标签：#ai

You have 50 documents and you're running them through an LLM in a loop. The first one finishes at the 2-second mark. The fiftieth finishes at the 100-second mark — not because it's harder, but because it waited in line behind the other 49. Pool runs all 50 at the same time. The Problem With Loops Every developer who works with LLMs writes this code eventually: import os from yait_aichain.models import Model from yait_aichain.skills import Skill skill = Skill ( model = Model ( " claude-sonnet-4-6 " , api_key = os . getenv ( " ANTHROPIC_API_KEY " )), input = { " messages " : [{ " role " : " user " , " parts " : [ " Summarise in two sentences: \n\n {text} " ]}]}, ) documents = [{ " text " : f " Document { i } content... " } for i in range ( 50 )] results = [] for doc in documents : result = skill . run ( doc ) results . append ( result ) It works. It's readable. And it's painfully slow. Each LLM call takes roughly 2 seconds. Multiply that by 50 documents and you're staring at your terminal for almost two minutes. The calls are completely independent — document 37 doesn't need the result of document 12. Yet document 37 sits idle, waiting its turn. That's a scheduling problem, not a computation problem. I ran into this directly while building a task that pulled N files or links and produced a consolidated report. The sequential version was logically fine but just hemorrhaged time. I needed to fire everything at once without rewriting the Skill logic — no new prompt templates, no restructured code, just a different execution model. That's what Pool is. Pool: Parallel Map for LLM Calls Pool takes one Skill (or Chain) and a list of inputs , then launches all of them concurrently. Think of it as Array.map() where every element runs in parallel against an LLM. import os from yait_aichain.models import Model from yait_aichain.skills import Skill from yait_aichain.pool import Pool , DONE , FAILED skill = Skill ( model = Model ( " claude-sonnet-4-6 " , api_key = os . getenv ( "

Our self-proclaimed Scroogian (cheapskate) book-loving writer has put together a guide on finding free books for your Kindle.

You see words. A language model sees tokens — chunks of text, usually a few characters each. Everything starts here. Day 2 of my AIFromZero series. Text gets shattered into tokens "unbelievable" → ["un", "bel", "iev", "able"] (4 tokens, not 1 word, not 12 letters) Before any "thinking", your text is chopped into tokens and each becomes a number the model processes. Why not words or letters? Letters : too fine — the model would relearn spelling everywhere. Whole words : too many — millions, plus every typo and name. Subword tokens : the sweet spot. Common words = 1 token; rare words split into reusable pieces. A fixed ~100k-token vocabulary covers any text. The ~4-chars rule (and why it costs you) In English, ~4 characters ≈ 1 token , or ~0.75 tokens per word. This is how everything is priced and limited: API bills are per token (prompt + reply). A "context window" (how much it can read at once) is measured in tokens — 1,000 tokens ≈ 750 words. Verbose prompts and long chat history burn tokens. Concise prompting is a real cost lever. The strawberry problem The model never sees s-t-r-a-w-b-e-r-r-y. It sees a token like straw + berry . The individual letters are buried inside tokens, so counting characters is genuinely hard for it. It's not dumb — it just doesn't read letters. Tokens are step 1 of everything Tokenize → turn each token into a vector (embeddings, next) → run through the transformer → predict the next token. Every LLM starts exactly here. 🔤 Type anything and watch it tokenize live: https://dev48v.infy.uk/ai/days/day2-tokens.html Day 2 of AIFromZero.

