https://cloudsail.com/blog/Dated guides with sources on AI investment decisions, coding agent rollout, process agents, retrieval, model and inference costs, and platform integration.enTue, 10 Nov 2026 08:00:00 +0000 https://cloudsail.com/blog/claude-code-company-rollout/https://cloudsail.com/blog/claude-code-company-rollout/Tue, 10 Nov 2026 08:00:00 +0000How to introduce Claude Code in a company: accounts and provider, managed settings, permissions, repository, a pilot with one team, training and works council. https://cloudsail.com/blog/works-council-ai-system-logging/https://cloudsail.com/blog/works-council-ai-system-logging/Sat, 07 Nov 2026 08:00:00 +0000What AI assistants, LLM gateways and coding agents log, which fields let an employer monitor staff, and how to configure them for a German works council. https://cloudsail.com/blog/shadow-ai-approved-platform/https://cloudsail.com/blog/shadow-ai-approved-platform/Sat, 07 Nov 2026 08:00:00 +0000Ban shadow AI and it moves to private phones. What an approved AI platform on your own servers or cloud account looks like, and how it enforces your AI policy. https://cloudsail.com/blog/on-premise-llm-for-companies/https://cloudsail.com/blog/on-premise-llm-for-companies/Sat, 07 Nov 2026 08:00:00 +0000How companies run an on-premise LLM so data never leaves the building: when it beats a private cloud, the layers from GPUs to agents, German model tests and sizing. https://cloudsail.com/blog/llm-gateway-explained/https://cloudsail.com/blog/llm-gateway-explained/Sat, 07 Nov 2026 08:00:00 +0000What an LLM gateway does, what it does not, when a company needs one, and how LiteLLM, Kong, Azure, Cloudflare, Portkey, OpenRouter and Agent Router compare. https://cloudsail.com/blog/litellm-production-security/https://cloudsail.com/blog/litellm-production-security/Sat, 07 Nov 2026 08:00:00 +0000After the 2026 LiteLLM incidents: how to check exposure, when you need Redis and Postgres, how to pin and upgrade safely, and what to tick off before go-live. https://cloudsail.com/blog/introducing-ai-in-your-company/https://cloudsail.com/blog/introducing-ai-in-your-company/Sat, 07 Nov 2026 08:00:00 +0000Start AI with one process that costs real hours. Pilot it against criteria agreed in advance, then go live on your own systems with an owner and IT on board early. https://cloudsail.com/blog/eu-data-residency-azure-openai-bedrock/https://cloudsail.com/blog/eu-data-residency-azure-openai-bedrock/Sat, 07 Nov 2026 08:00:00 +0000Which Azure OpenAI deployment types and Bedrock inference profiles keep prompts in the EU, who can read them, where an EU region stops helping, and how to enforce it. https://cloudsail.com/blog/copilot-or-own-ai-assistant/https://cloudsail.com/blog/copilot-or-own-ai-assistant/Sat, 07 Nov 2026 08:00:00 +0000When Microsoft 365 Copilot is enough, when your own AI assistant pays off for ERP, DMS and in-house data, and why you fix permissions first either way. https://cloudsail.com/blog/coding-agents-llm-gateway/https://cloudsail.com/blog/coding-agents-llm-gateway/Sat, 07 Nov 2026 08:00:00 +0000Claude Code can be pinned to your own LLM gateway, Copilot BYOK and Cursor only partly. Settings, what to log, team budgets and the usual failures, as of November 2026. https://cloudsail.com/blog/chatgpt-data-protection-options/https://cloudsail.com/blog/chatgpt-data-protection-options/Sat, 07 Nov 2026 08:00:00 +0000ChatGPT at work and data protection: private accounts, business tiers, own cloud account or own servers. What leaves the company at each step and what to check. https://cloudsail.com/blog/ai-readiness-check/https://cloudsail.com/blog/ai-readiness-check/Sat, 07 Nov 2026 08:00:00 +0000Is your company ready for AI? Follow one real case through your systems and check skills, data, systems, rules and people line by line. With a checklist. https://cloudsail.com/blog/ai-change-management/https://cloudsail.com/blog/ai-change-management/Sat, 07 Nov 2026 08:00:00 +0000Change management for AI: an honest