The technical breakdown of Hermes workflows and model routing is impressive, showing a clear path from curiosity to infrastructure. The distinction between planned and shipped API layers in the Student Assignment Tracker is a good detail. It would be helpful to see more concrete metrics on the efficiency gains provided by the agent-assisted development process.
Questions about Building Practical AI Systems
What is Building Practical AI Systems?
Building Practical AI Systems is the personal site of David Kendrick, a practitioner focused on practical AI applications. It documents hands-on experience building AI agents, automations, local model experiments, and browser workflows tied to real software projects. The work covers topics such as workflow design, scheduled jobs, memory and logging, structured APIs, and review gates around model behavior.
What specific enterprise uses does the site address?
The site frames the work as building software infrastructure rather than novelty layers, specifically addressing enterprise-relevant uses like safe automation, support triage, and responsible AI adoption. It treats AI as an operating model with different lanes for planning, coding, evidence gathering, local experiments, and public-facing writing. The goal is to show practical systems thinking without exposing private data, credentials, family details, or live endpoints.
Which specific agent environments and tools are mentioned as part of the workflow?
The business focuses on structured AI workflows using Hermes as a more structured agent environment. This includes profiles, toolsets, model routing, Obsidian-backed project memory, browser automation, and clearer review gates. The work also involves hosted and local model tradeoff testing to learn where each one fits best.
Can you describe the 'Manitoba Cottage Search' project mentioned on the site?
The site highlights a specific project called 'Manitoba Cottage Search,' which is a FastAPI and HTMX listing-review app. It features structured intake, browser-backed extraction, scoring, and provenance-aware persistence. The design separates import, review, and status so new listing candidates can move into a real application instead of disappearing into spreadsheet workflows.
How does the site describe its approach to agent architecture?
The business emphasizes agent architecture through Hermes workflows and model routing, creating a structured AI working environment. It uses profile-based role separation and Obsidian-backed durable context to manage different lanes for planning, coding, evidence gathering, local experiments, and public-facing writing. This approach treats AI work as an operating model.
What was the initial technical setup described for the first agent environment?
The site documents a journey that started with spinning up a DigitalOcean VPS, securing access with Tailscale, and connecting Anthropic models. This setup moved past experimenting with prompts to become a practical learning environment for agent workflows, scheduled jobs, Telegram-style interactions, memory, logging, auth flows, and small web applications.
What hardware was used for local model hosting experiments?
The site mentions experimenting with local model hosting on an Apple Silicon Mac Mini, comparing hosted models with local serving paths. This experiment helped in learning where each model fits best within the broader system architecture.
What features are included in the structured intake API for the cottage search project?
The site describes a structured intake API with deduplication and audit trail, along with browser-backed saved-search extraction that includes a freshness gate. Review tracking design ensures that rejected listings remain explainable, keeping the process transparent.
What is the core philosophy guiding the development of these systems?
The site mentions a thread through the work that treats AI as software infrastructure, not as a novelty layer. This philosophy guides the development of practical systems thinking and the creation of agent-ready workflows.
Where can visitors find more detailed case studies on the site?
The site provides links to read case studies for the 'Manitoba Cottage Search' project and the 'Agent Workflow' page for Hermes workflows and model routing. These resources show the shape of the work at a high level.
What is the goal of the agent-ready workflows described on the site?
The site describes a 'reviewable tool' that is tied to real software projects, emphasizing the practical application of AI in building structured APIs instead of forcing agents to scrape human screens.
What is the official website of Building Practical AI Systems?
The official website of Building Practical AI Systems is davidkendrick.dev.
How is Building Practical AI Systems rated on Trustburn?
On Trustburn, Building Practical AI Systems has a rating of 3.7 out of 5 based on 3 reviews.
Post your question publicly — other visitors, and whoever manages this page, can answer. It will be visible on this page.
Want to hear about this company again? Follow it and we’ll email you when a new review appears.
Based on reviews, listing data and the company's website. Verify anything important with the company directly.
The site outlines solid systems thinking with clear examples of agent workflows and local model benchmarks, though some API plans remain unimplemented.
The project examples, particularly the Manitoba Cottage Search and Student Assignment Tracker, illustrate a clear understanding of full-stack integration. Using FastAPI, HTMX, and SQLite for practical applications shows a preference for reliable, maintainable technologies. The agent-assisted development workflow documented in the assignment tracker adds an interesting layer to traditional coding processes. It is impressive how local models are benchmarked against hosted options to determine the best fit for specific tasks. This evidence-based approach to model selection suggests efficiency. One question remains regarding how these structured workflows scale in a larger enterprise environment. Does the current architecture support multi-tenant deployments easily? The mention of support triage and internal document assistants implies a direction toward enterprise relevance, but details on scalability would be valuable. Overall, the work appears well-constructed and thoughtfully designed.
Questions & Answers
No questions yet — be the first to ask.
Sign in to post
Your text is saved — you'll come right back to post.
Don't have an account? Create one
News & Press
No news yet. This is where the business shares its own updates and press.
Photos
No photos yet. Photos shared by reviewers will show up here.
About Building Practical AI Systems
Building Practical AI Systems is the personal site of David Kendrick, a practitioner who works on practical AI applications. The site documents hands-on experience building AI agents, automations, local model experiments, and browser workflows tied to real software projects. It covers topics such as workflow design, scheduled jobs, memory and logging, structured APIs, and review gates around model behavior. The work also addresses enterprise-relevant uses including safe automation, support triage, and responsible AI adoption, framed as building software infrastructure rather than novelty layers.
- Website
- davidkendrick.dev