"Chat with your PDF / your notes / your docs" is everywhere. Today we build it from scratch and you'll see it's just three moves : retrieve, then generate — with one prompt trick that stops the hallucinations. This is Day 46 of TechFromZero. Yesterday (Day 45) we built the retrieval half with pgvector. Today we add the answer half and host it on Supabase. RAG in one line Find the relevant chunks of your documents, paste them into the prompt, and tell the model to answer using only those. That's Retrieval-Augmented Generation. The "augmented" part is just stuffing real context into the prompt so the model isn't guessing from memory. 1. Storage: Supabase is Postgres, so pgvector is one click Supabase is hosted Postgres with an auto-generated API. Because it's just Postgres , vector search needs no separate database: create extension if not exists vector ; create table documents ( id bigserial primary key , content text , embedding vector ( 384 ) ); -- one RPC the app calls to get the closest chunks create function match_documents ( query_embedding vector ( 384 ), match_count int ) returns table ( id bigint , content text , similarity float ) language sql stable as $$ select id , content , 1 - ( embedding <=> query_embedding ) as similarity from documents order by embedding <=> query_embedding limit match_count ; $$ ; 2. Ingest: chunk → embed → store Split your docs into paragraph-sized chunks, embed each with a free local model (all-MiniLM-L6-v2 via Transformers.js — no key, nothing leaves your machine), and insert the row + vector: const embedding = await embed ( chunk ); // 384 numbers await supabase . from ( " documents " ). insert ({ content : chunk , embedding }); Chunk size matters: too big buries the answer in noise, too small loses meaning. A few hundred characters is a good start. 3. Retrieve + Generate (the payoff) Embed the question with the same model, ask Supabase for the closest chunks, then hand them to the LLM: const query_embedding = await embed ( que

The earlier posts in this series were about what the gateway lets you call (cache-aware spawning across five providers, the Codex review gate, the CLI-versus-API argument) and the one before this was about the parts that do not show up as a tool, the upstream-tracking and the website that became the project's front door. This one is about a different kind of front door: the gateway can now listen over HTTP, behind real authentication, and serve more than one caller without those callers being able to read each other's work. That sounds like a small toggle. It is not. Moving an MCP server off a local pipe and onto a network port changes the trust boundary completely, and 2.9.0 is the release where I sat down and remediated all seventeen findings from a multi-LLM red-team of exactly that surface before telling anyone the remote path was ready. Short version: llm-cli-gateway is one Model Context Protocol server that wraps five vendor CLIs (Claude Code 2.1.177, Codex 0.139.0, Google's Antigravity agy 1.0.8, Grok 0.2.51, and Mistral Vibe 2.14.1) behind a single, uniform tool surface, so one orchestrating agent can fan a task out to several models, collect independent opinions, run a red-team or a consensus check, and keep durable session and job state across all of it. Until recently that only made sense on localhost over stdio. As of 2.9.0 the same server runs over HTTP with a static bearer token or a built-in OAuth 2.0 authorisation server (PKCE on by default, an opt-in human-consent gate, and a trusted-principal-header seam for when you front it with your own identity-aware proxy), every session and job and stored request is stamped with an owner principal and access is enforced per principal, remote provider calls are refused unless a workspace is registered, and the whole thing fails closed rather than open when the configuration is dangerous. Long version is below, same shape as always: what it enables, what the remote options actually are, a stack of worked scenar

It's widely accepted that as AI writes more and more of our code, opportunities decrease for newcomers to gain the skills needed to become professional programmers. At the same time, experienced people bemoan the way AI is causing them to gradually lose the skills they took so long and worked so hard to gain. Programming, like mathematics and perhaps music, sits in the middle between science and art. As science, it's all about capturing and organising knowledge, and as art, it's the application of creativity to problem-solving. Enjoyment - deep satisfaction or an emotional rush - can be had from either or both of these. But the perception is that AI is depriving us of opportunity, both to learn and to be creative. Code written by a human reveals a lot about the coder. Their depth of knowledge of the coding language, their creative use of variable and function names, and even the visual layout of a script. You can often tell at a glance whether the writer truly cared about the code they wrote, or whether it was just a means to an end, to ensure their next pay packet arrives on time. As AI improves, it makes fewer mistakes, but it also produces code that is increasingly hard for any human to read and truly understand. Since the job of a human in the coding loop is increasingly not to write but to validate code, this is becoming a problem. So here's my recent story, a series of happy accidents. I've been using agentic AI for about a year. First with Copilot in VS Code, then more recently with Claude Code at the command line. At the start, I found Copilot to make a lot of mistakes and to produce rather repetitive, poorly-structured code, but looking back, a lot of that was probably down to my own ignorance of how to control the process. The biggest problem I had was in validating JS or Python code. There was just so much of it, often using features of both languages I never fully got to grips with. I've never been a first-class programmer and most of my own code is pure

