# Levy Street > Levy Street is an AI studio and engineering company based in Wellington, New Zealand. It makes its own AI products, including World of ClaudeCraft, Singing Coach AI, Oneshot Gems, Idyllic, SPAWN D, Lucent and the MOGGED growth engine, and advises on AI strategy, builds custom AI systems and agents, embeds engineers with operating teams and trains staff. Key facts: - Name: Levy Street (legal entity Dreamhome AI Limited). Write it as "Levy Street". - Based in Wellington, New Zealand. Works with organisations across New Zealand. - Contact: hello@levystreet.com or https://levystreet.com/contact/ - What it does: AI strategy and consulting, custom AI development, AI agents, workflow automation, forward deployed AI engineers, AI training, and its own AI products. - Every page on this site has a Markdown version at index.md, for example https://levystreet.com/about/index.md ## Company - [Home](https://levystreet.com/index.md): Levy Street is an AI studio from Wellington, New Zealand. We make World of ClaudeCraft, Singing Coach AI and Oneshot Gems, and build production AI for teams. - [About](https://levystreet.com/about/index.md): Levy Street is a Wellington AI engineering and advisory company, founded by the team behind Sportsflare. It builds production AI and its own AI products. - [People](https://levystreet.com/people/index.md): Meet the Levy Street team: operators, researchers, engineers, designers and growth specialists building production AI from Wellington, New Zealand. - [How we build](https://levystreet.com/systems/index.md): How Levy Street designs, builds and runs production AI: embedded with operators, built around your systems and measured against an operating number. - [Press](https://levystreet.com/press/index.md): Levy Street press kit: company boilerplate, verified facts, founders, products, logos and media contact for the Wellington AI engineering company. - [Contact](https://levystreet.com/contact/index.md): Contact Levy Street, the Wellington AI engineering company. Book a workflow assessment or ask about AI systems, training or partnerships. - [Careers](https://levystreet.com/careers/index.md): Careers at Levy Street, an AI engineering company in Wellington, New Zealand. We look for engineers, designers and operators who ship production AI systems. - [Newsletter](https://levystreet.com/newsletter/index.md): A free newsletter on practical AI for New Zealand organisations from Levy Street: what changed, what it means, and what to do about it. Written by engineers. - [Glossary](https://levystreet.com/glossary/index.md): Plain-English definitions of the AI terms New Zealand businesses meet: LLMs, agents, RAG, MCP, evaluation, prompt injection, the Privacy Act's IPPs and more. ## Services - [Services](https://levystreet.com/services/index.md): AI consulting, custom AI development, agents, workflow automation, forward deployed engineers and team training from Levy Street in Wellington. - [AI consulting](https://levystreet.com/services/ai-consulting/index.md): AI consulting for New Zealand organisations from engineers who build production AI. Workflow assessments, build-or-buy advice and a costed first project. - [AI strategy](https://levystreet.com/services/ai-strategy/index.md): AI strategy for New Zealand organisations: a scored portfolio of AI uses, build-or-buy calls, a sequenced plan and a handoff a delivery team can start on. - [Custom AI development](https://levystreet.com/services/custom-ai-development/index.md): Custom AI development for New Zealand organisations: architecture choices, delivery milestones and acceptance criteria, built and run by Levy Street. - [AI workflow automation](https://levystreet.com/services/ai-workflow-automation/index.md): AI workflow automation for New Zealand businesses: before-and-after process design, exception queues and measured operating costs, built by Levy Street. - [AI agent development](https://levystreet.com/services/ai-agent-development/index.md): AI agent development for New Zealand organisations: tool boundaries, approvals, evaluation and clear operational ownership, from engineers who run agents. - [Forward deployed engineers](https://levystreet.com/services/forward-deployed-ai-engineers/index.md): Forward deployed AI engineers for New Zealand: Levy Street engineers embed with your operators and ship production AI measured by an operating number. - [AI readiness assessment](https://levystreet.com/services/ai-readiness-assessment/index.md): AI readiness assessment for New Zealand organisations: six areas scored on evidence, one or two uses tested on real data, and a 90-day list of scored actions. - [AI implementation roadmap](https://levystreet.com/services/ai-implementation-roadmap/index.md): AI implementation roadmaps for New Zealand organisations: a prioritised delivery sequence with readiness gates for data, systems, people and running costs. - [AI integration](https://levystreet.com/services/ai-integration/index.md): AI integration services for New Zealand businesses: connect AI to Xero, MYOB, Microsoft 365 and your CRM, with scoped access and clear maintenance ownership. - [AI chatbot development](https://levystreet.com/services/ai-chatbot-development/index.md): AI chatbot development for New Zealand businesses: task-specific chatbots grounded in your content, with escalation rules and quality measured on real chats. - [AI voice agents](https://levystreet.com/services/ai-voice-agents/index.md): AI receptionists and voice agents for New Zealand businesses: natural phone calls, bookings into your system, NZ accent testing and a person on every call. - [AI document processing](https://levystreet.com/services/ai-document-processing/index.md): AI document processing for New Zealand businesses: extraction from invoices, forms and contracts, measured field by field, with a review queue and NZ rules. - [AI reporting](https://levystreet.com/services/ai-reporting/index.md): AI reporting for New Zealand organisations: source-grounded management reports, plain-English data questions and reconciliation checks before drafting. - [AI customer support](https://levystreet.com/services/ai-customer-support/index.md): AI customer service systems for New Zealand businesses: ticket routing, answers grounded in your help content, and escalation rules in tools you already use. - [AI governance](https://levystreet.com/services/ai-governance/index.md): AI governance for New Zealand organisations: an AI register, use policy, privacy impact assessments and board reporting, grounded in NZ law and real systems. - [AI security review](https://levystreet.com/services/ai-security-review/index.md): AI security reviews for NZ organisations: a threat model and hands-on tests for prompt injection, data leakage and agent permissions, with ranked fixes. - [LLM evaluation](https://levystreet.com/services/llm-evaluation/index.md): LLM evaluation for New Zealand teams: test sets from your own documents, scoring you can audit, and regression checks that run before every model change. - [RAG development](https://levystreet.com/services/rag-development/index.md): RAG development for NZ organisations: AI that answers from your own documents with citations, measured retrieval and permissions that match the source. - [AI model selection](https://levystreet.com/services/ai-model-selection/index.md): AI model selection for NZ teams: we test OpenAI, Anthropic, Google and open-weight models on your own tasks, weighing