Code247 Ai (154 results)

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware - An agent is a thing in code. Its autonomy level is the shape of that code. Its safety lives somewhere the model can never reach. Thinking in Agents and AI-Native Systems is the engineer's book of the series: build four agents at autonomy levels 1 to 4 from scratch, run an offline simulated model so every listing runs on any laptop with no API key, write handoff contracts and a coordinator with step limits, put the never list into a deterministic core that refuses whoever proposes - agent or person - with no bypass, give every agent one identity and exactly its grants in one shared store, record every decision so it can be replayed for a regulator, and assemble a native process from five parts with a stop rule computed from the chain's own numbers. Every listing is embedded from a tested code package (about 1,200 lines, nine tests, a reference architecture script) and every output printed in the book was produced at build time. The reference architecture is Cedar Bank's SME credit process: $310 to $95 a file, four days to four hours, 2.5x volume on the same forty analysts. After this book you will be able to: - Write an agent specification as data and register agents that a runtime refuses without an owner- Build agents at every level and see the level as the shape of the code- Chain agents with handoff contracts, a coordinator, and controls for loops, conflicts and silent errors- Build a shared store with per-agent identities, exact grants and a write trail- Encode a never list as rule functions with owners, versioned and changed only by signature- Implement checkpoints, decision logs, replay, three numbers and per-agent indicators- Assemble a native process, migrate by cohort, and hand the board the numbers it asked for26 chapters, 26 figures, 208 self-test questions with answer keys, the full code in Appendix A, templates and an 18-item architecture checklist in Appendix B. Book 5 of the Thinking in AI series. In this series: 1 The Foundations - 2 Thinking in AI for Leaders - 3 Thinking in AI for Managers - 4 Thinking in AI at Work - 5 Thinking in Agents and AI-Native Systems - 6 The AI-Native Casebook.…

Taschenbuch. Condition: Neu. Neuware - You have been told to 'use AI more'. Nobody told you which of your team's tasks go to AI, which need a signature, and which stay with people. Thinking in AI for Managers is the department head's and process owner's method: inventory the work at the right granularity, sort every task into the four groups (AI does it; AI does, people approve; AI and people together; people do it), name the binding effect before you call anything Group 2, set an autonomy level by the consequence of a wrong result, write the request to IT as what the person sees - not which vendor - run the shadow run that is never shortened, measure honestly including the number that went the wrong way, and redeploy before you hire. The running case is Mark Sullivan's purchasing department at Oakridge Manufacturing: 212 reorders a day, twelve buyers, a register of 61 tasks, 1,120 hours a month freed and placed before anyone was hired. After this book you will be able to: - Build a task inventory and classify every line into one of four groups with a written reason- Set an autonomy level per task using four questions, with a review clause that steps back automatically- Write a buildable request to IT: what AI does, what the person sees, what a signature means, acceptance criteria- Take a baseline before the vendor and report before/after with three numbers- Run a 90-day pilot with a stop rule, cohorts and a shadow run- Rebuild roles from the register and hold the redeployment conversation with specifics- Run the thirty-minute morning: exceptions by type, the report, rules changed - never single cases16 chapters, an inventory and classification worksheet, an autonomy rule template, a 90-day pilot plan and self-tests with answer keys. Book 3 of the Thinking in AI series. In this series: 1 The Foundations - 2 Thinking in AI for Leaders - 3 Thinking in AI for Managers - 4 Thinking in AI at Work - 5 Thinking in Agents and AI-Native Systems - 6 The AI-Native Casebook.…