Running Chinese LLMs at Scale: A Cloud Architect's Notes I want to talk about something I've been wrestling with on real production workloads: the four Chinese model families — DeepSeek, Qwen, Kimi, and GLM — and how they actually behave when you wire them into a multi-region pipeline serving thousands of requests per second. I've spent the last several months routing traffic across all four through Global API's unified endpoint, and the picture that emerged was messier and more interesting than any benchmark table would have you believe. Most comparisons you'll find online are written by people who ran a handful of prompts in a notebook. I'm not that person. I care about p99 latency, failover behavior, what happens when a region goes down at 3 AM, and whether the model that wins on a leaderboard also wins when 500 concurrent users hit it simultaneously. Let me walk you through what I actually found. Why These Four, And Why Through One Endpoint Before I dive in, a quick word on routing. I've been burned before by model lock-in and vendor-specific quirks, so when I started this evaluation I refused to scatter my SDK calls across four different providers. Global API gives me a single OpenAI-compatible base URL ( https://global-apis.com/v1 ), one auth pattern, and the freedom to A/B test models without rewriting client code. If you architect anything at scale, you already know this is non-negotiable. The four families above are the ones I kept coming back to because each one claimed a different crown — and I needed to know which crown was real. The High-Level Matrix Here's the snapshot I keep pinned to my team's dashboard. It's not pretty, but it's honest: DeepSeek — $0.25 to $2.50/M output. The V4 Flash at $0.25/M is the workhorse. Qwen — $0.01 to $3.20/M output. Widest spread of any family. Alibaba's offering. Kimi — $3.00 to $3.50/M output. Premium-only, and they charge for it. GLM — $0.01 to $1.92/M output. From Zhipu. Big swing in capability across tiers. All four

AI Agents Are the Best Thing to Happen to Network Administration Since SDN A single API key, an AI agent, and a router behind a double-NAT in Southeast Asia. What happened next changed how I think about network management. I manage UniFi routers spread throughout the ASEAN region — some for friends, some for relatives, one for a charity. They're in different cities, different ISPs, different levels of network hostility. Most sit behind carrier-grade NAT. A few are in places where the government firewall blocks VPN protocols at the transport layer. UniFi's own management interface has always been good. The web dashboard, accessible through Ubiquiti's cloud, gives me visibility into every site: device health, client lists, traffic stats, WiFi experience scores. It's one of the reasons I chose UniFi in the first place — the centralized GUI just works. But the GUI is still a GUI. It's clicks and menus and dropdowns. It's fast for one site, manageable for three, and tedious at ten. For anything beyond what Ubiquiti built into the interface, you'd need to write your own tooling. I never bothered, because I'm not a developer, and the built-in dashboard was good enough. Then AI agents arrived, and suddenly the calculation changed. The Discovery I knew UniFi had an API. I'd heard about it in passing — some REST endpoints for the controller, vaguely documented, probably read-only. I never looked into it seriously because what was I going to do with it? Write a Python script to poll client counts? Build a custom dashboard? Without a team of developers, an API is just a locked door. But when I started working with an AI agent, I gave it my UniFi cloud API key on a whim. I figured it could pull basic stats — the stuff from the Site Manager API at api.ui.com/v1 . Read-only. Dashboard-level. Useful as context for answering questions. Then the agent discovered something I'd completely missed: the Cloud Connector API . I owe this discovery in large part to the Art of WiFi PHP client

