Jads Blog

Jeff Adler's engineering blog. AI, agentic engineering, leadership, iOS architecture, and technical deep dives from Google, Dropbox, and Reddit.

Adopt the Agent. Build the Loop. 2026-05-29 -- Everyone is debating Cursor vs Claude Code. Wrong fight. The agent is a commodity. The only thing worth building is the loop around it.
The Manager Layer is Next 2026-05-16 -- AI isn't replacing managers. It's exposing which ones were never load-bearing to begin with.
Tokenmaxxing Is What Happens When You Measure Wrong 2026-05-15 -- Mandating AI tool usage is correct. Measuring it is a trap. The right move is to measure outcomes, not inputs.
Relaunching This Blog for the AI Era 2026-03-04 -- Jeff Adler's engineering blog, eight years of iOS architecture and leadership. Now relaunching for the AI era.
The Year the Agent Stopped Being a Demo 2025-12-16 -- 2025 was the year agents went from demo to operational. Some 2024 predictions held. Some didn't. Three bets for 2026, and the gap that hasn't closed.
AI Is Redefining What \"Senior Engineer\" Means 2025-11-07 -- Senior engineer used to mean better code. AI is redefining it to mean system design, taste, and judgment. A Dropbox director's take.
What Two Years of Agentic Coding Did to My Org Chart 2025-10-28 -- Sixteen months after the agent-managers thesis, here's what actually happened to my org chart. Some held up. Some didn't. Hiring shifted hardest.
Your Agent's Bottleneck Is Your Documentation 2025-08-12 -- GPT-5 landed muted. The model layer is stabilizing. The differentiator moved one layer down to your information architecture for agents.
Voice Is Going to Fail for the Same Reason Chat Did 2025-05-28 -- Voice as a default surface is failing the way chat did. Voice as an embedded wedge is the actual win, and the early-2025 data already shows it.
AI Makes You 3x Faster, Then Review Explodes 2025-04-04 -- AI tools help Dropbox engineers ship 3-5x faster, but without the right processes you just create a traffic jam at code review.
Reasoning Is a Commodity Now 2025-02-26 -- DeepSeek R1 and Claude 3.7 made reasoning a commodity in five weeks. The architectural question moves from which provider to where you absorb the cost.
Should We Write Code for LLMs Now? 2025-01-24 -- If LLMs increasingly read and modify our code at Dropbox, should our conventions optimize for them instead of humans?
Your User Is Not You 2024-12-16 -- The biggest AI product mistake of 2024 was shipping for ourselves. Users don't know what to prompt and bounce on the first reply. Empathy gap is PMF gap.
Agentic Workflows Are Harder Than You Think 2024-10-11 -- The gap between an impressive agent demo and production at Dropbox is enormous. Here's what bridging it actually takes.
Commit Your Specs (We're Almost There) 2024-09-27 -- o1 is the first model where the spec might be the load-bearing artifact, not the code. I am hesitant. The QA bar this requires is one the industry hasn't built.
From Copilot to Pipeline 2024-07-30 -- Cursor and Copilot are the trust-but-verify phase of AI coding. The next phase is a pipeline of agents that writes the code while you review.
Engineers Are Becoming Agent Managers 2024-06-07 -- AI is flattening the engineering output curve at Dropbox. Engineers are managing agents now. Org structures need to change.
Stop Picking Models. Pick What's Stable Around Them. 2024-03-28 -- Claude 3 dethroned GPT-4. Devin redefined the agent. Teams that hardcoded a model are scrambling. The architecture that survives picks anything but the model.
Stop Building Chat. Your Users Don't Want Another Box. 2024-02-14 -- Every B2B company is racing to bolt an "Ask AI" chat onto their product. Most users bounce on the first reply. The chat box is product engineering laziness.
Staff to Senior Manager: 90% the Same Job 2023-11-15 -- I went from Staff Engineer to Senior Engineering Manager at Dropbox in three months and was surprised how much I was already doing.