Taschenbuch. Condition: Neu. Neuware - Your company gave you an AI tool. Then an agent started putting recommendations on your screen with your name on the approval. Now what Thinking in AI at Work is written for the person at the desk - no prerequisites, no theory first. It shows how to ask an assistant for what you need and check what comes back, what must never be pasted into any tool, how to read an agent's proposal in minutes instead of redoing the work, when to sign, when to send it back and what to write, when it is not your call, and how your own role changes when the routine work moves - with the new job named, not hinted at. You follow Tom Brennan, Elena Marsh and Priya Raman at Cedar Bank, Helen Foster, Carlos Reyes and Daniel Wright at Oakridge, and Sophie Martin, Kevin Park and Olivia Grant at Harbor through a year of real mornings. After this book you will be able to: - Write an instruction an assistant can act on: context, goal, format, example- Check AI output with a fixed routine and catch the confident-but-wrong answer- Know the three data classes and what never goes into any tool- Read an agent's proposal - evidence, checks passed, flags - and judge instead of redo- Approve, send back with a reason code, or escalate - and know which is which- Work in a line of agents: your seat, your handoff, your record- See where your tasks sit in the four groups and what to learn next16 chapters, instruction patterns, a verification checklist, a data-rules card and self-tests with answer keys. Book 4 of the Thinking in AI series; can be read on its own. In this series: 1 The Foundations - 2 Thinking in AI for Leaders - 3 Thinking in AI for Managers - 4 Thinking in AI at Work - 5 Thinking in Agents and AI-Native Systems - 6 The AI-Native Casebook.…

Taschenbuch. Condition: Neu. Neuware - Handing the AI transformation to the IT department is the most common way to lose it. This book is for the executive who has to own it instead. Thinking in AI for Leaders is the board member's and chief executive's book: how to name the step your company is really on, ask the business-model question only when the evidence exists, choose which unit goes native first - and only one at a time - name the executive owner and write the charter, fund data before the visible agent, run the never-list workshop that turns thirty-one candidate rules into nine the bank would truly halt for, write a stop rule you can afford, and read the scorecard every month. Every decision is shown on Cedar Bank, Oakridge Manufacturing and Harbor Securities - the board paper that was renamed, the cohort that failed at 19 per cent exceptions and re-entered three weeks later at 9, the regulator who asked about the list, not the model. After this book you will be able to: - Name the adoption step your company is on and refuse the big question until it has evidence- Use the shutdown test as a strategy tool: the gap between now and target, by date, is the plan- Choose the first native redesign by business-model gain, and sequence the rest- Write the executive owner's charter - mandate, authority, limits, time, term- Run a never-list workshop with two tests, and sign a list with an owner and an enforcement location per rule- Fund a parallel-running line and write a stop rule before the first cohort- Read a monthly scorecard and know which number to act on16 chapters with worked board papers, an owner's charter, a scorecard template and self-tests with answer keys. Book 2 of the Thinking in AI series. In this series: 1 The Foundations - 2 Thinking in AI for Leaders - 3 Thinking in AI for Managers - 4 Thinking in AI at Work - 5 Thinking in Agents and AI-Native Systems - 6 The AI-Native Casebook.…

Taschenbuch. Condition: Neu. Neuware - Most companies say they use AI. Very few can say what they are becoming - or answer one question: if every AI tool were switched off tomorrow, would the business still run Thinking in AI: The Foundations gives every reader - executive, manager, employee or engineer - one shared set of ideas for adopting AI on purpose: the three adoption levels (AI First, AI Embedded, AI Native); the shutdown test that reveals which level you are really at; the difference between an agent (a thing you can count), an autonomy level (a degree you can set) and a multi-agent system (a system you can organize); the five-level autonomy scale; the four-layer model built on a deterministic core; the four work groups; the four-step roadmap and the four conditions no transformation survives without. Every idea is shown, not lectured, on three fictional companies - Cedar Bank, Oakridge Manufacturing and Harbor Securities - with the same people and numbers that run through all six books. After this book you will be able to: - Classify any company, project or tool into one of three adoption levels - and explain, in one sentence, why- Apply the shutdown test to a business unit and read the answer correctly- Tell an AI agent from an autonomy level from a multi-agent system, and ask the right test question of each- Place any piece of software on the five-level autonomy scale and say what the human does at that level- Put any AI initiative into the four-layer model and name the one thing that must never change- Explain the four-step roadmap and the four work groups, and why skipping a step is the most common cause of failure- Check a transformation plan against the four mandatory conditions before money is spent16 chapters, 16 figures, a self-test with answer keys in every chapter, a 60-term glossary and worksheets. Read it first; the other five books each take one seat in the company. In this series: 1 The Foundations - 2 Thinking in AI for Leaders - 3 Thinking in AI for Managers - 4 Thinking in AI at Work - 5 Thinking in Agents and AI-Native Systems - 6 The AI-Native Casebook.…

