Share
Embed
The Agentic Era Of Mobile Engineering
Key learnings on how LLMs are transforming engineering teams, the challenges of working with non-deterministic AI, and practical approaches for integrating AI effectively into team workflows.
Erik Jhordan Rey
April 10, 2026
Transcript
27 slides
-
The Agentic Era Of Mobile Engineering April 2026 Erik Jhordan
Rey, he/him/his Senior Manager, SDE -
2 1 Meeting Topic AI is changing how engineering works
2 How We apply AI in our Mobile Team 3 Safe and Smart AI Adoption (Guardrails) -
“AI doesn’t just change how we code, it changes how
we build software” Engineering AI -
6 AI speeds up understanding, humans ensure relevance • it
helps teams understand problems faster • generate structured requirements • identify risks early • AI supports decisions, it doesn’t replace them AI Capabilities • Set priorities and business context • Domain knowledge is critical • Validate outputs and assumptions • Make final decisions Human Role Engineering AI -
9 AI amplifies architectural vision and expands design possibilities •
Suggest High-level architecture choices • Recommends how to organize modules, classes, or services for maintainability • Generate diagrams, data models, and API contracts • Compare performance, cost, and complexity across options • Highlights potential edge cases, bottlenecks, or architectural risks early AI Capabilities • Choose the right architecture for context • Validate scope, assumptions and constraints • Balance technical and business trade-offs • make decisions based on business needs, team capabilities, and domain knowledge Human Role Engineering AI -
“Choosing the right architecture still depends on experience, context, and
business needs” Engineering AI -
12 AI accelerates coding, but humans ensure the code is
safe, maintainable, and reliable • Automates repetitive code setup, templates, and project structure • Autonomous Code Generation • Assist in Real-Time Coding • Highlights potential errors, suggests fixes, and improves code • Creates test cases to ensure coverage and reliability. AI Capabilities • Review and validate AI-generated code • Debug and fix mistakes • Make architectural and logical decisions • Ensure quality, security, and maintainability • AI writes code, humans ensure it works Human Role Engineering AI -
“AI tools change the work, but what makes a great
engineer remains largely unchanged.” Engineering AI -
16 A new experience that helps buyers and renters discover
homes through conversation, receive guidance and take real steps such as scheduling tours or connecting with a local agent. “We’re connecting the entire housing journey with AI in a way that hasn’t been possible before,” said Jeremy Wacksman, CEO of Zillow. https://www.zillow.com/news/zillow-debuts-ai-mode/ Zillow AI mode, bringing guided intelligence to every step of the housing journey Apply AI -
18 • Prompt engineering & AI model interaction • AI-assisted
code review & refactoring • AI-driven testing & automation • LLMs • MCPs • Skills • Ethical AI & responsible implementation AI Development AI-powered development tools for Android & iOS AI Development Tools Claude Opus 4.6 Claude Sonnet 4.6 Claude Haiku 4.5 GPT 5.4 GPT 5.3 Codex Composer 1.5 Model Cursor • IDE • CLI Claude Code • TUI • Desktop App Codex • CLI • Desktop App Apple Intelligence • Xcode Environment -
Piloting Emerging AI Tools for Development Learning what works in
practice Apply AI -
20 AI Testing Tools AI-Driven Testing • AI Testing tools
allow users to write UI tests without any code. • Automate functional testing and edge case detection • An AI agent follows your test steps and interacts with the app as a human would, performing actions and verifying results. • If the AI tester cannot complete an execution step or fails to confirm a verification step, it fails the test and explains why. Pros Cons • Tests written in natural language • Ability to run tests that mirror the user’s actual experience • Tests are easier to maintain and adjust • Visual reports of test results • Tests detail in plain english what went wrong when they fail • Non-determinism causing inconsistent result • Buggy behavior internal to the frameworks • LLM has to reason out every step, costing tokens and time • These tools are still changing and evolving and may be unstable Apply AI -
23 • Duration from task initiation to completion • Measures
if AI is helping the team deliver faster Cycle Time Change Failure Rate Measuring AI Success in Engineering Acceptance Rate • Check if faster commits create bottlenecks in Code Review or QA • Ensures overall workflow efficiency is improved • Ensures speed doesn’t compromise quality • Tracks errors or bugs introduced by AI-generated code • Frequency of AI-generated suggestions actually used by developers • Indicates practical usefulness, not just availability. Downstream Effects AI Adoption Decide what success looks like for your team! -
Common Pitfalls to Avoid AI amplifies what’s already there, good
or bad. AI Adoption -
25 AI is powerful, but safe, responsible use keeps your
team efficient and secure • Define ethical standards and AI usage guidelines • AI amplifies workflows, doesn’t fix broken processes • Train engineers to prompt, validate, and guide AI • Avoid unvetted tools that could leak data • Always review AI-generated code manually • Creates test cases to ensure coverage and reliability. AI Adoption • Monitor AI recommendations to ensure alignment with project goals • Document lessons learned to refine AI adoption strategy • Keep an eye on costs make sure the tools are worth it “Don’t just copy-paste AI outputs always verify, because LLMs aren’t always correct or the source of truth” -
26 Further Reading • Zillow Mexico Careers • The next
chapter of AI in real estate • How Zillow’s new AI mode works • Exploring Generative AI • To vibe or not to vibe • Agentic AI Foundation • Zillow AI mode Engineering AI