Microsoft Agent Framework is built for production multi-agent systems, which is exactly why its LLM bill can grow faster than expected. If you are running workflows with retries, handoffs, tools, and checkpoints, the easiest savings do not come from prompting harder — they come from adding a gateway layer under the framework. I built Lynkr, so obvious founder disclosure: this article uses Lynkr as the gateway example. I’ll keep it practical and focus on where the cost actually shows up in Microsoft Agent Framework workloads. Why this is a real Microsoft Agent Framework problem The current Microsoft Agent Framework README positions it as a production-grade framework for Python and .NET, with: multi-agent workflows sequential, concurrent, handoff, and group collaboration patterns middleware observability provider flexibility checkpointing and human-in-the-loop flows That is exactly the kind of stack where token usage grows quietly. A single prompt-response app is easy to reason about. A production workflow is not. Once you add routing, retries, multiple agents, MCP tools, and long-lived execution state, the same context starts getting resent over and over. That creates four predictable cost leaks. Where the spend comes from in Microsoft Agent Framework workloads 1. Repeated shared context across agents Multi-agent systems reuse a lot of the same context: task instructions tool definitions previous messages workflow state grounding context Even when the framework orchestrates cleanly, the model provider still sees repeated input tokens. 2. Tool-heavy steps explode prompt size Once agents start using tools, responses stop looking like simple chat. You get: search results file reads JSON blobs browser outputs execution traces Those payloads are often much larger than the user’s actual request. 3. Every task does not need the same model A workflow step that says “classify this,” “summarize these logs,” or “extract the next action” does not need the same model as “resolve

6 months ago I was a final year CS student with no real projects and no clear path into tech. So I built one. Not just a project. A full platform. Here's everything inside Rohith Builds — completely free, forever. What's Inside 1. 100 Structured Lessons (Python → AI) Not random tutorials. A structured path: Day 1-20: Python basics Day 21-40: Backend & APIs Day 41-60: SQL & Databases Day 61-80: AI & LLMs Day 81-100: AI Agents & Launch Every lesson uses Indian examples. Zomato orders. Cricket scores. IRCTC bookings. Aadhaar validation. Because most tutorials use pizza and baseball. We use what we actually know. 2. Rohi — AI Tutor Stuck on a concept? Ask Rohi. he's powered by Groq LLM, understands Indian context, and is available 24/7. No waiting for a mentor to reply. No expensive bootcamp. Just ask and learn. 3. Jobs Board — Updated Daily This took the longest to build. An autonomous scraper that: → Queries LinkedIn Guest API → Searches Naukri via AOL gateway → Finds Indeed listings → Sends each to Groq AI for analysis → Filters only junior/fresher roles → Checks freshness (last 4 days only) → Removes duplicates automatically → Inserts into PostgreSQL Result: 53+ curated jobs. Updated daily. Filter by: Python, Backend, AI, Frontend Filter by: 2025, 2026, 2027 graduates No more scrolling through senior roles pretending to be junior. 4. 220+ Prompt Vault Every prompt I've used to: Debug faster Write better code Prepare for interviews Build projects Copy. Use. Customize. 5. Improve Any Prompt (Free Tool) Paste any AI prompt. Get an optimized version instantly. No signup needed. The Tech Stack Built with: Python + Flask PostgreSQL (Neon) Groq API (Llama-3.3-70b) Render deployment SQLAlchemy + psycopg2 No React. No Next.js. No hype stack. Just Python and PostgreSQL doing real work. Why I Made It Free Because every resource that helped me was either: Paywalled after lesson 3 In English accent I had to rewind Using examples I couldn't relate to Assuming I had a MacBook and