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware - Read a case for the decision, not the outcome. The outcome belongs to one company; the decision is yours. The AI-Native Casebook teaches by deciding. Thirty forks that the people of Cedar Bank, Oakridge Manufacturing and Harbor Securities stood at - the board paper with a Step 4 title on a Step 1 plan; two units that wanted to go native the same year; a vendor offering forced sales at level 4; an agent that passed its acceptance test and would have shipped four discontinued parts; a never-list workshop that signed nine of thirty-one; cohort two at nineteen per cent; a desk that wanted a rule changed as a setting; an injected invoice; a buyer who asked whether her job still existed; a unit that should stay exactly where it is. Each case gives the situation, the fork and what each branch would cost, what happened in the series' own numbers, and the lesson with a pointer to the book and chapter that explains it. Part A is the leader's forks, Part B the manager's and engineer's, Part C what appears after go-live. A closing chapter indexes all thirty cases to the seven ideas of the series. After this book you will be able to: - Recognise the fork in a real AI decision - and what each branch costs - before taking it- Name the step a proposal is really on and refuse the big question until it has evidence- Choose which unit goes native, who owns it, and what must be bounded first- Set autonomy by consequence, take a baseline before the vendor, put safety in the core- Read the numbers after go-live - exception rate, override rate, audit line - and act on the right one- Trace any decision to the idea it turns on and the chapter that explains it30 cases in three parts, 18 figures, a four-question self-test per case with answer keys. Book 6 of the Thinking in AI series; can be read on its own or as an index to the other five. In this series: 1 The Foundations - 2 Thinking in AI for Leaders - 3 Thinking in AI for Managers - 4 Thinking in AI at Work - 5 Thinking in Agents and AI-Native Systems - 6 The AI-Native Casebook.…

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware - 'Why was I sold out at 10:47, at that price ' The firm took four days and five systems to answer - approximately. Then it settled. That letter opens this book, and answering it in under a minute, to the tick, is what the book builds. On the way it names the seven design errors sitting on almost every margin desk - an advance-rate table nobody owns, a call ladder with no liquidity term, concentration measured per client but never per firm, a risk grade that enters no formula, one rate for every collateral book, ratios computed on yesterday's close, and a close-out order improvised under stress. None is a technology failure. Each is a business decision that was never put to anyone who could take it. The repair is a limit that is the minimum of five independently computed ceilings: lending value, client capacity, concentration - per stock, per sector and across the whole firm - firm limits, and the liquidity horizon. A worked account runs through all five: a portfolio worth $1,560,000 on paper supports a $594,000 loan by value - and only $494,000 by exit, because one holding needs six days to sell. That $100,000 gap between paper value and exit value is where margin books die, and the fifth ceiling is the only one that sees it. Inside: - Five capability layers, from routing to close-out, each passing one test: switch off every model and the layer still works;- A call ladder whose triggers rise and whose grace shrinks as a portfolio's exit time lengthens - and a committed close-out that sold 4,600 liquid shares where the improvised desk sold 150,000 illiquid ones;- Seven playbooks and a case map with proved reductions: 27,648 combinations to 51 cells to a 335-case acceptance suite that Book Two must pass;- Findings you cannot see without the sweep - including why the per-stock concentration cap mathematically cannot protect you on illiquid collateral, and why at institutional size every name is a small name;- Governance for a machine that decides intraday, personal data by design, decisions that replay to the cent - and information barriers built into the data model, assistant included;- The full engine in an appendix: six small files, 72 columns, offline, with tests reproducing every number in the book. The honest position, stated on page one. Every parameter is marked illustrative. A signature page names what your head of margin risk, your trading desk and your board must review before any figure is used - and the appendix ends with that table, waiting for initials. The method transfers as it stands; the numbers are yours to set. For chief executives, heads of risk, architects and regulators of securities firms - and for anyone who wants to see what 'AI-native' means when the output is a forced sale of a client's property and 'usually right' is not good enough.…