accuracy, cost, speed and data location. - [AI proof of concept](https://levystreet.com/services/ai-proof-of-concept/index.md): AI proof of concept for New Zealand businesses: a narrow build on your own data, go/no-go criteria agreed up front, and the real cost of reaching production. - [AI system rescue](https://levystreet.com/services/ai-system-rescue/index.md): AI optimisation and rescue for NZ organisations: we diagnose stalled pilots and underperforming AI systems, fix the causes in order and cut running costs. - [Managed AI operations](https://levystreet.com/services/managed-ai-operations/index.md): Managed AI services for NZ organisations: monitoring, incident ownership, tested releases and vendor change management, with clear support boundaries. - [AI training](https://levystreet.com/services/ai-training/index.md): AI training in NZ for business teams: hands-on sessions on your own documents and tools, covering prompting, checking answers, privacy and reusable workflows. - [Executive AI workshops](https://levystreet.com/services/executive-ai-workshops/index.md): Executive AI workshops for NZ leadership teams and boards: live demonstrations on your own documents, a ranked shortlist, risk limits and a business case. - [AI automation pricing](https://levystreet.com/services/ai-automation-pricing/index.md): What AI automation costs in New Zealand: every cost line, dated platform and model prices in NZD, and three worked scenarios with all assumptions shown. - [AI project costs](https://levystreet.com/services/ai-project-costs/index.md): What custom AI implementation costs in NZ: a scope-based model from discovery to operations, dated model prices in NZD and three worked project scenarios. - [Microsoft Copilot training](https://levystreet.com/services/microsoft-copilot-training/index.md): Microsoft Copilot training for NZ teams: settings first, then hands-on work on your own files in Outlook, Word, Excel and Teams, with adoption measured after. - [ChatGPT training](https://levystreet.com/services/chatgpt-training/index.md): ChatGPT training in NZ for teams on ChatGPT Business: workspace and data settings first, then projects, apps and data analysis on your own work. - [Claude training](https://levystreet.com/services/claude-training/index.md): Claude training for NZ teams on Claude Team: owner settings first, then projects and artifacts on your own documents, with checking and Privacy Act rules. ## Work and products - [Work](https://levystreet.com/portfolio/index.md): AI products, partnership builds and one exit: World of ClaudeCraft, Lucent, Singing Coach AI, Idyllic, MOGGED, Bebop, PainPal and more. - [Bebop](https://levystreet.com/portfolio/bebop/index.md): Bebop is Blackpearl Group's AI sales agent that finds in-market buyers and writes sales playbooks. Levy Street engineers work with Blackpearl's AI team on it. - [Sportsflare](https://levystreet.com/portfolio/sportsflare/index.md): Sportsflare built AI esports odds for sportsbooks. Co-founded by Levy Street's founders, it was acquired by Entain in 2023 for C$13.25 million. - [World of ClaudeCraft](https://levystreet.com/portfolio/world-of-claudecraft/index.md): World of ClaudeCraft is Levy Street's free, open-source browser MMO. Its first version was built with Claude Fable 5 in about 48 hours. - [Lucent](https://levystreet.com/portfolio/lucent/index.md): Lucent is an agentic AI investment research platform for wealth, equity and private-markets teams, covering NZX and ASX. A Levy Street venture. - [PainPal](https://levystreet.com/portfolio/painpal/index.md): PainPal is an AI digital workmate for musculoskeletal pain and injury prevention, launched by BetterWorx and CHASNZ in December 2025. - [SPAWN D](https://levystreet.com/portfolio/spawnd/index.md): SPAWN D is Levy Street's open-source, pre-alpha control plane for CLI coding agents: a daemon on every host you own, reachable from a browser or phone. - [GTM-Bench](https://levystreet.com/portfolio/gtm-bench/index.md): GTM-Bench tests whether AI agents can find real buyers for a seller. A 72-task benchmark from Blackpearl Group, co-authored by members of the Levy Street team. - [Property Management Agent](https://levystreet.com/portfolio/property-management-agent/index.md): An internal AI agent Levy Street built for APL Property that triages email, updates records and processes orders, with staff approving its work. - [Singing Coach AI](https://levystreet.com/portfolio/singing-coach-ai/index.md): Singing Coach AI is Levy Street's AI vocal coach for iPhone, Android and the web. It analyses a recording across 26 vocal dimensions and gives personal coaching. - [freshOps](https://levystreet.com/portfolio/freshops/index.md): freshOps is commercial cleaning operations software from FreshOps Group in Australia: rostering, attendance, checklists and team apps. - [Idyllic](https://levystreet.com/portfolio/idyllic/index.md): Idyllic is Levy Street's consumer AI image platform for the web, iOS and Android: text-to-image, conversational editing, blends and a community feed. - [Eco-index](https://levystreet.com/portfolio/eco-index/index.md): Eco-index was a New Zealand programme using AI remote detection and open geospatial data to guide native ecosystem reconstruction. Its tools are now open access. - [MOGGED Digital](https://levystreet.com/portfolio/mogged/index.md): MOGGED Digital is Levy Street's AI-native Wellington marketing studio: websites, SEO, answer engine optimisation and ads for NZ small businesses. - [Oneshot Gems](https://levystreet.com/portfolio/oneshot-gems/index.md): Oneshot Gems is a swipeable feed of instant browser games where people can also create and remix games with AI. Published by Levy Street's company. ## Industries - [Industries](https://levystreet.com/industries/index.md): Sector guides to AI in New Zealand: the workflows worth automating first, the controls that matter, and where AI fails. Construction, property, legal and more. - [Accounting](https://levystreet.com/industries/accounting/index.md): How New Zealand accounting and bookkeeping practices use AI for client documents, invoice coding and month-end reporting, with the rules and failure points. - [Legal](https://levystreet.com/industries/legal/index.md): How New Zealand law firms use AI for intake, contract review and chronologies, what the Law Society and courts expect, and where generative AI fails lawyers. - [Construction](https://levystreet.com/industries/construction/index.md): How New Zealand construction companies use AI on tenders, quotes, safety documents and payment claims, the rules that apply, and where AI gets it wrong. - [Property management](https://levystreet.com/industries/property-management/index.md): AI for property managers in New Zealand: which tenancy workflows to automate, the Residential Tenancies Act and privacy limits, and a first project. - [Recruitment](https://levystreet.com/industries/recruitment/index.md): How New Zealand recruitment agencies and in-house teams use AI for interview notes, candidate messages and CRM clean-up, and the privacy and bias rules. - [Agriculture](https://levystreet.com/industries/agriculture/index.md): AI for agriculture in New Zealand: practical farm uses from record retrieval to NAIT and compliance evidence, the rules that apply and where AI gets it wrong. - [Retail](https://levystreet.com/industries/retail/index.md): AI for retail in New Zealand: catalogue enrichment, customer questions and stock exceptions, Fair