AI Makes Software Estimation Faster at Best, Riskier at Worst

·1 views

AI tools have reshaped software project estimation by compressing best-case timelines through faster code generation, but they have also made worst-case scenarios harder to predict. The time saved generating code often shifts to reviewing AI output that appears correct but can introduce subtle bugs, especially at system integration points. Three-point estimation — a PERT-based technique using optimistic, most-likely, and pessimistic figures weighted by a standard formula — remains a reliable framework for quoting realistic timelines to clients. However, developers must resist the temptation to lean on the more attractive AI-assisted optimistic number when committing to deadlines. The core challenge is that AI redistributes work rather than eliminating it, making the pessimistic estimate more complex and the middle-ground quote wobblier than before.

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)

Beginner SEO Diary: How to Build a Keyword List Using Only Free Tools

A blogger documenting their SEO learning journey shares the process they followed on Day 4 to build their first keyword list using free tools like Google Keyword Planner, AnswerThePublic, and Search Console. Industry data suggests long-tail phrases of three or more words account for roughly 92% of all searches and convert at more than double the rate of short, generic terms. The writer emphasizes that matching a keyword to search intent is more important than chasing high search volume, noting that a mismatch between content format and what Google already ranks can prevent a page from ranking regardless of writing quality. They recommend manually reviewing actual search results for each shortlisted keyword, a step they describe as the most useful filter that many beginner guides overlook. Volume figures from free tools, particularly Google Keyword Planner, are advised to be treated as rough estimates rather than precise guarantees.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

How One Developer Built a Home Server from an Old Laptop for Free

A developer set up a home server using an old, unused laptop to explore low-cost ways for students to host software projects without paying for cloud services. Inspired by a conversation with a friend, he installed Ubuntu Server LTS on the laptop after wiping Windows and creating a bootable USB drive. The setup required no graphical interface, relying entirely on terminal commands for configuration and package installation. With the server running, he decided to self-host his personal website and experiment with a media server. The project took an unexpected turn when he discovered Docker and containers, opening up a broader range of possibilities.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

How to Fix Common Code Splitting Mistakes and Boost Web App Performance

Code splitting is a widely used technique that breaks large JavaScript bundles into smaller, on-demand chunks to improve initial load times, but improper implementation can slow applications down rather than speed them up. Common pitfalls include over-splitting, which generates too many small chunks and increases network request overhead, and under-splitting, where chunks remain too large to deliver real performance gains. Developers also face issues with incorrect splitting points that create waterfall loading delays, as well as shared module duplication that inflates total download size across multiple chunks. Large third-party libraries can further undermine splitting efforts by bloating otherwise targeted chunks. Identifying these problems requires careful analysis of bundle structure and performance metrics before applying targeted fixes.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

How CQRS Pattern Fixed a Construction App's Weekly Crash Under Peak Load

A field workforce reporting platform repeatedly crashed every Friday at 4:30 PM when thousands of construction workers simultaneously submitted weekly logs, overwhelming the database with concurrent read and write requests. Engineers redesigned the system using Command Query Responsibility Segregation (CQRS), routing write requests through Azure Service Bus so background workers could process submissions at a steady pace rather than all at once. The change eliminated end-of-week downtime by spreading a five-minute traffic spike across a manageable fifteen-minute processing window. However, the shift to asynchronous processing introduced eventual consistency, causing project managers to see outdated dashboards and prompting frustrated workers to resubmit forms multiple times, generating duplicate records. The team addressed this by adding unique submission IDs to prevent duplicates and displaying a processing status indicator in the UI so users understood their data was being handled.

0 comments Read more at DEV Community