What I Got Wrong About RAG 2023-09-15 -- RAG isn't what makes your AI product good -- retrieval quality is. Most teams learn that the hard way by optimizing the wrong layer first.
Shipping AI When Nothing Works Yet 2023-08-25 -- The messy reality of shipping LLM-powered features at Dropbox with real users, real latency budgets, and a cost model that doesn't work.
The Evaluation Gap Is the AI Gap 2023-07-28 -- Teams that build real eval infrastructure first ship better AI products -- not because they have better models, but because they know when their model is wrong.
Prompt Engineering Is Not a Job 2023-06-13 -- Hiring a prompt engineer is a category error. The real job is eval design, statistical rigor, and infrastructure -- and no one is posting that JD.
Latency Is a Feature Decision, Not an Infrastructure Problem 2023-04-21 -- How you handle latency determines whether users trust your AI product. Make that decision at design time, not in production.
ChatGPT Just Made AI a Product Engineering Problem 2022-12-08 -- ChatGPT launched eight days ago. Mobile engineers thinking "this doesn't affect me" are wrong. The product engineering reset starts here.
Mandates Fail, Examples Win 2022-11-04 -- Scaling engineering practices to a large iOS org. Why mandates fail and consistency comes from tooling, not enforcement.
Aggressive iOS Version Drops Are a Platform Mistake 2022-09-27 -- iOS 16 shipped two weeks ago. Most teams about to drop iOS 14 are trading short-term cleanup for long-term user pain. Stage features, don't drop versions.
We Ditched HLS and Cut Playback Errors by 22% 2022-08-12 -- How I rebuilt Reddit video playback by ditching HLS for an LRU-based prefetch cache, cutting playback errors meaningfully.
SwiftUI 4 Is Production-Ready. For Some of You. 2022-06-17 -- WWDC 2022 closed the biggest SwiftUI gaps. SwiftUI 4 works in production. For greenfield. Mixed-mode is the right call for everyone else.
SliceKit: Composition and Testing (Part 2) 2022-05-06 -- How SliceKit's composition model works in practice, plus the testing strategy that got the Reddit iOS org on board.
Modularization Is a Migration, Not a Decision 2022-04-14 -- Most mobile orgs treat modularization as a Q3 project. It's actually a multi-year migration with org-design implications. Treat it like infra.
SliceKit: Declarative UI at Reddit (Part 1) 2022-03-18 -- How I built a declarative UI framework for the iOS org at Reddit, and why consistency at scale requires opinionated tooling.
The Build-Time Tax Is the Most Underpriced Cost in Mobile 2022-02-22 -- At 100+ engineer mobile orgs, build time is the single biggest tax on engineering productivity and the most under-invested area on every roadmap I see.
Apple Silicon Cut My Mobile Build Times in Half 2021-12-09 -- Six weeks on an M1 Pro and my real iOS build runs roughly twice as fast as the 2019 Intel MacBook it replaced. The laptop-refresh ROI writes itself.
Why I Left Dropbox for Reddit 2021-11-19 -- After three years at Dropbox, I joined Reddit to lead iOS platform engineering for the 12th most-visited website.
iOS 15 Made Privacy a Product Engineering Problem 2021-10-21 -- iOS 15 plus ATT turned privacy from compliance into product engineering. Teams treating it as legal-team work are losing revenue this quarter.
Swift Concurrency Changes How You Hire iOS Engineers 2021-08-18 -- WWDC 2021 shipped async/await and Actors. Most iOS hiring rubrics still test for legacy GCD. The teams that update their interviews are the ones that win.
The Best Frameworks Come from Product Engineers 2021-06-25 -- The best mobile frameworks come from product engineers solving their own problems at Dropbox, not platform teams building in a vacuum.
Cross-Platform Code Is a Tax You Pay Twice 2021-05-06 -- The default reflex is to share iOS and Android code to save effort. Most teams pay that effort twice: once in the abstraction, once in the special-cases.