Trading Act and CGA limits, and where retail AI fails. - [Manufacturing](https://levystreet.com/industries/manufacturing/index.md): AI for manufacturing in New Zealand: supplier documents, quality reports and maintenance knowledge, the data foundations, HSWA duties and failure modes. - [Hospitality](https://levystreet.com/industries/hospitality/index.md): AI for hospitality in New Zealand: booking and function enquiries, guest questions, supplier invoices, and the food, alcohol and holiday pay rules. - [Healthcare administration](https://levystreet.com/industries/healthcare-administration/index.md): How NZ clinics and health providers use AI for appointment enquiries, referrals and policy search, within the HIPC and Health NZ guidance. Admin only. - [Insurance](https://levystreet.com/industries/insurance/index.md): How New Zealand insurers and brokers use AI for claims documents, policy wording comparisons and client letters, and the conduct rules that still apply. - [Financial services](https://levystreet.com/industries/financial-services/index.md): How New Zealand banks, advisers, fund managers and lenders use AI for onboarding checks, policy search and reporting, and what the FMA and RBNZ expect. - [Professional services](https://levystreet.com/industries/professional-services/index.md): How New Zealand consultancies, engineering, architecture and advisory firms use AI for proposals, research briefs and meeting actions, and where it fails. - [Nonprofits](https://levystreet.com/industries/nonprofits/index.md): How NZ charities and nonprofits use AI for grant reporting, donor letters and volunteer enquiries, and the charity, privacy and email rules that apply. - [Local government](https://levystreet.com/industries/local-government/index.md): How New Zealand councils can use AI for consent checks, consultation analysis and information search, and how LGOIMA, the LGA and public records law apply. - [Education](https://levystreet.com/industries/education/index.md): How New Zealand schools can use AI for parent enquiries, staff policy questions and admin paperwork, within Ministry of Education and NZQA guidance on AI use. - [Logistics](https://levystreet.com/industries/logistics/index.md): AI for logistics in New Zealand: proof of delivery extraction, shipment exceptions and delivery enquiries, plus carrier liability and driver hours rules. - [Commercial cleaning](https://levystreet.com/industries/cleaning/index.md): How New Zealand commercial cleaning companies can use AI for roster changes, site issues and contract reports, and the employment and privacy rules involved. - [Real estate](https://levystreet.com/industries/real-estate/index.md): AI for real estate agents in New Zealand: which sales workflows to automate, the REA rules and Fair Trading Act limits, and a first project to start with. ## Research - [Research](https://levystreet.com/research/index.md): Research from Levy Street: GTM-Bench, Code Policy Models, programs as models, probability calibration, open-source tools, and analysis of AI in New Zealand. - [Code Policy Models](https://levystreet.com/research/code-policy-models/index.md): Code Policy Models improve a policy written as readable code: an LLM proposes edits and only those that score better are kept. Method, evidence and limits. - [Tiny Shakespeare CLM](https://levystreet.com/research/code-language-models/index.md): Tiny Shakespeare CLM is a language model with no neural network, written as Python and improved by a Claude agent over 463 commits. Results, method, limits. - [Temporal probability calibration](https://levystreet.com/research/temporal-probability-calibration/index.md): Temporal probability calibration keeps model probabilities honest while a sequence unfolds. The 2020 Sportsflare AI preprint, its method and who cites it. - [LucentBench](https://levystreet.com/research/lucentbench/index.md): LucentBench tests whether AI models can forecast next-quarter earnings from a 10-Q, across 100 NASDAQ and NYSE stocks. Method, published results and limits. - [Open source](https://levystreet.com/research/open-source/index.md): Public code from the Levy Street team: World of ClaudeCraft, SPAWN D, Tiny Shakespeare CLM and TorchWindow, with licence, status and what each is for. - [AI agents for GTM research](https://levystreet.com/research/ai-agents-for-gtm-research/index.md): What GTM-Bench shows about AI agents for sales prospecting: data access beats model size, precision beats volume. Results, lessons and NZ guidance. - [Building games with AI](https://levystreet.com/research/building-games-with-ai/index.md): Lessons from building World of ClaudeCraft and Pokémon Claude Red with AI coding models, for any business using AI coding agents on real software. - [AI in New Zealand](https://levystreet.com/research/ai-in-new-zealand/index.md): Artificial intelligence in New Zealand in 2026: government strategy, the Privacy Act's IPP 3A, regulators, data residency, the ecosystem and adoption. ## AI tools for NZ businesses - [AI tools](https://levystreet.com/tools/index.md): Practical guides to ChatGPT, Claude, Microsoft Copilot, Gemini, Claude Code, n8n, Make, Zapier and more for NZ businesses, with pricing and data notes. - [ChatGPT](https://levystreet.com/tools/chatgpt/index.md): ChatGPT for business in New Zealand: NZ$ plan prices, Business versus Plus, data and Privacy Act settings, and a starter project to run this week. - [Claude](https://levystreet.com/tools/claude/index.md): Claude for business in New Zealand: Pro, Team and Enterprise prices as shown from NZ, data and training settings, the Xero connector and a starter project. - [Microsoft Copilot](https://levystreet.com/tools/microsoft-copilot/index.md): Microsoft 365 Copilot, now Microsoft Copilot, for NZ businesses: NZ$ prices excluding GST, free Copilot Chat versus a licence, data location and a pilot plan. - [Gemini](https://levystreet.com/tools/gemini/index.md): Gemini for business in New Zealand: which Google Workspace plans include it, NZ$ prices, how Google handles your data, and a Sheets starter project. - [NotebookLM (Gemini Notebook)](https://levystreet.com/tools/notebooklm/index.md): NotebookLM is now Gemini Notebook. A guide for NZ businesses: plans and source limits, Workspace data handling, and a staff policy notebook to build. - [Claude Code](https://levystreet.com/tools/claude-code/index.md): Claude Code for New Zealand teams: what Anthropic's coding agent does well, plan prices as shown from NZ, how your code is handled, and a starter project. - [Cursor](https://levystreet.com/tools/cursor/index.md): Cursor AI for New Zealand teams after the SpaceX deal: the new USD plans, Privacy Mode and data use, buying it from NZ, and a starter project to try. - [n8n](https://levystreet.com/tools/n8n/index.md): n8n automation for New Zealand businesses: euro pricing per execution, self-hosting in an NZ region, licence rules, Xero setup and a starter AI workflow. - [Make](https://levystreet.com/tools/make/index.md): Make automation for New Zealand businesses: how credit billing works, plan prices, AI modules, data zones, and a ServiceM8 enquiry scenario to build. - [Zapier](https://levystreet.com/tools/zapier/index.md): Zapier AI for New Zealand businesses: USD plan prices, how AI steps and MCP calls use tasks, where data is stored, and a Xero reminder Zap to build this week. - [Copilot Studio](https://levystreet.com/tools/copilot-studio/index.md): Copilot Studio for NZ organisations: what Microsoft Copilot includes, Copilot Credit pricing with a worked example, where data sits and a first agent. - [Google AI Studio](https://levystreet.com/tools/google-ai-studio/index.md): Google AI Studio for New Zealand teams: what the free tier does with your prompts, Gemini API prices in USD, NZ data notes and a hands-on extraction project. ## Comparisons - [Compare](https://levystreet.com/compare/index.md): Side-by-side comparisons for NZ businesses: ChatGPT vs Claude, Copilot vs ChatGPT, n8n vs Make vs Zapier, custom AI vs off-the-shelf, and more. - [ChatGPT vs Claude](https://levystreet.com/compare/chatgpt-vs-claude/index.md): ChatGPT vs Claude for NZ businesses: Business and Team seat prices shown from NZ, data handling, Xero and ServiceM8 links, and three tests to run yourself. - [Copilot vs ChatGPT](https://levystreet.com/compare/copilot-vs-chatgpt/index.md): Copilot vs ChatGPT for NZ businesses: NZ$ prices, free Copilot Chat versus paid seats, grounding in Microsoft 365 data, data location and a same-task test. - [Gemini vs ChatGPT](https://levystreet.com/compare/gemini-vs-chatgpt/index.md): Gemini vs ChatGPT for NZ teams: what Google Workspace already includes, NZ$ seat prices, data terms, Sheets and Gmail tests, and when a second assistant pays. - [Claude vs Gemini](https://levystreet.com/compare/claude-vs-gemini/index.md): Claude vs Gemini for document work in NZ: long-context reading, Gemini Notebook source limits, prices shown from NZ, data terms and three document tests. - [n8n vs Make](https://levystreet.com/compare/n8n-vs-make/index.md): n8n vs Make for NZ businesses: executions versus credits, prices in October 2026, self-hosting in a NZ cloud region, AI agent features and a test plan. - [Zapier vs n8n](https://levystreet.com/compare/zapier-vs-n8n/index.md): Zapier vs n8n for NZ businesses: tasks versus executions, AI step billing, US$ and euro prices from October 2026, self-hosting and a same-workflow test. - [Make vs Zapier](https://levystreet.com/compare/make-vs-zapier/index.md): Make vs Zapier for NZ businesses: credits versus tasks, what each counts, prices in October 2026, what happens at the limit, and a Shopify cost example. - [Copilot Studio vs custom agents](https://levystreet.com/compare/copilot-studio-vs-custom-agents/index.md): Copilot Studio vs custom AI agents for NZ organisations: Copilot Credit rates, API token costs, processing region, a worked cost example and a test plan. - [AI agency vs internal hire](https://levystreet.com/compare/ai-agency-vs-internal-hire/index.md): AI agency vs hiring an AI engineer in NZ: a break-even rule, three-year cost arithmetic, a paid work-sample test, employment law points and failure modes. - [AI consultant vs developer](https://levystreet.com/compare/ai-consultant-vs-software-developer/index.md): AI consultant vs software developer in NZ: who answers which question, decision rules, when advice pays for itself, a same-brief test and failure modes. - [Custom AI vs off-the-shelf](https://levystreet.com/compare/custom-ai-vs-off-the-shelf/index.md): Build vs buy AI for NZ businesses: off-the-shelf, configured and custom options, payback arithmetic, a 30-case buy-first test and the costs that come later. - [AI chatbot vs voice agent](https://levystreet.com/compare/ai-chatbot-vs-voice-agent/index.md): AI chatbot vs voice agent for NZ businesses: per-outcome and per-minute pricing, +64 number costs, accent and place-name testing and a cost example. - [AI workshop vs course](https://levystreet.com/compare/ai-workshop-vs-self-paced-course/index.md): AI workshop vs self-paced course for NZ teams: what each suits, staff-time arithmetic, free NZ guidance, a split-group test and how to measure adoption. - [Automation vs AI agents](https://levystreet.com/compare/automation-vs-ai-agents/index.md): Automation vs AI agents for NZ businesses: decision rules, per-run cost arithmetic, platform pricing, NCSC NZ agent guidance and a 50-case test. - [RAG vs long context](https://levystreet.com/compare/rag-vs-long-context/index.md): RAG vs long context in 2026: when million-token models beat retrieval, per-query costs on Claude and GPT-6.1 Sol, caching, permissions and a fair test. ## Learn - [Learn](https://levystreet.com/learn/index.md): Free AI lessons for NZ business teams: short, tested exercises on prompting, checking AI answers, privacy, agents, automation and team rollout. - [AI courses in New Zealand](https://levystreet.com/learn/ai-courses-in-new-zealand/index.md): AI courses in NZ compared for October 2026: university qualifications, NZQA micro-credentials, free vendor training and practical lessons, by goal and cost. - [AI foundations for small business](https://levystreet.com/learn/ai-foundations-for-small-business/index.md): A free AI for small business lesson for NZ owners: what chat assistants do well, where they fail, and a 25-minute test on your own weekly tasks. - [Choosing your first AI workflow](https://levystreet.com/learn/choosing-your-first-ai-workflow/index.md): Pick your first AI business use case with a five-point scoring method, a worked example from a Christchurch builder and a 30-minute exercise. - [Writing useful business prompts](https://levystreet.com/learn/writing-useful-business-prompts/index.md): Write ChatGPT prompts for business that work first time: a five-part structure, a Dunedin electrician's worked example and a 20-minute exercise. - [Checking AI answers against sources](https://levystreet.com/learn/checking-ai-answers-against-sources/index.md): AI hallucinations explained for New Zealand teams, with a claim-by-claim checking method, a worked privacy breach example and a 25-minute exercise. - [Giving AI useful context](https://levystreet.com/learn/giving-ai-useful-context/index.md): What AI context means and how to build a one-page context pack so ChatGPT, Claude, Copilot or Gemini answers from your own facts, not guesses. - [Building a reusable prompt library](https://levystreet.com/learn/building-a-reusable-prompt-library/index.md): How to build a prompt library your team will use: entry fields, variables, test inputs, owners and review dates, with a worked real estate example. - [Protecting customer information](https://levystreet.com/learn/protecting-customer-information-when-using-ai/index.md): AI privacy for NZ businesses: what the Privacy Act 2020 and the Privacy Commissioner expect, a traffic-light data rule and a de-identification exercise. - [Creating a team AI use policy](https://levystreet.com/learn/creating-a-team-ai-use-policy/index.md): How to create an AI policy for a New Zealand team: eight sections, a drafting prompt, a Wellington consultancy example and a five-question scenario test. - [Calculating time saved by AI](https://levystreet.com/learn/calculating-time-saved-by-ai/index.md): Measure AI time savings honestly: time the task before and after, count review and rework, and turn net hours into dollars, with a worked NZ example. - [Evaluating an AI pilot](https://levystreet.com/learn/evaluating-an-ai-pilot/index.md): AI evaluation for business pilots: build a test set, write pass criteria before you run it, score critical errors and decide go, adjust or stop. - [Choosing business AI subscriptions](https://levystreet.com/learn/choosing-business-ai-subscriptions/index.md): Choose AI tools for business without overspending: compare Copilot, Gemini, ChatGPT and Claude plans in NZD, check data terms and run a same-task test. - [Training