Agentic Engineering Patterns >

Many developers worry that outsourcing their code to AI tools will result in a drop in quality, producing bad code that's churned out fast enough that decision makers are willing to overlook its flaws.

If adopting coding agents demonstrably reduces the quality of the code and features you are producing, you should address that problem directly: figure out which aspects of your process are hurting the quality of your output and fix them.

Shipping worse code with agents is a choice. We can choose to ship code that is better instead.

Avoiding taking on technical debt

I like to think about shipping better code in terms of technical debt. We take on technical debt as the result of trade-offs: doing things "the right way" would take too long, so we work within the time constraints we are under and cross our fingers that our project will survive long enough to pay down the debt later on.

The best mitigation for technical debt is to avoid taking it on in the first place.

In my experience, a common category of technical debt fixes is changes that are simple but time-consuming.

• Our original API design doesn't cover an important case that emerged later on. Fixing that API would require changing code in dozens of different places, making it quicker to add a very slightly different new API and live with the duplication.
• We made a poor choice naming a concept early on - teams rather than groups for example - but cleaning up that nomenclature everywhere in the code is too much work so we only fix it in the UI.
• Our system has grown duplicate but slightly different functionality over time which needs combining and refactoring.
• One of our files has grown to several thousand lines of code which we would ideally split into separate modules.

All of these changes are conceptually simple but still need time dedicated to them, which can be hard to justify given more pressing issues.

Coding agents can handle these for us

Refactoring tasks like this are an ideal application of coding agents.

Fire up an agent, tell it what to change and leave it to churn away in a branch or worktree somewhere in the background.

I usually use asynchronous coding agents for this such as Gemini Jules, OpenAI Codex web, or Claude Code on the web. That way I can run those refactoring jobs without interrupting my flow on my laptop.

Evaluate the result in a Pull Request. If it's good, land it. If it's almost there, prompt it and tell it what to do differently. If it's bad, throw it away.

The cost of these code improvements has dropped so low that we can afford a zero tolerance attitude to minor code smells and inconveniences.

AI tools let us consider more options

Any software development task comes with a wealth of options for approaching the problem. Some of the most significant technical debt comes from making poor choices at the planning step - missing out on an obvious simple solution, or picking a technology that later turns out not to be exactly the right fit.

LLMs can help ensure we don't miss any obvious solutions that may not have crossed our radar before. They'll only suggest solutions that are common in their training data but those tend to be the Boring Technology that's most likely to work.

More importantly, coding agents can help with exploratory prototyping.

The best way to make confident technology choices is to prove that they are fit for purpose with a prototype.

Is Redis a good choice for the activity feed on a site which expects thousands of concurrent users?

The best way to know for sure is to wire up a simulation of that system and run a load test against it to see what breaks.

Coding agents can build this kind of simulation from a single well crafted prompt, which drops the cost of this kind of experiment to almost nothing. And since they're so cheap we can run multiple experiments at once, testing several solutions to pick the one that is the best fit for our problem.

Embrace the compound engineering loop

Agents follow instructions. We can evolve these instructions over time to get better results from future runs, based on what we've learned previously.

Dan Shipper and Kieran Klaassen at Every describe their company's approach to working with coding agents as Compound Engineering. Every coding project they complete ends with a retrospective, which they call the compound step where they take what worked and document that for future agent runs.

If we want the best results from our agents, we should aim to continually increase the quality of our codebase over time. Small improvements compound. Quality enhancements that used to be time-consuming have now dropped in cost to the point that there's no excuse not to invest in quality at the same time as shipping new features. Coding agents mean we can finally have both.

Tags: coding-agents, ai-assisted-programming, generative-ai, agentic-engineering, ai, llms

sziring shares a report from Business Insider: Silicon Valley has long competed for talent with ever-richer pay packages built around salary, bonus, and equity. Now, a fourth line item is creeping into the mix: AI inference. As generative AI tools become embedded in software development, the cost of running the underlying models -- known as inference -- is emerging as a productivity driver and a budget line that finance chiefs can't ignore. Software engineers and AI researchers inside tech companies have already been jousting for access to GPUs, with this AI compute capacity being carefully parceled out based on which projects are most important. Now, some tech job candidates have begun asking about what AI compute budget they will have access to if they decide to join. "I am increasingly asked during candidate interviews how much dedicated inference compute they will have to build with Codex," Thibault Sottiaux, engineering lead at OpenAI's Codex, the startup's AI coding service, wrote on X recently. He added that usage per user is growing much faster than overall user growth, a sign that AI compute is becoming even scarcer and more valuable. That scarcity is reshaping how engineers think about their work and pay. "The inference compute available to you is increasingly going to drive overall software productivity," said OpenAI President Greg Brockman. The report cites a recent compensation submission from a software engineer that listed "Copilot subscription" as part of the pay and benefits. "OpenAI and Anthropic should create recruitment sites where their clients can advertise roles, listing the token budget for the job alongside the salary range," said Peter Gostev, AI capability lead at Arena, a startup that measures the performance of models. Tomasz Tunguz of Theory Ventures predicts AI inference will be the fourth component of engineering compensation, alongside salary, bonus, and equity. "Will you be paid in tokens? In 2026, you likely will start to be," Tunguz said.

Read more of this story at Slashdot.

Seattle-area Microsoft employees who are showing up in the office three days a week are also showing up on roadways and impacting commuters’ speeds, according to new data from traffic analysis company Inrix.

Inrix measured travel speeds on eastbound and westbound SR 520 and southbound and northbound I-405 during the weeks of Feb. 23 and March 2. Many of Microsoft’s more than 50,000 employees in the region rely on the roadways and bridges connecting Seattle and the Eastside to the company’s headquarters campus in Redmond, Wash.

The data shows speeds on 520 dropped across all days during the first week, with speeds on Tuesday, Wednesday, and Thursday showing the slowest travel speeds over just over 30 mph.

Morning commute speeds between Tukwila and Bellevue fell as much as 35% and as much as 25% between Lynnwood and Bellevue. The evening commute saw speeds drops as much as 27% between Bellevue and Tukwila on Friday while speeds fell 21% northbound between Bellevue and Lynnwood, Inrix reported.

Microsoft isn’t dictating from above which three days people will need to be in the office. Specifics are left to individual teams and managers. Some groups may require more than three days, and certain customer-facing roles like field sales and consultants are exempt.

The region’s roadways could get some relief when Sound Transit’s Crosslake Connection opens March 28, finally linking Seattle and the Eastside by light rail across Lake Washington — connecting downtown Seattle to downtown Bellevue and the Redmond Technology station at Microsoft headquarters.

Previously: Microsoft’s new RTO policy starts Feb. 23, bringing Seattle-area workers back 3 days a week

An overview of Gemini Embedding 2, our first fully multimodal embedding model that maps text, images, video, audio and documents into a single space.

Meta is acquiring Moltbook, a Reddit-like platform where AI agents can make and comment on posts, as first reported by Axios. In a statement to The Verge, Meta spokesperson Matthew Tye confirmed the Moltbook team will join Meta Superintelligence Labs as the company looks for "new ways for AI agents to work for people and businesses."

Matt Schlicht and Ben Parr launched Moltbook earlier this year, offering a "social" network for autonomous agents powered by the open-source AI assistant OpenClaw (formerly Moltbot). The platform went viral earlier this year for a number of posts - including one that asks questions about AI consciousness - thou …

Read the full story at The Verge.

Adobe is also adding new AI-powered image-editing features to Firefly.