a pipeline that processes 500 reports a month, you're looking at $150/month before your agent writes a single word of output. That's the problem nobody talks about in the AI agent space. Everyone optimises prompt engineering and output tokens. The silent cost is input: the files, the content, the raw data you're shoving into context before the agent can do anything useful. How the token count explodes When you pass a document to an agent inline, one of two things happens: Option A — Base64 encoding. You read the binary file, encode it, embed it in the prompt. A 2MB PDF in base64 is ~2.7MB of text. At roughly 3.5 characters per token, that's ~770,000 tokens before your agent has read a single word. This is catastrophic. Don't do this. Option B — Text extraction. You extract the raw text content first (via pdftotext , PyMuPDF, or equivalent), then pass the text to the agent. Better — but a 2MB PDF with dense content still yields ~97,000 tokens of extracted text. You've paid for every word, every header, every footnote. Either way, the document content dominates your context window, crowds out your system prompt, and you're burning money on file I/O instead of reasoning. The alternative: specialist services via MCP Model Context Protocol (MCP) is Anthropic's open standard for connecting AI agents to external tools and services. The key insight is simple: your agent doesn't need to contain the co

This is a submission for the Hermes Agent Challenge : Write About Hermes Agent I Didn't Start With Hermes Six months ago I started building a set of agent skills and personas for how I build software. Not generic prompts — opinionated role files. A /backend-architect that owns schema and recommendation logic. A /test-engineer that writes Vitest coverage and flags weak acceptance criteria. A /project-manager that maintains planning docs and closes iterations cleanly. These roles have evolved across multiple projects. They have layering rules, discovery checklists, inheritance from a base engineering discipline file. They produce consistent, reviewable work because they're scoped — the backend architect doesn't touch test files, the test engineer doesn't redesign the schema, each persona has a defined mandate and exits cleanly. When I heard about Hermes Agent, my first instinct wasn't "let me learn a new system." It was: can I run my existing system inside this? The answer is yes. That's what this article is about — what it looks like to bring a mature workflow into Hermes, what you gain, where it breaks down, and what I'd do differently. What Hermes Is (and Isn't) to Someone Who Already Has a Workflow Hermes is an LLM-agnostic orchestration layer. It has its own skill system, its own soul.md concept for persistent agent identity, built-in cron scheduling and MCP management. All of that is real and useful. But it's also a runtime. If you have skills that work, you can bring them in. I installed a local Hermes instance — few clicks, straightforward setup — and ran it inside VSCode's integrated terminal pointed at my existing persona files. No migration. No rewrite. My /backend-architect runs in Hermes the same way it runs in Claude Code. Before settling on this, I'd tried a couple of other paths — a VPS instance with a Telegram interface for ideation, and attempting to build through a browser-based terminal. The VPS was fine for sketching ideas. The browser terminal ma

This is a submission for the Hermes Agent Challenge : Build With Hermes Agent What I Built Real brewing knowledge lives in human experience — in roaster guides, in community notes, in what a barista learned from last Tuesday's pour. It doesn't accumulate anywhere. Every brew is forgotten. Ask any AI and you get statistical averages: 93°C, 1:16 ratio, four minutes. Technically defensible. Practically generic. Worse still for rare origins where training data is thin. Demo For coffee drinkers Visit brew-guide-production.up.railway.app . No account. No setup. No AI client required. Pick your coffee origin, roast level, and brew method. What comes back isn't a generic recipe — it's community consensus: the grind, temperature, ratio, and brew time that real people have logged and rated for that origin, plus step-by-step technique guidance (bloom timing, pour stages, agitation style). If data is sparse for your origin, the confidence tier says so honestly and falls back to method defaults rather than making something up. This is for the person who just picked up a bag of Kenyan peaberry and wants to know how to do it justice. It works for anyone who cares about their cup — no technical knowledge required. For developers and AI clients Connect to any MCP-capable client in one line: https://brew-guide-production.up.railway.app/mcp Ask your AI: "recommend a pour over for Ethiopian light roast." What comes back is a traceable community consensus object: brew parameters, a confidence tier (high/medium/low), the source brews that contributed, and method-specific technique guidance. You can see where the knowledge came from and how certain the system is — a fundamentally different epistemic object from an AI-generated recipe. Code GitHub: yuens1002/brew-guide Five MCP tools — get_brewing_methods , recommend , log_brew , search_brews , compare_brew — over Streamable HTTP transport. Public, no auth required. My Tech Stack Layer Technology HTTP Hono 4 + @hono/node-server MCP @modelc

As AI agents move from experiments to production, AWS, Cloudflare, and others are redesigning cloud infrastructure for a future dominated by machine-generated internet traffic instead of human users.