answer on jobs, staff fears, the works council, shadow mode and staged rollout, shared results and training on real work. https://cloudsail.com/blog/ai-business-case/https://cloudsail.com/blog/ai-business-case/Sat, 07 Nov 2026 08:00:00 +0000Whether AI pays off depends on one process and your own figures. Measure today's cost per case, count build, run and review costs, then compare. With a worksheet. https://cloudsail.com/blog/ai-agency-or-own-team/https://cloudsail.com/blog/ai-agency-or-own-team/Sat, 07 Nov 2026 08:00:00 +0000AI agency, consultancy or in-house team: when each fits, what to ask a partner, which evidence to request, and how a build with handover moves AI to your team. https://cloudsail.com/blog/ai-act-deployer-checklist/https://cloudsail.com/blog/ai-act-deployer-checklist/Sat, 07 Nov 2026 08:00:00 +0000EU AI Act duties for companies that use AI: dates after the Digital Omnibus, AI literacy, transparency, high-risk uses and a technical deployer checklist. https://cloudsail.com/blog/llm-regression-testing/https://cloudsail.com/blog/llm-regression-testing/Wed, 04 Nov 2026 08:00:00 +0000How to regression test an LLM application when prompts, models or documents change: what goes in the suite, how to compare versions and when to stop a release. https://cloudsail.com/blog/how-to-test-ai-systems/https://cloudsail.com/blog/how-to-test-ai-systems/Wed, 04 Nov 2026 08:00:00 +0000How to test LLM applications, RAG assistants and AI agents: which checks are deterministic, which are statistical, and where the test cases come from. https://cloudsail.com/blog/ai-acceptance-criteria/https://cloudsail.com/blog/ai-acceptance-criteria/Wed, 04 Nov 2026 08:00:00 +0000Acceptance criteria for LLM applications, chatbots and agents: failure types, severities, thresholds set in advance and a pass rate calculator. https://cloudsail.com/blog/llm-judge-validation/https://cloudsail.com/blog/llm-judge-validation/Sun, 01 Nov 2026 08:00:00 +0000Before you report LLM judge scores, test the judge on hand-labelled examples, measure true positive and true negative rates and correct the pass rate. https://cloudsail.com/blog/llm-eval-sample-size/https://cloudsail.com/blog/llm-eval-sample-size/Sun, 01 Nov 2026 08:00:00 +0000Why 80% against 83% on 100 cases proves little, how to compare prompts on the same cases and how many cases a real difference needs, with worked numbers. https://cloudsail.com/blog/llm-error-analysis/https://cloudsail.com/blog/llm-error-analysis/Sun, 01 Nov 2026 08:00:00 +0000Before writing evals, read about 100 traces with a domain expert, note the first failure in each, group the notes into failure types and count them. https://cloudsail.com/blog/coding-agents-compared-for-teams/https://cloudsail.com/blog/coding-agents-compared-for-teams/Fri, 30 Oct 2026 08:00:00 +0000Claude Code, Cursor, GitHub Copilot and OpenAI Codex compared for engineering teams: team plans, usage billing, models, data handling and how to choose. https://cloudsail.com/blog/coding-agent-rollout-cost/https://cloudsail.com/blog/coding-agent-rollout-cost/Fri, 30 Oct 2026 08:00:00 +0000Seats, usage, review time and the costs teams forget: what Claude Code, Cursor, GitHub Copilot or Codex cost for 50 engineers, with a worked example. https://cloudsail.com/blog/coding-agent-cost-calculator/https://cloudsail.com/blog/coding-agent-cost-calculator/Fri, 30 Oct 2026 08:00:00 +0000Work out what Claude Code, Cursor, GitHub Copilot or Codex costs your team a month and per merged change, review time included, at October 2026 prices. https://cloudsail.com/blog/choosing-an-ai-consultancy/https://cloudsail.com/blog/choosing-an-ai-consultancy/Fri, 30 Oct 2026 08:00:00 +0000What to ask a coding agent or AI consultancy before hiring it, and the evidence to request for each answer, from the baseline to the first step. https://cloudsail.com/blog/small-model-task-selection/https://cloudsail.com/blog/small-model-task-selection/Thu, 22 Oct 2026 08:00:00 +0000How to pick high-volume tasks a