an internal AI champion](https://levystreet.com/learn/training-an-internal-ai-champion/index.md): What an AI champion does, how to pick the right person, and a six-week plan to train them, with a role card prompt and a Timaru freight example. - [Rolling out AI to a small team](https://levystreet.com/learn/rolling-out-ai-to-a-small-team/index.md): AI adoption for small NZ teams: a 30-day plan for the people side, from an honest kickoff to weekly practice, buddy pairs and a day-30 pulse check. - [Managing AI mistakes and escalation](https://levystreet.com/learn/managing-ai-mistakes-and-escalation/index.md): Practical AI risk management for NZ teams: grade AI mistakes by severity, follow a six-step escalation ladder, and know when a breach must be notified. - [Using AI for meeting preparation](https://levystreet.com/learn/using-ai-for-meeting-preparation/index.md): AI meeting preparation in 20 minutes: turn your notes, emails and numbers into a one-page brief with an agenda, decisions, questions and checked facts. - [Turning meeting notes into actions](https://levystreet.com/learn/turning-meeting-notes-into-actions/index.md): Use AI meeting notes properly: turn rough notes or a transcript into an action table with owners, dates and source quotes, and flag what is unclear. - [Drafting emails with human review](https://levystreet.com/learn/drafting-emails-with-human-review/index.md): AI email writing with a safety net: brief the draft, run a five-point review for facts, promises, tone, privacy and sign-off, then send it yourself. - [Researching a customer account](https://levystreet.com/learn/researching-a-customer-account/index.md): Learn AI sales research the careful way: build a one-page brief on a NZ prospect from public sources, with every claim linked and privacy rules respected. - [Creating a business proposal with AI](https://levystreet.com/learn/creating-a-business-proposal-with-ai/index.md): Use AI as a proposal generator without invented claims: draft a NZ business proposal from your own scope, prices and evidence, then check every promise. - [Analysing a spreadsheet with AI](https://levystreet.com/learn/analysing-a-spreadsheet-with-ai/index.md): AI spreadsheet analysis you can trust: profile the data first, ask specific questions, make the AI show its working, and check two numbers by hand. - [Summarising documents with citations](https://levystreet.com/learn/summarising-documents-with-citations/index.md): Get an AI document summary you can check: every point cites a page and an exact quote, then spot-check the citations. Practise on a real NZ guidance PDF. - [Building a searchable business knowledge base](https://levystreet.com/learn/building-a-searchable-business-knowledge-base/index.md): Build a small AI knowledge base from your own policies and procedures, test it with 10 known questions, and fix what breaks before NZ staff rely on it. - [Using AI to prepare management reports](https://levystreet.com/learn/using-ai-to-prepare-management-reports/index.md): Use AI for management reporting without invented explanations: draft monthly variance commentary from your P&L and managers' notes, then tie out every number. - [Using AI for customer support drafts](https://levystreet.com/learn/using-ai-for-customer-support-drafts/index.md): Use AI for customer support safely: draft replies from a one-page rules sheet, keep a person sending every message, and grade drafts on real NZ emails. - [Writing a workflow specification](https://levystreet.com/learn/writing-a-workflow-specification/index.md): AI workflow design starts on paper: use an AI interviewer to write a one-page workflow spec with inputs, steps, exceptions, human review and measures. - [Automating a simple approval workflow](https://levystreet.com/learn/automating-a-simple-approval-workflow/index.md): A practical workflow automation lesson: write a purchase approval rule as a decision table, test it on paper with AI-written test cases, then build it. - [Connecting AI to your CRM](https://levystreet.com/learn/connecting-ai-to-your-crm/index.md): A practical AI CRM integration lesson: connect an assistant to HubSpot, Pipedrive or Salesforce read-only, verify answers, then allow one supervised write. - [Understanding AI agents](https://levystreet.com/learn/understanding-ai-agents/index.md): Agentic AI explained for NZ business teams: what makes an AI agent different from a chatbot, and how to map an agent's actions by impact before trusting it. - [When a workflow does not need an agent](https://levystreet.com/learn/when-a-workflow-does-not-need-an-agent/index.md): AI agents vs automation: a decision rule for choosing between deleting a step, rules automation, one AI step or an agent, tested on tasks from your own week. - [Evaluating AI generated code](https://levystreet.com/learn/evaluating-ai-generated-code/index.md): An AI code review routine for non-specialists and developers: read, test and second-check AI-written scripts before running them on real business data. - [Building your first internal AI tool](https://levystreet.com/learn/building-your-first-internal-ai-tool/index.md): Before reaching for an AI app builder, build a saved assistant for one narrow job with instructions, reference files and a 10-case test, then share it safely. - [Testing a business AI assistant](https://levystreet.com/learn/testing-a-business-ai-assistant/index.md): A practical AI testing lesson for business teams: build a 20-case test set with expected answers and pass rules, run it three times, and score the results. - [Creating an AI budget](https://levystreet.com/learn/creating-an-ai-budget/index.md): Work out your real AI costs: a 12-month budget covering seats, usage, automation, people's time and contingency, with dated NZ prices and stated assumptions. - [Writing AI vendor requirements](https://levystreet.com/learn/writing-ai-vendor-requirements/index.md): AI procurement for NZ businesses: turn rough notes into numbered, testable vendor requirements covering accuracy, data handling, security, cost and exit. - [Planning a safe AI rollout](https://levystreet.com/learn/planning-a-safe-ai-rollout/index.md): Plan a safe AI rollout for a New Zealand team: a one-page plan with approved uses, data rules, pilot gates and stop conditions, drafted with AI in 30 minutes. - [Measuring adoption after training](https://levystreet.com/learn/measuring-adoption-after-training/index.md): Measure AI training ROI for an NZ team: a 30/60/90-day plan for usage, behaviour and timed outcomes, plus a worked example turning survey data into dollars. - [Using AI for market research](https://levystreet.com/learn/using-ai-for-market-research/index.md): AI market research for NZ businesses: size a niche from Stats NZ data, check competitors and tag every claim as found, calculated, assumed or unverified. - [Source-checked content drafts](https://levystreet.com/learn/creating-a-source-checked-content-draft/index.md): AI content marketing without made-up claims: build a source pack, make the AI cite it sentence by sentence, and cut what you cannot substantiate under NZ law. - [Customer FAQs from support tickets](https://levystreet.com/learn/building-customer-faqs-from-support-tickets/index.md): Use AI as an FAQ generator for your NZ business: cluster support tickets, draft answers only from policy, and catch answers that misstate consumer rights. - [Working with images in business AI](https://levystreet.com/learn/working-with-images-in-business-ai/index.md): Multimodal