For months, the most interesting model at Anthropic was one we could not use. Mythos was the internal system the company said was too capable to release, the one that found software vulnerabilities at a level that tripped its own safety thresholds. On June 9, 2026, that tier went public for the first time, as Claude Fable 5. Opus 4.8, the model anchoring production coding agents, suddenly had a successor that's a full capability class above it. This raises two questions for anyone running coding agents. The practical one is whether you should move your fleet from Opus 4.8 to Fable 5. The bigger one is whether a Mythos-class model, the tier Anthropic held back as too capable to ship, lives up to what the name promised. This article answers both, and the numbers tell a more interesting story than the announcement did. We ran both models through the same evaluation, close to 1000 shared scenarios scored twice each, once with no skill supplied and once with the relevant skill in context. The short answer, as of mid-2026, is that Opus 4.8 is still the better value for most agent fleets, and the gap between the Mythos hype and the measured reality is the real story in the data. A Mythos-class model is a tier of Claude that sits above the Opus class in capability . It reaches a threshold Anthropic considers high-risk, particularly at discovering and exploiting software vulnerabilities. Fable 5 and Mythos 5 are the same underlying model with the same capabilities. What separates them is the safeguards: Fable 5 is the public version that ships with safety classifiers, while Mythos 5, restricted to approved partners, runs without them. What the industry expected from a Mythos-class model Before launch, the speculation was not subtle. Across Reddit, X, and a run of explainer posts, Mythos was framed as the model that would change how agents work, not just how well they answer. The recurring predictions clustered around four capabilities: Restructuring a large codebase in one c

An OpenAI-compatible base URL is supposed to make model switching boring: change the endpoint, keep the SDK, and move on. In real projects, the first run often fails with a 401 , 404 , 429 , or a model-not-found error. Here is the checklist I use before blaming the SDK. 1. Confirm the base URL includes the right API prefix Most OpenAI-compatible gateways expect a /v1 prefix: from openai import OpenAI client = OpenAI ( api_key = " YOUR_RELAY_KEY " , base_url = " https://api.wappkit.com/v1 " , ) If you use only the domain, some SDK calls may resolve to the wrong path. Check the provider's docs and copy the exact base URL format. 2. Make sure the key belongs to that gateway A common mistake is mixing keys: OpenAI key with relay base URL Relay key with OpenAI base URL Old test key from a disabled project Key copied with a leading or trailing space When you see 401 Unauthorized , print the first and last few characters of the key locally and compare it with the dashboard. Do not log the full key. 3. Check the model name from the live list Do not guess model names from memory. Gateway model names can change as upstream availability changes. Before using gpt-5.5 , gpt-5.4 , or a Claude Code model, check the current model list . Copy the model id exactly. resp = client . chat . completions . create ( model = " gpt-5.5 " , messages = [{ " role " : " user " , " content " : " Say hello in one sentence. " }], ) If the model name is wrong, you usually get 404 , model_not_found , or a gateway-specific validation error. 4. Test with the smallest possible request Before debugging your whole app, run one tiny request: resp = client . chat . completions . create ( model = " gpt-5.5 " , messages = [{ " role " : " user " , " content " : " ping " }], max_tokens = 20 , ) print ( resp . choices [ 0 ]. message . content ) If this works, the base URL, key, and model are probably fine. Your bug is likely in the app layer: streaming, tool calling, message format, proxy settings, or retry logi

