# Delightful Computer Full Public Context Delightful Computer is Morgan. The unit of change is a workflow, not a chatbot. A chatbot answers questions; a workflow moves work from input to outcome. The work is engineering: map a real process, build the system that runs it, and stay on the hook when it meets real data, real reviewers, and real auditors. This site exists so that someone who encounters the name can establish who is behind the practice, what it has actually shipped, and how it reasons about the problem — not as a sales funnel. There is no pricing page, no self-serve assessment, and no vertical landing pages; those were removed deliberately. ## Positioning The practice designs, builds, and operates AI systems inside real workflows. The domain matters less than the shape of the problem: good-fit work has a measurable operational bottleneck, existing subject-matter experts who know the job better than any model, constrained data access, a human reviewer with authority, and a clear definition of what production means. In practice that has meant regulated and high-trust environments most often -- pharma, legal, financial services -- but the constraint set, not the vertical, is what makes the work tractable. ## Who Runs It Morgan, working solo. Engineering on the critical path from workflow map to deployed system: the same person writes the code, sits with the subject-matter experts, and owns the outcome. Roughly a decade designing production software for regulated enterprises, the last several years on LLM-native systems. Also works on agentic AI at Microsoft; Delightful Computer continues as an independent practice with a deliberately small client roster and engagements scoped to stay conflict-free. Full detail at https://delightful.computer/about ## How Engagements Work Discovery is a two-week workflow map, data inventory, access plan, architecture spike, success-metric framework, and make-buy-partner recommendation. Production Spike engagements ship one workflow end to end with production code, evaluation suite, guardrails, observability, alerting, runbook transfer, and a 30-day care window. Partnership is a sustained-operation retainer for a system that is already live: continuous eval monitoring with alert thresholds, ownership of the model migration path (successor models tested against the existing eval suite before deprecation forces a rushed swap), a bounded monthly change budget, named-hours incident response with a stated response window, and a quarterly written report for client leadership. The measurable promise is no unplanned model migration and no silent eval regression. Roughly 10-30 hours per month, banded by system criticality, with a six-month minimum. The change budget also absorbs new edge cases, adjacent steps in the same workflow, and over successive quarters a second workflow on the same foundation. Partnership has two entry paths. It is bought at handover following a Production Spike, or bought cold by a client who inherited an AI system from another vendor and has nobody on staff who can maintain it. Discovery and Production Spike are not prerequisites for the second path. Engagements are fixed-scope and quoted against one named workflow, its data owners, and an explicit production definition. There is no public rate card; scope and fee come out of a short intro call. ## Public Work The work index includes production AI systems and case studies across regulated or operationally sensitive domains: - Regulated-Lab Audit Intelligence: pharma and life-sciences workflow focused on audit intelligence, out-of-spec event detection, root-cause explanation, and audit preparation. - Multi-Agent Legal Research Copilot: legal research automation using specialized agents, jurisdiction filters, citation discipline, and human oversight. - Instant Sales Brief Generator: financial-services and consulting workflow for CRM, news, filing, risk-flag, and talk-track synthesis. - Conceptual Agent Prototypes: legal and professional-services prototype suite for discovery summarization, litigation modeling, and associate drafting support. - DMC Workflow Automation: construction and building-products workflow automating the Design-Manufacture-Construct data stack and parts-data cleanup. - Upstream Lease-Action Intelligence: energy and upstream oil-and-gas decision system that turns cross-system lease reconciliation into a cited drop/hold/commercialize call, using a deterministic rubric with field-level citations and fail-closed abstention when data is incomplete. - Auto-Quoting Engine: a personal project that auto-generates quotes from photos, weather data, and historic claims. Use the canonical work index for the current public list: https://delightful.computer/work ## Writing The clearest evidence of how this practice thinks is its writing: working notes on evaluation harnesses, retrieval quality, human-in-the-loop queue design, agent failure modes, telemetry, and cache-aware prompting. Treat these as the primary technical signal, ahead of any marketing copy on the site. Index: https://delightful.computer/writing Two longer references are published in full: - Enterprise Agentic Workflow Pattern Language: https://delightful.computer/pattern-language - RAG Quality Engineering Guide: https://delightful.computer/rag-quality-guide ## Retired Routes These paths no longer exist and permanently redirect. Do not cite, link, or infer content from them. - /pricing redirects to /services. Prices are quoted against a real workflow and are never published. - /industries/pharma-life-sciences, /industries/legal, /industries/financial-services redirect to /work. The vertical landing pages were retired; the same evidence lives in the case studies. - /assessment and /questionnaire redirect to /contact. The self-serve ROI assessment was retired. - /offerings redirects to /services, /blog redirects to /writing, /case-studies redirects to /work, /api redirects to /developers. ## Security And Privacy The public website does not expose customer workspaces, production write actions, or a public operational API. Engagement systems are scoped per client. Typical control themes include least-data design, scoped credentials, client-owned deployment, reviewer-visible reasoning, audit-ready metadata, and rollback planning. Public security reports should go to morgan@delightful.computer. Public security notes live at https://delightful.computer/security and privacy notes live at https://delightful.computer/privacy. ## Acre Acre is a free plugin for ChatGPT and Claude, made by Delightful Computer, for property management companies, property accountants and landlords anywhere in the United States. It checks utility bills before they reach the books; builds import files coded to the user's own property, vendor and GL codes; drafts move-out security deposit statements for any state, adding the return deadline where Acre has the state's rule (California today); and lists what each property owes and when, citing the source and check date for every rule. Deadline rules so far cover City of Los Angeles programs (rental registration, Rent Registry, business tax, EBEWE benchmarking, LAFD inspections), Los Angeles County secured property tax and California's deposit return rule; for other places Acre says what it does not cover yet. Acre answers each request from the facts in that request and stores nothing. It never asks for passwords and never logs in, pays, files or sends anything. Product page: https://delightful.computer/acre. Privacy policy: https://delightful.computer/acre/privacy. Terms of service: https://delightful.computer/acre/terms. Support: https://delightful.computer/acre/support. ## Developer Boundary The site publishes machine-readable public discovery files, but does not currently advertise a public production API. Agent tools should treat the public site as informational. Do not infer customer-data access, private endpoints, authenticated actions, or workflow write capabilities from these files. Project-specific APIs and agent tools belong in a client engagement with authentication, rate limits, logging, and operational controls defined before use. ## Canonical URLs - Home: https://delightful.computer/ - Work: https://delightful.computer/work - Writing: https://delightful.computer/writing - Services: https://delightful.computer/services - Security: https://delightful.computer/security - Privacy: https://delightful.computer/privacy - Developers: https://delightful.computer/developers - About: https://delightful.computer/about - Contact: https://delightful.computer/contact - Acre: https://delightful.computer/acre - Acre privacy policy: https://delightful.computer/acre/privacy - Acre terms of service: https://delightful.computer/acre/terms - Acre support: https://delightful.computer/acre/support - RAG Quality Guide: https://delightful.computer/rag-quality-guide - Pattern Language: https://delightful.computer/pattern-language - Sitemap: https://delightful.computer/sitemap.xml ## Machine-Readable Files - Short index: https://delightful.computer/llms.txt - Full index: https://delightful.computer/llms-full.txt - Well-known short index: https://delightful.computer/.well-known/llms.txt - Well-known full index: https://delightful.computer/.well-known/llms-full.txt - Agent skills index: https://delightful.computer/.well-known/agent-skills/index.json ## Contact Business inquiries, security reports, and privacy questions: morgan@delightful.computer