AI for NZ businesses: read photos, receipts, whiteboards and charts with AI, measure its error rate on your images, and keep people out of frame. - [Understanding RAG](https://levystreet.com/learn/understanding-retrieval-augmented-generation/index.md): Retrieval augmented generation explained for NZ teams: how RAG finds passages in your documents, why answers fail, and an exercise playing the retriever. - [API or subscription](https://levystreet.com/learn/choosing-between-an-api-and-a-subscription/index.md): AI API vs subscription for NZ businesses: compare per-seat plans with per-token API costs for one workflow, with worked arithmetic, data handling and upkeep. - [Understanding AI tool permissions](https://levystreet.com/learn/understanding-ai-tool-permissions/index.md): AI agent permissions explained for NZ teams: what read, write and send access allow, whose identity an AI acts as, and how to build a permission matrix. - [Building a no-code AI prototype](https://levystreet.com/learn/building-a-no-code-ai-prototype/index.md): No-code AI for NZ teams: build a reusable assistant for one task in ChatGPT, Claude, Gemini or Copilot, test it on 10 cases and make a go or no-go call. - [Documenting an AI workflow](https://levystreet.com/learn/documenting-an-ai-workflow-for-handover/index.md): AI documentation for NZ teams: write a handover runbook for an AI workflow with exact prompts, owners, review steps and test cases, using AI to interview you. - [Managing AI subscriptions and access](https://levystreet.com/learn/managing-ai-subscriptions-and-access/index.md): AI access management for NZ businesses: register who uses which AI tools, on which plans and connections, then clean up personal and leaver accounts. - [Recognising prompt injection](https://levystreet.com/learn/recognising-prompt-injection/index.md): Prompt injection explained for NZ teams: how hidden instructions in emails, web pages and files steer AI tools, with a safe exercise you can run today. - [Evaluating AI translation quality](https://levystreet.com/learn/evaluating-translation-quality-with-ai/index.md): AI translation for NZ businesses: check machine translations with an error checklist, back-translation and risk tiers, and when te reo Māori needs review. - [AI accessibility checks](https://levystreet.com/learn/using-ai-for-accessibility-checks/index.md): AI accessibility checks for NZ teams: review a web page against WCAG 2.2 AA with AI, map issues to success criteria, and know which tests still need people. - [Planning human review for AI output](https://levystreet.com/learn/planning-human-review-for-ai-output/index.md): Human review of AI output for NZ teams: decide which outputs people check, who checks them, how many and where, and stop review becoming a rubber stamp. - [Learning to code with AI](https://levystreet.com/learn/learning-to-code-with-ai/index.md): A free AI coding course lesson for NZ business people and junior developers: what to learn first, a guided build, code review, security and when to stop. ## Resources - [Resources](https://levystreet.com/resources/index.md): Free AI resources for NZ businesses: ROI and cost calculators, an AI readiness checklist, an AI use policy template, risk register and more. - [AI readiness checklist](https://levystreet.com/resources/ai-readiness-checklist/index.md): A free AI readiness checklist for New Zealand businesses: 30 yes-or-no checks across workflow, data, privacy, people, tools and budget, with scoring rules. - [AI business case template](https://levystreet.com/resources/ai-business-case-template/index.md): Free AI business case template for New Zealand businesses: ten sections, costs and benefits in NZD, a stop rule and a worked example with the arithmetic. - [AI ROI calculator](https://levystreet.com/resources/ai-roi-calculator/index.md): AI ROI calculator in NZ dollars: estimate hours saved, net monthly benefit, payback period and first-year return for an AI project, with a worked example. - [AI use policy template](https://levystreet.com/resources/ai-use-policy-template-for-nz-smbs/index.md): Free AI use policy template for New Zealand small businesses, written for the Privacy Act 2020 and its 14 privacy principles, including IPP 3A. Edit, adopt. - [AI vendor evaluation checklist](https://levystreet.com/resources/ai-vendor-evaluation-checklist/index.md): AI vendor evaluation checklist for NZ buyers: 32 checks in seven groups, knock-out rules and a weighted scoring sheet for comparing shortlisted AI tools. - [AI project brief template](https://levystreet.com/resources/ai-project-brief-template/index.md): Free AI project brief template: 14 numbered fields that tell a builder or vendor what an AI system must do, with a completed New Zealand example. - [AI workflow mapping template](https://levystreet.com/resources/ai-workflow-mapping-template/index.md): Free workflow mapping template for AI projects: record each step, owner, system, time and exception, then mark every step automate, assist, human or remove. - [AI risk register template](https://levystreet.com/resources/ai-risk-register-template/index.md): AI risk register template mapped to NIST AI RMF, the OWASP 2026 LLM Top 10 and ISO/IEC 42001, with scoring scales and 12 starter risks for NZ organisations. - [AI pilot scorecard](https://levystreet.com/resources/ai-pilot-scorecard/index.md): Free AI pilot scorecard: eight weighted criteria scored 0 to 4, hard gates and a decision rule to scale, fix or stop an AI trial, with a retail example. - [AI training needs assessment](https://levystreet.com/resources/ai-training-needs-assessment/index.md): Free AI training needs analysis template: six skill areas, a 0 to 3 proficiency scale, role targets and a gap score that ranks training priorities for teams. - [AI prompt library template](https://levystreet.com/resources/ai-prompt-library-template/index.md): Free AI prompt library template for teams: 12 fields per prompt, an index, five starter prompts for NZ businesses and rules that keep the library current. - [AI quality review checklist](https://levystreet.com/resources/ai-quality-review-checklist/index.md): Free AI quality review checklist: 24 checks in six groups, three review levels matched to risk, a sampling rule and a defect log for NZ businesses. - [AI incident log template](https://levystreet.com/resources/ai-incident-log-template/index.md): Free AI incident log template: 16 fields, a four-level severity scale, near misses and the New Zealand privacy breach notification test for AI incidents. - [AI data inventory template](https://levystreet.com/resources/ai-data-inventory-template/index.md): Free AI data inventory template for NZ organisations: 15 columns, a sensitivity scale and the Privacy Act 2020 questions, including IPP 3A and offshore use. - [AI adoption survey](https://levystreet.com/resources/ai-adoption-survey/index.md): Free AI adoption survey for staff: 18 questions with fixed response scales on use, time saved, safe use, confidence and barriers, plus scoring rules. - [AI agent acceptance checklist](https://levystreet.com/resources/ai-agent-acceptance-checklist/index.md): AI agent testing checklist: 36 acceptance checks in eight groups, mapped to the OWASP 2026 agentic Top 10 and NCSC guidance, with sign-off rules. - [AI cost calculator](https://levystreet.com/resources/ai-cost-calculator/index.md): AI cost calculator for NZ businesses: estimate LLM API running costs in NZD