We Shipped SwiftUI. It Crashed 1% of Users. 2021-03-12 -- We rolled out SwiftUI to a slice of Dropbox users and hit a 1% crash rate. Here's why we chose UIKit and declarative frameworks instead.
A Year of Remote Engineering, Now What 2021-02-24 -- Twelve months ago I said remote was better for deep work but worse for everything else. A year of data later, the take mostly held, with two things I got wrong.
Six Weeks on M1: The iOS Engineering Verdict 2020-12-15 -- Six weeks on an M1 Mac mini. iOS engineering verdict, build times noticeably faster, fans silent, Rosetta absorbing the tooling tax. Refresh in 2021.
GPU Shaders Gave Us a 10x on Document Scanning 2020-11-20 -- How my team at Dropbox achieved a 10x performance gain in mobile document scanning by moving from CPU-bound C++ to GPU shaders.
iOS 14's One-Day Launch Was a Platform Mistake 2020-09-24 -- Apple shipped iOS 14 with less than 24 hours' notice. The widget-launch chaos was a structural process failure, not a pandemic-driven one-off.
Apple Silicon Is the Real Story of WWDC 2020 2020-07-14 -- WWDC 2020 had iOS 14, widgets, Big Sur. The story was none of those. Apple Silicon is the biggest mobile-engineering hardware shift this decade.
Remote Is Better for Deep Work 2020-06-19 -- Remote is more productive for deep work at Dropbox. In-person unlocks everything else. CI/CD becomes your most important teammate.
The Contact Tracing API Is a Mobile Platform Inflection 2020-04-21 -- Apple and Google's COVID Exposure API isn't really about contact tracing. It's two OS makers cooperating on a privacy protocol in 11 days.
Acquisition to App Store in Three Months 2020-03-13 -- How my team at Dropbox shipped HelloSign Mobile in three months post-acquisition by finding the right boundary between web and native.
Mac Catalyst Is Not Ready Yet 2020-02-05 -- Six months after Catalyst shipped, the production verdict is in. "Check a box in Xcode" collapses on platform conventions. Don't ship a Catalyst app in 2020.
Breaking Apart an iOS Monolith 2019-12-06 -- How I broke apart a Dropbox iOS monolith using protocol-first design, turning a tangled dependency graph into independent modules.
The Declarative Skeleton That Cut Code by ~40% 2019-10-25 -- How I built a declarative app skeleton that eliminated boilerplate, cut code substantially, and made test coverage a byproduct of good architecture.
Five RxSwift Patterns That Actually Work 2019-07-19 -- Five practical RxSwift patterns for well-architected iOS apps, from reactive mutable lists to the retain cycle trap nobody warns you about.
Minerva: The Coordinator Pattern Done Right 2019-05-10 -- Part 3: Testable navigation by separating what to present from how to present it. Deep linking becomes a natural consequence.
Minerva: Kill Your Imperative List Code 2019-03-22 -- Part 2: CellModels turn iOS list management from imperative data source manipulation into declarative state descriptions.
Minerva: An iOS Framework Nobody Asked For 2019-02-08 -- Part 1: Why I built an open-source iOS coordinator and list framework, what existing solutions get wrong, and the core protocol.
Google to Dropbox: Smaller Scale, Higher Velocity 2019-01-18 -- Moving from Google to Dropbox meant smaller scale but higher velocity. Embracing open source changed what was architecturally possible.
Leaving Google After Two and a Half Years 2018-11-02 -- After building iOS at Google, I'm moving on. What I learned about massive refactors, build systems, and when a company is too big.
VIPER Is Half Dead in Modern Swift 2018-07-14 -- VIPER was a breakthrough in iOS architecture, but modern Swift makes half its layers unnecessary. A protocol-driven alternative.
iOS Architecture at Google 2018-03-22 -- What iOS development looks like inside Google: internal promises frameworks, years resisting Swift, and everything built from scratch.
Inside iOS at Google Scale 2018-01-19 -- Inside the build system, source control, and tooling that powers iOS development at Google, and why long builds change everything.