small language model can handle reliably: a screening checklist, a test before training and a cost comparison at your volume. https://cloudsail.com/blog/self-hosted-ai-deployment-choice/https://cloudsail.com/blog/self-hosted-ai-deployment-choice/Thu, 22 Oct 2026 08:00:00 +0000Should you self-host an AI workload or keep a managed model endpoint? Test access requirements, operating capacity and measured workload costs with a worksheet. https://cloudsail.com/blog/repository-readiness-for-coding-agents/https://cloudsail.com/blog/repository-readiness-for-coding-agents/Thu, 22 Oct 2026 08:00:00 +0000What a repository must provide before coding agents work reliably: a clean-checkout setup, tested commands, short instructions and a CI job that proves them. https://cloudsail.com/blog/rag-vs-fine-tuning-decision/https://cloudsail.com/blog/rag-vs-fine-tuning-decision/Thu, 22 Oct 2026 08:00:00 +0000Use retrieval when the model needs facts that change and fine-tuning when it needs consistent behaviour on a narrow task. Criteria, examples and an order of work. https://cloudsail.com/blog/rag-stale-documents/https://cloudsail.com/blog/rag-stale-documents/Thu, 22 Oct 2026 08:00:00 +0000Deleted and edited documents often keep answering in RAG systems. Design sync for updates, deletions and renames, then prove freshness with canary documents. https://cloudsail.com/blog/rag-spreadsheets-and-tables/https://cloudsail.com/blog/rag-spreadsheets-and-tables/Thu, 22 Oct 2026 08:00:00 +0000Look-ups can be retrieved as text; totals, filters and comparisons need a query. Route table questions correctly and make every computed answer traceable. https://cloudsail.com/blog/rag-source-readiness/https://cloudsail.com/blog/rag-source-readiness/Thu, 22 Oct 2026 08:00:00 +0000Decide which documents a RAG assistant should use: build a source inventory with owners and versions, trace real questions to answers and find gaps first. https://cloudsail.com/blog/rag-retrieval-evaluation/https://cloudsail.com/blog/rag-retrieval-evaluation/Thu, 22 Oct 2026 08:00:00 +0000Trace each wrong RAG answer to its first failure: missing from the index, ranked too low, cut before the prompt, or misused by the model. A step-by-step diagnosis. https://cloudsail.com/blog/rag-production-handover/https://cloudsail.com/blog/rag-production-handover/Thu, 22 Oct 2026 08:00:00 +0000A RAG handover checklist: connectors and sync jobs, versioned configuration, evaluation sets and baselines, runbooks, named owners and a rehearsal before sign-off. https://cloudsail.com/blog/rag-multilingual-search/https://cloudsail.com/blog/rag-multilingual-search/Thu, 22 Oct 2026 08:00:00 +0000Will your assistant find the same evidence in every language? Build a parallel test set, compare query and document languages, and fix terms and analyzers. https://cloudsail.com/blog/rag-document-permissions/https://cloudsail.com/blog/rag-document-permissions/Thu, 22 Oct 2026 08:00:00 +0000Stop a RAG assistant from showing people documents they cannot open: carry access lists into the index, filter by user identity and test with denied users. https://cloudsail.com/blog/rag-citation-quality/https://cloudsail.com/blog/rag-citation-quality/Thu, 22 Oct 2026 08:00:00 +0000A citation that points to a real passage can still fail to support the claim. Check answers claim by claim against cited evidence and set acceptance criteria. https://cloudsail.com/blog/rag-chunking-for-business-documents/https://cloudsail.com/blog/rag-chunking-for-business-documents/Thu, 22 Oct 2026 08:00:00 +0000Split business documents for RAG without cutting answers off from their conditions: follow document structure, carry context into chunks and test the options. https://cloudsail.com/blog/quantization-quality-evaluation/https://cloudsail.com/blog/quantization-quality-evaluation/Thu, 22 Oct 2026 08:00:00 +0000Before swapping in a quantised model, rerun task, format, long-input and behaviour tests, measure serving at your latency target and compare cost per GPU.