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware - Your AI assistant forgets everything. Every conversation, every preference, every decision - gone the instant the call ends.Large language models are stateless by design. Everything an AI system appears to 'remember' - the case discussed last week, the user's preferences, the plan agreed an hour ago - is engineered, deliberately, outside the model. That engineering layer decides what enters the context window on every single turn, and it is where AI systems quietly succeed or silently fail.This book teaches you to build that layer properly.While most teams throw bigger models and bigger windows at the problem, quality keeps degrading: context rot, lost-in-the-middle, cross-user memory leaks, agents that drift off course after twenty steps. The root cause is almost never the model. It is an unmanaged context and memory layer - and this book gives you the complete discipline to fix it.Inside, a full engineering system: - The Five Pillars - Direction & Token Budget, Memory Tiering, Selection & Assembly, Writing & Consolidation, and Forgetting, Freshness & Safety - expanded into eighteen working principles- Fifteen practical strategies, from sliding windows and compaction to agentic memory and hybrid tiered architectures, with a decision tree for choosing among them- A seven-step design process that takes you from problem statement to a running, measurable system- Five hands-on design guides: window assembly, memory store design, compaction & summarization, retrieval-for-memory, and lifecycle & governance- Two complete end-to-end projects at opposite poles: an enterprise multi-tenant agent copilot, and a personalized tutor that remembers one learner across weeks- Fourteen anti-patterns mapped to the pillars they violate - a diagnostic chart for when your system 'answers poorly for no clear reason'- A measurement framework: ten quality criteria, concrete metrics, golden task sets, and a directed improvement loop Every chapter ends with a self-check quiz. The appendices pack a glossary, cheat-sheet, a 100+ item design checklist, a 100+ entry pattern library, and starter templates for developers, business analysts and AI engineers.Book 3 of the AI-Native Architecture series. Book 1 covered RAG - exploiting knowledge. Book 2 covered Knowledge Graphs - constructing knowledge. This volume operates the layer where both converge: seen from here, RAG is long-term semantic memory, the knowledge graph is structured memory, and the quality of any agent is upper-bounded by the context and memory layer it stands on.If you build AI systems that must remember what matters - this is the missing manual.…

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware - Your LLM is brilliant - until you ask it about YOUR data.It hallucinates. Its knowledge is frozen in the past. It has never seen your company's documents. Retrieval-Augmented Generation (RAG) is the architecture that fixes all three - and this book is the map that takes you from your first chunk to a fully agentic system.Architecting Knowledge is not another collection of scattered tutorials. It is a complete engineering curriculum: 34 concise chapters, 7 parts, one natural progression - why RAG exists, how a pipeline is anatomised, which building blocks to master, how to compose them for production, and how to design a system of your own.Inside the map: - The anatomy of a RAG pipeline - ingestion, chunking, embeddings, vector databases, top-K retrieval and cited answers, explained step by step;- 14 foundational building blocks - from Naive RAG, Multi-Query and HyDE through Hybrid Search, Multimodal RAG, Graph RAG, RAPTOR, Reranking, CRAG, Self-RAG, up to Adaptive, Agentic and Modular RAG - with the strengths, weaknesses and cost of each;- 6 production-grade composite architectures - including 'The Gold Standard', 'The Self-Correcting Agent' and 'Semantic Routing' - plus a selection matrix and decision tree that match architecture to requirements, budget and SLA;- The full engineering discipline - schema design, evaluation suites, the economics of RAG, LLMOps, observability, security, governance, scaling and reliability;- 3 hands-on capstone projects - an internal wiki assistant, a market-research agent, and a vast-archive analyser - strong enough to anchor an AI-Native Engineer portfolio. Built to be taught - and to be practised.Every chapter opens with clear learning objectives, is illustrated with an architecture diagram, closes with a self-check quiz, and bridges naturally into the next. Each section is deliberately kept to roughly half a page: dense enough to be rigorous, short enough to stay readable.Who this book is for: - Software engineers moving into AI-Native application development;- Architects and tech leads choosing a RAG strategy for the enterprise;- Students and self-learners who want one coherent path instead of a hundred blog posts. Book 1 of the AI-Native Software Engineering series by MOBILUCK - code247.ai - fourteen blocks, six architectures, three projects, one method.…