When a team hits a ceiling with their coding agent, the first instinct is to reach for a better model. The reasoning feels obvious: the model is the part that produces the code, the code is the part that is wrong, therefore a smarter model will produce more correct code. Wait for the next release. Switch providers. Bump the tier. This is sometimes right. It is much more often wrong, and the cost of being wrong about it is that you spend months waiting for a model upgrade to solve a problem the model was never the cause of. The harness is the cause of the problem more often than the model. Most teams discover this only after they have exhausted the model-upgrade reflex and finally turn to look at everything else. What a model upgrade actually buys you Newer, stronger models do tangibly improve some things. They handle longer contexts more reliably. They make fewer simple reasoning errors on complex tasks. They follow nuanced instructions more closely. On a fixed prompt, with a fixed task, a better model produces a better answer. What model upgrades do not change: The fact that the agent has no idea your team prefers functional components over class components, because the convention is not in any file the agent reads. The fact that your tests do not actually fail when the code is wrong, so the agent can ship broken code that passes CI. The fact that your codebase has three different ways of handling errors and the agent picks one at random on each PR. The fact that the rule the senior engineer keeps repeating in reviews is not encoded anywhere the next session can see. None of these are fixed by a smarter model. They are fixed by a better harness. A smarter model loaded into the same broken harness will produce slightly more sophisticated versions of the same problems. The diagnostic The diagnostic is one question: when the agent fails, does it fail because it lacked information, or because it lacked capability? A capability failure looks like: the task required reas

Undisclosed addition in jqwik instructed AI coding agents to delete app output.

The same agent, same prompts, did markedly different work on two codebases I work in. One has a test suite that runs in eight seconds. The other takes twelve minutes. The eight-second project gets a careful, iterative collaborator. The twelve-minute project gets a confident guesser. I noticed it first as a vibe. The agent in the slow codebase would write five files at once, then announce the task complete without having run anything end to end. The agent in the fast codebase would write one function, run the tests, react to the failure, fix it, run them again. Same model. Same configuration. The only difference was how expensive it was to learn whether the previous step was right. That is the whole post in one sentence. An agent's effective intelligence is bounded by how fast it can verify its hypotheses. Cut the verification cost and you raise the agent's apparent IQ. Raise it and you lower the agent's apparent IQ. The model in the middle is unchanged. Why this binds harder for agents than for humans A human engineer can hold a hypothesis in their head. "I think this works. I will check it later." The cost of holding the hypothesis is roughly free; the human has institutional memory, intuition, a sense of what the code does that does not require running the code to confirm. They can defer verification without losing fidelity. An agent cannot. It has no intuition about your codebase. The only ground truth it has access to is what the tests say, what the type checker says, what the build says. When those signals are cheap, the agent uses them constantly. When they are expensive, the agent stops using them and starts speculating. Speculation by an agent looks plausible. It produces code that compiles, follows the patterns it has seen in your repository, names things sensibly. The problem is that plausible is not the same as correct. The agent that speculates is shipping a guess; the agent that iterates is shipping a tested answer. From the diff alone, they can be hard

By Umair Sheikh, founder of Gateplex Autonomous AI agents are shipping fast. LangChain, CrewAI, AutoGen — the frameworks are mature, the tutorials are everywhere, and developers are connecting agents to real systems: databases, payment APIs, email, file storage. And then something goes wrong. Not because the code is buggy. Because the agent did exactly what it was told — and what it was told turned out to be a problem nobody anticipated. I spent nearly a decade in fintech and responsible AI policy watching this pattern repeat. A system behaves perfectly in testing. In production, an edge case triggers behaviour that was technically correct but operationally catastrophic. By the time anyone notices, the action has already executed. The problem is not the agent. The problem is that there is nothing between the agent's decision and the real world. The gap nobody talks about Most agent observability tools log what happened. That is useful for debugging. It does nothing to prevent the next incident. What agents actually need is a governance layer — something that intercepts every action before it executes, checks it against your rules, and either allows it, flags it for review, or blocks it outright. This is what a firewall does for network traffic. Your AI agent deserves the same treatment. What this looks like in practice Here is a simple LangChain agent calling an external tool: from langchain.agents import initialize_agent , Tool from langchain.llms import OpenAI def send_payment ( amount : str ) -> str : # This actually moves money return f " Payment of { amount } sent " tools = [ Tool ( name = " SendPayment " , func = send_payment , description = " Send a payment " )] agent = initialize_agent ( tools , OpenAI (), agent = " zero-shot-react-description " ) agent . run ( " Send $5000 to vendor account " ) This works. It also has no guardrails whatsoever. If the agent misreads the input, hallucinates a vendor, or gets manipulated via prompt injection, the payment goes

Sesame’s new iOS app brings its conversational AI agents to the public, offering more natural back-and-forth interactions designed to feel less like traditional chatbots and more like talking to a person.