for 11 OpenAI, Anthropic and Google models, with caching, batch and exchange rate. - [AI roadmap template](https://levystreet.com/resources/ai-roadmap-template/index.md): AI roadmap template for New Zealand businesses: goals, a prioritisation score, three horizons over 12 months, gate measures and a quarterly review agenda. - [AI procurement questions](https://levystreet.com/resources/ai-procurement-questions/index.md): AI procurement checklist for New Zealand buyers: 30 questions to put to AI suppliers in writing, what good answers look like, contract clauses, red flags. - [AI learning plan for teams](https://levystreet.com/resources/ai-learning-plan-for-teams/index.md): Free 12-week AI learning plan for teams: four stages, about two hours per person a week, practice on real work, free lessons and an individual plan card. - [AI tools pricing NZ](https://levystreet.com/resources/ai-tools-pricing-nz/index.md): AI tools pricing for NZ businesses as listed on 19 October 2026: ChatGPT, Claude, Microsoft Copilot, Gemini, Copilot Studio, n8n, Make, Zapier and API prices. ## Engineering - [Engineering](https://levystreet.com/engineering/index.md): Practical AI engineering guides from Levy Street: reliable agents, evaluations, RAG, structured outputs, tool calling, MCP, observability and cost control. - [Designing reliable AI agents](https://levystreet.com/engineering/designing-reliable-ai-agents/index.md): AI agent architecture for production: when a workflow beats an agent, the control loop, tool contracts, state, stop rules and the failure modes to design out. - [AI agent evaluations](https://levystreet.com/engineering/ai-agent-evaluations/index.md): AI agent evaluation in practice: a task suite with pass/fail criteria, state checks before LLM judges, repeated trials, pass^k, cost and release gates. - [RAG retrieval evaluation](https://levystreet.com/engineering/rag-retrieval-evaluation/index.md): RAG evaluation done properly: build a labelled question set, measure recall@k, MRR and nDCG for retrieval, then faithfulness and correctness for answers. - [RAG versus fine-tuning](https://levystreet.com/engineering/rag-versus-fine-tuning/index.md): RAG vs fine-tuning for business AI: what each changes, a same-task test to decide, worked cost arithmetic with October 2026 prices, and provider limits. - [Choosing a vector database](https://levystreet.com/engineering/choosing-a-vector-database/index.md): Vector database comparison for business RAG: pgvector, Qdrant, Weaviate and managed services, the filtering trap, memory sizing and how to benchmark. - [Structured outputs and validation](https://levystreet.com/engineering/structured-outputs-and-validation/index.md): How to get reliable LLM structured output in production: provider schema limits, semantic validation, bounded repair loops and malformed-output tests. - [AI tool calling in production](https://levystreet.com/engineering/ai-tool-calling-in-production/index.md): AI tool calling in production: tool contracts, strict schemas, side-effect classes, idempotent retries, approval boundaries and the failure modes to test. - [MCP for business systems](https://levystreet.com/engineering/mcp-for-business-systems/index.md): How to build or adopt an MCP server for business systems under the 2026-07-28 spec: stateless design, OAuth scopes, confirmations, state handles and audit. - [Agent memory design](https://levystreet.com/engineering/agent-memory-design/index.md): How to design AI agent memory: memory types, a write policy with provenance, Privacy Act retention and deletion, poisoning controls and recall tests. - [Agent permissions and approvals](https://levystreet.com/engineering/agent-permissions-and-approvals/index.md): AI agent security through least privilege: identity models, action tiers and approval rules, approval screens that resist rubber-stamping, and abuse tests. - [Prompt injection testing](https://levystreet.com/engineering/prompt-injection-testing/index.md): Prompt injection testing for AI apps and agents: map every input channel, build a test catalogue, plant canaries, measure attack success and rank defences. - [AI audit logs](https://levystreet.com/engineering/ai-audit-logs/index.md): How to build an AI audit trail: trace structure, OpenTelemetry GenAI fields, what to record or hash, tamper evidence, NZ privacy rules and incident replay. - [Monitoring AI quality](https://levystreet.com/engineering/monitoring-ai-quality/index.md): AI observability beyond uptime: a quality drift dashboard with metric definitions, sampled grading, judge calibration and alerts that respect sample size. - [Reducing AI latency](https://levystreet.com/engineering/reducing-ai-latency/index.md): LLM latency explained and reduced: TTFT versus output speed, a latency budget, the optimisations that work, fast tiers, NZ network routes and how to measure. - [Managing LLM costs](https://levystreet.com/engineering/managing-llm-costs/index.md): LLM cost optimisation with real numbers: cost per successful task, token accounting, caching break-even, batch and flex discounts, routing and budget guards. - [Human review queues](https://levystreet.com/engineering/human-review-queues/index.md): How to design human in the loop AI review queues: routing by risk and calibrated confidence, audit sampling, reviewer agreement and the metrics that matter. - [Agent retry and failure recovery](https://levystreet.com/engineering/agent-retry-and-failure-recovery/index.md): AI agent reliability in practice: classify failures, retry with backoff and jitter, make tool calls idempotent, checkpoint runs and fall back safely. - [Evaluating long context models](https://levystreet.com/engineering/evaluating-long-context-models/index.md): How to evaluate a long context LLM on your own documents: depth and length grids, multi-hop and absence tests, distractor stress tests and effective context. - [Multi-agent systems in practice](https://levystreet.com/engineering/multi-agent-systems-in-practice/index.md): When multi-agent systems beat a single agent, how to run a fair controlled comparison, what to measure, and the coordination failures to design against. - [Building private AI search](https://levystreet.com/engineering/building-private-ai-search/index.md): How to build private AI search over company documents: permission-aware retrieval, ACL sync from source systems and a leakage test suite that proves it. - [Running open models for business](https://levystreet.com/engineering/running-open-models-for-business/index.md): When a local LLM makes sense for a business, how to size GPU memory from weights and KV cache, choose a serving stack, test capability and compare costs. - [AI data preparation](https://levystreet.com/engineering/ai-data-preparation/index.md): Data preparation for AI systems: an ingestion pipeline with quality gates for extraction, normalisation, deduplication, versioning, privacy and chunking. - [OCR versus vision models](https://levystreet.com/engineering/ocr-versus-vision-models/index.md): OCR vs AI vision models for document extraction: how they fail differently, how to benchmark field accuracy, CER and cost per page, and when to combine them. - [Voice agent quality evaluation](https://levystreet.com/engineering/voice-agent-quality-evaluation/index.md): Voice AI evaluation for phone and voice agents: word error rate, entity accuracy, turn latency budgets, scenario suites and escalation precision and recall. - [AI model