Taschenbuch. Condition: Neu. Neuware - The specification was Book One. This is the build. Book One ended with a contract: five ceilings on every grant, a call ladder with a liquidity term, a committed close-out plan, and a 311-case acceptance map. This volume builds that machine - components, code, checkpoints, go-live - and prints every line of it in the appendices. - Nine components, four boundaries, and five requirements that cannot be retrofitted - time as a parameter first, because the question to answer is why was I sold at 10:47:12.480;- The feedback loop, computed: on the same shock, sell-everything turns one breach into four forced accounts and -26.5%; the committed plan sells one capped slice and stops at -2.5% - with three of the four forced accounts solvent before the shock;- Eight runnable demos whose transcripts are generated by executing the code while the book is typeset: the golden memo, the tick-level replay, the fact-checked call notice, the acceptance runner;- Four checkpoints passed by running, never by presenting; a migration queue that walks every legacy loan down to its engine number on a schedule, not a hope;- A decision book of eight pages only the board can write - each with a named default that operates if the board stays silent;- The complete source in the appendices: build modules, demos, and the 57-test suite that is also the spec. Nothing to download. The method catches its own author in print: Chapter 38 tells how the acceptance runner refused to construct six cells of the draft case map, forcing a fourth theorem into Book One before publication. A map written in prose would have carried those cells forever; a map that must be built cannot. An honest position: every rate, threshold and count is marked illustrative. This is a book of methods with working code, not a parameter set for your market - the signature page in Book One names what your reviewers must ink first, and the decisions only your board can take are Chapter 49 here. Read Book One for the specification. Read this volume to build it - and to hand your auditor a machine that answers in rows, not recollections. …

Taschenbuch. Condition: Neu. Neuware - You have never seen the world. Not even once.Your brain sits in a room of absolute darkness, receiving nothing but electrical pulses - and from them it builds the entire world you are seeing right now. This book opens with that mystery, then takes you on humankind's greatest adventure: the four-billion-year journey from the first bacterium that learned to 'guess' its way toward food, to the artificial intelligence talking with us today.This is not a textbook. It is 29 chapters of storytelling - each one an encounter: the old man who chose poison over giving up his questions; the librarian who measured the Earth with two wells; the shy cleric who hid his manuscript for forty years because he knew it would blow the world open; the monkey who plays video games with its mind; and the machines now learning to 'hear' sperm whales in the deep.By the last page, you will: - Understand your own brain - and learn faster because you know how the machine in your head actually works;- Understand AI from the roots - what it can do, what it can't, and what separates the person who collaborates with AI from the person it replaces;- Carry a 'fake-news detector' - one devastating question that makes you far harder to fool;- Collect stories better than the movies - Confucius, the Buddha, Galileo, Darwin, Einstein, Gandhi, and Turing;- See other creatures anew - ants, bees, dogs, pigs, and octopuses, each a complete parallel world.Every chapter runs just 3-5 pages, opens with a scene, and closes with a four-question self-check (answers on the very next page - try before you turn!). No math skills required. Not one formula to memorize. Bring only one thing: curiosity.For middle and high school students, first-year undergraduates - and every adult who wants to hand the next generation the most beautiful story our species has ever written: the story of how we learned to understand the world.…