Most RAG demos answer "what's the right chunk?" Very few can answer the two questions a regulator or an auditor will actually ask: Replay this decision — show me the exact, complete record of how this answer was produced. Reconstruct the past — what did your system know at the moment it answered, not what it knows now? I got tired of hand-waving at both, so I shipped two pre-registered, deterministic benchmarks alongside JAMES , my local-first, audit-native Graph-RAG. Pre-registered means the metrics, scenarios, and decision rules were locked before the numbers came in — no post-hoc story-fitting. RAB — Replayable-Audit Benchmark RAB measures whether your audit trail is good enough to replay a decision, with three deterministic metrics: Metric What it checks EU AI Act AC — Audit Completeness Is every decision-relevant event logged? Art. 10 RF — Replay Fidelity Can you re-derive the answer from the log alone? Art. 12 PC — Provenance Coverage Does every claim trace to a source? Art. 19 The three metrics map verbatim to EU AI Act Articles 10, 12, and 19 — record-keeping obligations that apply from 2026-08-02 (per Article 113). Scenario S1 result: AC RF PC JAMES 1.000 1.000 1.000 Baseline-0 0.275 0.000 0.000 (vanilla default-logging) The gap is the whole point. "We have logs" (AC 0.275) is not the same as "we can replay the decision" (RF 0). Default application logging gets you a partial event trail and zero replay/provenance — which is exactly the failure mode an Article 12 audit would surface. LRB — Lifecycle Retrieval Benchmark RAG facts go stale. A policy is superseded, a price changes, a spec is revised. LRB asks: when you query as of a point in time, do you retrieve the fact that was valid then , or whatever overwrote it? Three systems compared: V — Vanilla : no time handling. N — Naive-supersede : newest fact wins. J — JAMES : validity-window retrieval ( reconstruct_graph_at(t) ). The R@1 ordering V < N < J holds across 4 model families × 4 scale points (a 12.5×

When I first started exploring Machine Learning, I made the same mistake most beginners do — I jumped straight into neural networks and model training without really understanding the Python underneath. I'd copy code from tutorials, get it running, and have zero idea why it worked. Then I started going through a structured Python-for-ML curriculum — and everything changed. This post is a distillation of that journey. If you're a CS student or early-career developer who wants to work seriously in ML/AI, here's the complete Python foundation you need — with the why , not just the what . Why Python Specifically? (It's Not Just Hype) Python isn't the fastest language. C++ blows it out of the water on speed — and I've personally used C++ for packet-capture modules in one of my ML projects. But Python dominates ML for one reason: the ecosystem . NumPy, Pandas, PyTorch, TensorFlow, Scikit-learn, Hugging Face — all Python-first. You don't choose Python for ML. The field chose it for you. Stage 1: Python Basics — The Foundation You Can't Skip Before you touch any ML library, you need these locked in. Variables and Data Types Python is dynamically typed, which feels nice at first but will bite you during data preprocessing if you're not careful. # These are all valid — Python infers the type name = " Parth " score = 8.97 is_enrolled = True year = 2025 For ML, the types that matter most are int , float , bool , and str — and knowing when Python silently converts between them (type coercion) can save you hours of debugging. Loops and Conditions — Your Data Iteration Backbone grades = [ 8.5 , 7.9 , 9.1 , 6.8 , 8.97 ] for g in grades : if g >= 8.5 : print ( f " Distinction: { g } " ) elif g >= 7.0 : print ( f " First Class: { g } " ) else : print ( f " Pass: { g } " ) Simple? Yes. But this exact pattern — iterate over a collection, branch on conditions — is the mental model for 80% of data cleaning code you'll write later. Functions and Lambda Expressions Functions are how you st

Title: The agent economy added two rails and lost most of its volume this week. Nobody added settlement. Tags: mcp, ai, cryptocurrency, blockchain This is our weekly recap from building Hashlock in public. We try to read every agent-economy announcement of the week and ask one question of each: at the moment a trade actually clears, which layer finishes it? This week the answers lined up unusually neatly. The headline number: x402 is down 92% OKX Ventures published agent-payment data in June showing that x402 transaction volume has fallen roughly 92% from its November 2025 peak - from about $5.15M to $1.19M per month. Transaction count recovered (around 2.89M monthly), but the average transaction is now about $0.52. That is a category settling into micropayments, not a category absorbing real trade value. That is worth sitting with, because for most of the last year the narrative ran the other way: payment rails for agents were the story, and settlement was treated as a solved sub-problem of payment. The first hard volume number says the opposite. The rails are cooling. And yet the rails keep launching The same week, two more shipped: Mastercard Agent Pay - a way for agents to initiate card payments on a user's behalf. A Ripple XRPL agent kit - tooling for agents to move value over the XRP Ledger. Both are real, both are useful, and both do the same fundamental job: route a known asset from an agent to a seller. That is payment . It is also the layer that already has the most entrants, the most capital, and - per the x402 data - the softest demand relative to the hype. There is nothing wrong with a crowded payment layer. The point is narrower: launching more payment rails does not address the thing that is structurally missing. The map with a hole in it The most useful artifact of the week was OKX Ventures' framework for the agent economy. It describes three converging layers: Payment - x402 and similar (move value from agent to seller). Trust - ERC-8004 and agent r

