Developer replaces traditional portfolio with AI chatbot that answers questions about him

·1 views

A developer named Aaron rebuilt his personal portfolio site to feature an AI chatbot as the primary interface, allowing visitors to ask open-ended questions about his background, projects, and experience. Rather than presenting a fixed layout of project cards and work history, the site retrieves relevant information from Markdown files and generates contextual responses on demand. The frontend is built with SvelteKit, while backend orchestration runs through n8n. Shortly after launching, real users began submitting complex queries that caused serverless functions on Netlify to time out, effectively killing responses mid-stream. Aaron responded by moving the chat endpoint to self-controlled infrastructure to handle longer-running requests more reliably.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Atlas Sanctum Aims to Build AI Platform for Regenerative Social and Ecological Impact

Atlas Sanctum is a conceptual AI-powered platform submitted for a weekend developer challenge, designed to channel generosity into measurable real-world outcomes for people and ecosystems. The project envisions technology as infrastructure for coordination, connecting those who can help with those who need it, while tracking social and ecological results rather than purely financial ones. The platform is structured around modular components covering knowledge sharing, trust verification, opportunity discovery, and resource coordination. A demo and open-source codebase have been published on GitHub as part of its early development. Developers intend to evolve it into an open ecosystem of regenerative intelligence tools, APIs, and applications.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Python Fundamentals Revisited: Why Data Structures and Built-ins Matter Most

A data science learner returning to Python after SQL practice found that deliberately revisiting core concepts — lists, dictionaries, sets, loops, and functions — significantly improved their coding efficiency. Choosing the right data structure based on the question being asked, rather than habit, led to noticeably shorter and cleaner code. Built-in functions like sorted(), max(), and min() with the key argument replaced multi-line manual loops with single-line solutions. The author also reframed functions not just as tools for code reuse but as a way of naming and communicating intent, making scripts easier to read later. Switching from notebook-based coding to running scripts directly in the terminal was highlighted as an underrated but practical skill that surfaces errors more reliably.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Narrow AI Agents Outperform All-in-One Bots, Developer's Production Tests Show

A software developer shared lessons from building a general-purpose AI agent that could handle web search, file editing, Slack messaging, database queries, and more, only to find it consistently produced errors in production. The agent's failures — including wrong calendar bookings, a repo-reformatting pull request, and outdated pricing answers — all stemmed from having too many tools and no clear, singular purpose. The developer replaced the bloated agent with tightly scoped specialist agents, each given a one-sentence job description and only the tools strictly necessary for that task. In one documented case, a billing specialist agent with just three tools dramatically reduced wrong answers by restricting the model to a small, verified dataset and blocking it from browsing unreliable public sources. The core finding is that AI agent reliability improves not by upgrading the underlying model, but by narrowing the agent's scope, tools, and success criteria.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

How Claude Code's allow, ask, and deny permissions actually work together

Claude Code's permission system uses three rule lists — allow, ask, and deny — evaluated in a fixed order where deny always wins, regardless of how specific an allow rule is. Rules are sourced from multiple settings files including project-level and user-level configs, with deny overriding across all scopes. Common pitfalls include glob patterns that match unintended commands, wrapper-stripping behavior that can bypass rules, and compound shell commands where each segment must be independently authorized. Bare tool-name denials remove a tool from the model's context entirely, while pattern-based denials like Bash(rm *) still leave the tool visible but block matching commands. The guide recommends never allowing unrestricted Bash access and treating every permission denial as useful data to refine policy over time.

0 comments Read more at DEV Community