migration](https://levystreet.com/engineering/ai-model-migration/index.md): How to run an LLM migration safely: inventory call sites, freeze a regression suite, gate on quality, cost and latency, then shadow, canary and roll back. - [Versioning prompts and evaluations](https://levystreet.com/engineering/versioning-prompts-and-evaluations/index.md): Prompt versioning for production AI: what to version, config hashes on every call, semantic versions for prompts, immutable eval sets and fast rollback. - [AI incident response](https://levystreet.com/engineering/ai-incident-response/index.md): AI incident response for production systems: severity levels, kill switches, evidence to preserve, scoping queries, privacy breach steps and postmortems. - [Testing AI with synthetic data](https://levystreet.com/engineering/testing-ai-with-synthetic-data/index.md): Using synthetic data to test AI systems: where generated test cases help, why they overstate accuracy, and how to measure the gap against a real holdout set. - [AI product feedback loops](https://levystreet.com/engineering/designing-ai-product-feedback-loops/index.md): How to design an AI feedback loop: which user signals to capture, a fixed feedback taxonomy, measuring feedback quality and closing each loop with a test. - [Data residency for AI systems](https://levystreet.com/engineering/data-residency-for-ai-systems/index.md): AI data residency in New Zealand: what the Privacy Act and IPP12 require, NZ and Australian cloud regions, and which AI providers process data in-region. ## Integrations - [Integrations](https://levystreet.com/integrations/index.md): How to connect AI to Xero, MYOB, Microsoft 365, Google Workspace, HubSpot, Shopify, ServiceM8 and more, with permissions and approvals done properly. - [AI and Xero](https://levystreet.com/integrations/xero/index.md): How to connect AI to Xero in 2026: JAX, the Accounting API and remote MCP server, bill extraction with approval, granular scopes, developer fees and limits. - [AI and Microsoft 365](https://levystreet.com/integrations/microsoft-365/index.md): How to connect AI to Microsoft 365 in 2026: Copilot licences, Microsoft Graph, Work IQ MCP and permission-aware SharePoint search, with NZ costs and limits. - [AI and Google Workspace](https://levystreet.com/integrations/google-workspace/index.md): How to connect AI to Google Workspace in 2026: Gemini, the Gmail API and Workspace MCP, shared inbox triage and reply drafts staff approve, with NZ costs. - [AI and HubSpot](https://levystreet.com/integrations/hubspot/index.md): How to connect AI to HubSpot in 2026: Agent Hub, credits, the MCP server and API, and account research enrichment with source URLs, under NZ privacy law. - [AI and Salesforce](https://levystreet.com/integrations/salesforce/index.md): How to connect AI to Salesforce in 2026: Agentforce pricing, hosted MCP servers, API limits and an opportunity briefing workflow that respects sharing rules. - [AI and Pipedrive](https://levystreet.com/integrations/pipedrive/index.md): How to connect AI to Pipedrive in 2026: built-in AI, the official MCP server, OAuth scopes, token-based API limits and a lead follow-up draft workflow. - [AI and Shopify](https://levystreet.com/integrations/shopify/index.md): How to connect AI to Shopify in 2026: Sidekick, the Inbox agent, the GraphQL Admin API and UCP, and product support that answers from live catalogue data. - [AI and WooCommerce](https://levystreet.com/integrations/woocommerce/index.md): How to connect AI to WooCommerce in 2026: REST API keys, the MCP Adapter preview and a safe order enquiry assistant that verifies customers before it answers. - [AI and Zendesk](https://levystreet.com/integrations/zendesk/index.md): How to connect AI to Zendesk in 2026: AI agents, Copilot, resolution billing tiers, OAuth scopes and a ticket workflow with draft replies and escalation. - [AI and Intercom](https://levystreet.com/integrations/intercom/index.md): How to connect AI to Intercom in 2026: Fin pricing per outcome, escalation rules, the MCP server and AU hosting limits, and daily answer quality checks. - [AI and Slack](https://levystreet.com/integrations/slack/index.md): How to connect AI to Slack in 2026: Slack AI by plan, the Slack MCP server and Real-time Search API, and a staff knowledge assistant with channel controls. - [AI and Microsoft Teams](https://levystreet.com/integrations/microsoft-teams/index.md): How to connect AI to Microsoft Teams in 2026: Copilot, Copilot Studio credits, the Agents SDK, team-level permissions and an assistant on approved data. - [AI and Notion](https://levystreet.com/integrations/notion/index.md): How an AI Notion integration works in 2026: Notion AI plans, the API, webhooks and hosted MCP, with knowledge refresh and search designs for NZ teams. - [AI and SharePoint](https://levystreet.com/integrations/sharepoint/index.md): An AI SharePoint integration guide for NZ: Copilot licensing, Microsoft Graph, Sites.Selected, the Work IQ MCP server and permission-aware search design. - [AI and Google Drive](https://levystreet.com/integrations/google-drive/index.md): An AI Google Drive integration guide for NZ: Gemini in Drive, Drive API scopes and quotas, the Drive MCP preview, and source-linked document Q&A design. - [AI and Airtable](https://levystreet.com/integrations/airtable/index.md): An AI Airtable integration guide for NZ: Omni and AI credits, the Web API, webhooks and MCP server, plus record enrichment with validation and approval. - [AI and monday.com](https://levystreet.com/integrations/monday/index.md): An AI monday.com integration guide for NZ: AI credits, the GraphQL API, webhooks and Platform MCP, and a project status summary that cites its sources. - [AI and Asana](https://levystreet.com/integrations/asana/index.md): An AI Asana integration guide for NZ teams: AI Teammates and AI request pricing, the API, webhooks and MCP server, and turning meeting notes into tasks. - [AI and Jira](https://levystreet.com/integrations/jira/index.md): An AI Jira integration guide for NZ: Rovo credits and overage billing from 3 December 2026, the Rovo MCP server, REST API, webhooks and triage designs. - [AI and ServiceM8](https://levystreet.com/integrations/servicem8/index.md): An AI ServiceM8 integration guide for NZ trades: AI Smart Helpers, the API, webhooks and official MCP server, and a job-notes-to-quote draft with approval. - [AI and Simpro](https://levystreet.com/integrations/simpro/index.md): An AI Simpro integration guide for NZ: Simpro Lightning, JustAsk and its AI agents, the OAuth2 REST API, and service request triage with dispatcher approval. - [AI and Cin7](https://levystreet.com/integrations/cin7/index.md): An AI Cin7 integration guide for NZ: ForesightAI, AMA and document recognition, the Core and Omni APIs and webhooks, plus inventory exception summaries. - [AI and MYOB](https://levystreet.com/integrations/myob/index.md): An AI MYOB integration guide for NZ: Iris by MYOB, the read-only MYOB Connector, the Business API, and a supplier invoice review workflow with approval. - [AI and Power BI](https://levystreet.com/integrations/power-bi/index.md): An AI Power BI integration guide for NZ: Copilot capacity rules, executeQueries, Power BI MCP servers and Fabric IQ, and narratives with reconciled metrics. ## Optional - [Full text of every page](https://levystreet.com/llms-full.txt) - [Sitemap](https://levystreet.com/sitemap.xml) - [Industry workflow guides](https://levystreet.com/industries/)