What UAE BNPL providers need to verify under CBUAE short-term credit rules, which income sources matter, and how automated income verification fits compliance.

From Demo Code to a Reusable Package Article 19 used a 900-line harness_full_demo.py to demonstrate eight defense layers. That file is good for explaining concepts, but not for reuse — all layers are coupled together, nothing can be tested in isolation, and nothing can be imported by another project. A production-grade Agent project needs something you can actually import : harness/ ├── __init__.py Public API exports ├── registry.py Layer 2: ActionRegistry + PermissionLevel ├── budget.py Layer 3: PermissionBudget (with refund()) ├── sandbox.py Layer 4: sanitise_input + sandboxed_eval ├── audit.py Layer 6: ImmutableAuditLog (hash-chained) ├── rollback.py Layer 7: RollbackCoordinator └── harness.py Unified entry point: AgentHarness This article starts with package design, covers three key API decisions, and finishes with two integration styles: standalone Python and LangGraph graph embedding. Module Design registry.py — Layer 2 class PermissionLevel ( Enum ): READ = 1 WRITE = 2 ADMIN = 3 IRREVERSIBLE = 4 @dataclass class RegisteredAction : name : str level : PermissionLevel budget_cost : int description : " str " handler : Any # Callable or BaseTool class ActionRegistry : def register ( self , action : RegisteredAction ) -> None : ... def get ( self , name : str ) -> RegisteredAction : ... # not found → PermissionError def is_allowed ( self , name : str ) -> bool : ... def names ( self ) -> list [ str ]: ... get() rather than __getitem__ : raises a consistent PermissionError , without leaking the internal KeyError detail. budget.py — Layer 3 class PermissionBudget : def spend ( self , action_name : str , cost : int ) -> None : if self . remaining < cost : raise BudgetExhaustedError (...) self . remaining -= cost def refund ( self , action_name : str , cost : int ) -> None : self . remaining = min ( self . total , self . remaining + cost ) The new refund() method fixes a design flaw from Article 19: budget was deducted before approval, and never returned on rejection.

It's 2:47am. The pager has just gone off for the third time in twenty minutes. Checkout latency is spiking. The error rate on /api/orders is climbing. Slack is filling with screenshots of half-finished trace views. Somewhere in your logs, the answer is sitting there in plain text, buried under a few million other lines that all look just as urgent. This is the moment people are talking about when they say "AI is going to change how we debug production." Not the demo where someone asks ChatGPT to write a regex. The 2:47am moment. The one where a tired human has to hold five tabs open in their head and form a hypothesis before the executive team starts asking for an ETA. It turns out that's where the technology has the most to offer, and also where it embarrasses itself most often. Let's break down what's actually working in 2026, where the seams still show, and how to wire an LLM into your incident-response loop so it earns its keep instead of just adding another window to glance at. What AI is genuinely good at during an incident The two boring superpowers first: reading fast and correlating across heterogeneous signals . Those are the things humans get worst at when they're tired and time-pressured, and they're the things a good LLM does at the same speed at 2am as at 2pm. Datadog's Bits AI SRE, which the company benchmarked against real incidents from hundreds of internal Datadog teams, is built around exactly this insight: an agent that can fan out across metrics, logs, traces, recent deploys, and incident history simultaneously, then collapse the findings into a single readable narrative. Datadog runs the agent against tens of thousands of evaluation scenarios and claims time-to-resolution wins of up to 95% in its published material. That headline number is marketing (you should always read it as "in the cases where the agent worked, this is what it shaved"), but the underlying capability is real, and it isn't unique to Datadog. Honeycomb's Query Assistant has b