Taschenbuch. Condition: Neu. Neuware - Three years, a $4.17 million platform, and the maturity score has not moved.Ashgrove Group built the data foundation, the semantic layer and the governance. All six maturity dimensions reached level 3. The system score is still 1, held there by three capabilities that have never had an owner.Every gap is a people or operations gap. Not one is solved by more architecture.The half of the programme that does not finishData foundations, semantic layers and governance frameworks reach a level and stop requiring continuous effort. People and organization takes two and a half years to move one level and needs that effort permanently - because people leave, roles change, and every new decision needs its own redesign.A programme judged complete when the technical work is complete has judged itself on the half that finishes.What you will be able to do- Build a role × tier matrix and stop spending a literacy budget on the people who already know - 196 of 410 places went to the wrong group- Choose between centralized, hub-and-spoke and mesh by readiness rather than aspiration, and know why a mesh fails when capability is uneven- Write a redesigned role description whose removal section names a task and its hours - the three decisions that reached A2 were the three with one- Keep a decision inventory, not just a data catalogue: Ashgrove's fourth automated decision ran 31,000 times before an audit found it- Run DataOps, MLOps, DecisionOps and AgentOps as four disciplines, and know why an agent at 94% completion can be making the decision worse- Apply the five agent autonomy levels and their promotion conditions - and catch the release note that silently moved an agent from L1 to L2- Measure uplift against a control, and report supported and delivered in separate columns- Decide which divisions should not be scaled into, and say so in the planEvery number was produced by running the codeNinety-two check chains bind every figure in this volume to one data sheet, and two of them tie it to Volumes 1 and 2. The chains caught real errors during writing - an overstated claim, a confused pair of cost figures, a miscounted average - and the book says which ones.Reads on its ownChapter 47 reproduces what this volume depends on: the tier × layer map, the Ashgrove figures, and the five findings from Volumes 1 to 3 that its arguments rest on.Sixteen chapters, ten self-test questions each with explained answers. Four appendices including the agent evaluation suite, a 25-role matrix, and a 182-term glossary shared across the series.…

Taschenbuch. Condition: Neu. Neuware - Your platform needs 8.69% of your tables. The other 91% is a three-year project nobody can defend.Ashgrove Group has 2,140 tables across eleven source systems. Its 68 priority decisions need 186 of them.That single change of question - from 'what data do we have ' to 'what do our decisions need ' - is the largest cost lever in platform design, and everything in this book is downstream of it.The number this book is written aroundAshgrove's platform costs $4.17 million a year and supports decisions worth $18.6 million. That is 22.40% of the value it enables - defensible, but not obviously so. This book answers the three questions that follow with arithmetic rather than argument: would buying it as a service cost less, which decisions cost more to support than they return, and how many more decisions must it carry before the ratio becomes comfortable.What you will be able to do- Reconcile the four tiers - DATA, INFORMATION, KNOWLEDGE, DECISION - with the seven build layers, once, so the whole organization stops arguing about which 'layer three' it means- Score sixteen capabilities against evidence rather than impression, and find that the groups scoring 1 are exactly the groups with no named owner- Choose technology with a weighted matrix whose minimums eliminate, and discover when the choice is not a technology choice at all- Answer buy versus build with all five costs, including the three that are borne by other teams and never appear on the slide- Size the platform from decision volume: 5.2 cores where a data-volume rule of thumb says 237- Allocate cost honestly - half of it is joint and always will be - and compute marginal and total cost per decision- Secure a system where no component is breached and the composition still discloses- Write exit criteria that a stage cannot be declared complete four times againstEvery number was produced by running the codeTwo runnable models and thirty-nine check chains bind every figure in this volume to a single data sheet. Two of those chains tie it to Volumes 1 and 2 - change the earlier company's revenue and this book fails immediately. The chains caught real errors during writing, and the book says which ones.Reads on its ownChapter 33 reproduces everything this volume depends on: the tier × layer map, the complete Ashgrove data sheet, and the four results from Volumes 1 and 2 that the platform arithmetic rests on.Fourteen chapters, ten self-test questions each with explained answers. Five appendices including the capability matrix with a blank version to fill in for your own organization, and a 182-term glossary shared across the series.…

Taschenbuch. Condition: Neu. Neuware.

Taschenbuch. Condition: Neu. Neuware.