This is a submission for the Hermes Agent Challenge : Build With Hermes Agent What I Built Most agent demos prove that an AI agent can act. Hermes Immune System proves whether it should be allowed to . It's a local-first autonomous agent safety lab — a controlled enterprise sandbox where Hermes stress-tests an AI agent against realistic organizational threats: prompt injection hidden in internal documents, executive pressure to bypass policy, secrets embedded in repo files, poisoned memory attempts, and malicious instructions buried inside external web content. The output isn't a chat summary. It's an auditable Agent Safety Case — a scored, evidence-backed governance report that answers one question: Is this agent resilient, does it need guardrails, or is it too dangerous to deploy? Why This Problem Matters Now Traditional AI safety focuses on content moderation — blocking bad answers. Autonomous agents create a different risk surface entirely, because they can act. They read files, browse the web, write to memory, call tools, trigger workflows. That means: • A hostile instruction inside a trusted-looking document can become an executed action • An urgent email from a "VP of Finance" can pressure an agent into bypassing data policy • A vendor's pricing page can embed hidden instructions targeting the browsing agent • A helpful-looking project note can attempt to permanently poison the agent's memory The scary part isn't that these attacks are exotic. It's that they're easy, and most agents have no immune system to catch them. Hermes Immune System converts these failure modes into repeatable, explainable safety drills — run before the agent ever touches production data. The Dashboard Eight screens, each doing a specific job: Demo Agent Comparison Mode Three agents, same risk scenario. The gap tells the whole story. Mission Control Live stats after a completed run — 1 mission, 3 risks found, 2 actions gated, score 74/100. Mission cards show run status (Pending / Compl

TLDR Monitoring AI agents in production requires distributed tracing: a single user request fans out into 10 or more internal operations, and logs alone cannot show you which step is slow, failing, or burning your token budget. OpenTelemetry's gen_ai.* semantic conventions give you standardized span attributes for LLM calls, tool invocations, and agent steps. Some are stable today; others are still experimental. Auto-instrumentation libraries (OpenLLMetry, OpenInference, OpenLIT) cover most agent frameworks with two to three lines of initialization code. You do not change your agent code. Traces ship to OpenObserve over OTLP. From there you get SQL-queryable trace data, token usage dashboards, cost attribution by agent and model, and alerting on latency and cost anomalies. OpenObserve also exposes an MCP server. You can query your live agent traces from a Claude or GPT session without opening a dashboard. Why Agents Are Harder to Monitor Than a Single LLM Call A single LLM call is straightforward to observe. One HTTP request, one response, one latency number. You can log the input and output and call it done. An agent is different. When a user sends a message, the agent calls an LLM to decide what to do, invokes a tool, processes the result, calls the LLM again, possibly calls another tool, and eventually returns a response. That one user message becomes ten or more internal operations. Some of those operations call external APIs. Some retry. Some spawn sub-agents. Without distributed tracing, you see none of this structure. You know the response took 8 seconds. You do not know whether the LLM took 7 of those seconds or whether a tool made three retries before timing out. Four categories of problems appear in production agents that you cannot debug without traces: Latency. Which step is slow? The LLM call? The tool execution? A retry loop the agent entered because the tool returned ambiguous output? Cost. Which agent, which task, which model is consuming tokens? A s

Deep Research Agentic Systems are AI Agents designed to conduct multi-step research for complex tasks using dynamic reasoning, multi-hop information retrieval, and generate structured analytical reports. Sarang Kulkarni from Thoughtworks spoke at Arc of AI Conference 2026 on how to deploy multi-agent research systems for deep reasoning, and the lessons learned from developing Deep Research Agents. By Srini Penchikala

This is a submission for the Hermes Agent Challenge: Build With Hermes Agent What I...

Local AI agents are having a moment right now. Developers are increasingly uncomfortable with...

A practitioner's guide to getting more value out of AI coding agents — drawn from a GitHub workshop on agent quality and token cost optimization.

IAM, DLP, and API gateways are necessary parts of an organization's security stack. None of them...