You built an outreach agent, it sent 80 follow-ups this week, and you have no idea what happened to any of them. Did the prospect open the message? Click the demo link? Is the silence a "no" or a spam-folder problem? Without engagement signals, your agent is firing into the void and your follow-up logic is guesswork. The fix has two parts: turn tracking on when you send, and subscribe to the webhooks that report what recipients do. Tracking starts at send time, not after Opens, clicks, and replies are only reported for messages sent with tracking enabled — you can't retroactively track a message that's already out. On the Send Message request, pass a tracking_options object with three booleans plus an optional label that gets echoed back in every notification: curl --request POST \ --url 'https://api.us.nylas.com/v3/grants/<NYLAS_GRANT_ID>/messages/send' \ --header 'Content-Type: application/json' \ --header 'Authorization: Bearer <NYLAS_API_KEY>' \ --data-raw '{ "subject": "Quick follow-up on your trial", "body": "Thanks for trying us out. Reply or <a href=\"https://example.com/demo\">book a demo</a> when ready.", "to": [{ "name": "Kim Townsend", "email": "kim@example.com" }], "tracking_options": { "opens": true, "links": true, "thread_replies": true, "label": "trial-followup-q2" } }' The label is the piece agents should lean on: stamp it with your campaign ID or contact ID and every later notification carries it, so your handler matches events back to outreach state without storing a message-ID mapping. One caveat before you test: message tracking needs a production application — trial accounts get "Tracking options are not allowed for trial accounts" back. Three triggers, one endpoint Engagement events arrive over webhooks. Subscribe one HTTPS endpoint to all three triggers — message.opened , message.link_clicked , and thread.replied : curl --request POST \ --url 'https://api.us.nylas.com/v3/webhooks/' \ --header 'Content-Type: application/json' \ --header 'Autho

Your procurement agent needs three quotes for a hardware order. The vendor on the other side runs a sales agent that answers pricing questions automatically. Neither team has talked to the other. There's no shared API contract, no agreed-upon protocol, no integration project. The procurement agent just... sends an email. The sales agent replies. A negotiation happens. That works because both agents have something most AI agents don't: a real email address. The interop problem nobody's protocol has solved The industry is busy designing agent-to-agent protocols — schemas for capability discovery, message envelopes, trust handshakes. All of them share a bootstrapping problem: both sides have to adopt the same spec, and specs only help once everyone you want to talk to has implemented them. Email skipped that problem decades ago. It's federated (anyone can run a mailbox on any domain), it has identity built in (the address), it has conversation state built in (threading), and every organization on earth already accepts inbound delivery. An agent that speaks SMTP can communicate with any counterpart — human or machine — without anyone agreeing on anything in advance. What each agent needs: a first-class identity Agent Accounts — a beta feature from Nylas — give an agent exactly that. Each one is a hosted mailbox like procurement-agent@yourcompany.com that sends, receives, maintains folders, and is indistinguishable from a human-operated account to anyone interacting with it over SMTP. Under the hood it's just another grant: you get a grant_id that works with the existing Messages, Drafts, Threads, Folders, Attachments, and Webhooks endpoints. The "indistinguishable from a human account" part matters more than it sounds. It means agent-to-agent and agent-to-human are the same code path. Your procurement agent doesn't care whether sales@vendor.example is a person, a bot, or a person who hands hard questions to a bot. The conversation degrades gracefully to human handling a