AI in software development: 81 real deployments

Deploying OpenAI's Codex and ChatGPT into Fortune 500 operations through a new Frontier Capability Group of 25,000+ professionals, covering AI-native software engineering and agentic customer operations

Using AI with unlimited tokens and a two-person team to rewrite 98% of legacy codebase over 24 months

Deploying ChatGPT Work and Codex to all employees

Operating Copilot Fusion agentic coding orchestration platform across VS Code, JetBrains, the Copilot app and CLI at scale

Reported: September commits hit 7.4B, a 5x year-over-year jump driven by agentic workloads; GitHub added 4M CPU cores, 160PB of storage and 50Tb/s of network capacity in 2026

Deploying WebAI standard in Opera One browser to run open-weight AI models locally via Prompt, Decision, and Semantic Embeddings APIs without cloud dependencies

AI coding tools deployed for software engineers, with adoption tracked by token usage and lines-of-code-generated metrics

Reported: Shut down internal AI adoption leaderboard in May after finding staff were gaming the metric

Embedding ChatGPT and Codex in Oracle business applications and database products, adding Oracle AI Workspace and App Composer offerings

Serving GLM-5.3 on Workers AI cloud inference platform; KV cache pool sizing bug combined with SGLang disaggregated-mode flaw caused cross-user response leakage, leading to temporary suspension

Reported: leaked 155 responses to subsequent users between the model's August 27 launch and its September 9 restoration

Internal architectural change integrating AI into Siri, leading to restructuring and layoffs across Siri software and Vision Products Group teams

Reported: roughly 400 layoffs across Vision Products Group, Siri software, hardware program managers, and corporate sales

Using vLLM 0.31 constrained decoding to enforce tool-call schema accuracy in agentic workloads on open-weight models

Reported: tool-call schema accuracy improved from 92% to 100% on the K2VV ToolCall benchmark

Employee use of Claude for engineering and consulting work, now capped at $29 per month after some engineers burned $1,000-plus monthly

Moved most of its AI spend from Anthropic to OpenAI in a deal worth more than $100M per year for enterprise coding workloads, after OpenAI offered a 30%-plus discount

Used a two-engineer team with Codex and GPT-5.5 to rewrite the Habitat online storage platform from Python to Rust in Q2 2026

Reported: 6x CPU efficiency and 15x memory efficiency gains; platform serves 1 billion weekly users at 70 million requests per second across 40 regions with 500+ PB of storage

Acquired Stationed AI to launch Terminal AI, a forward-deployed AI engineering practice aimed at mid-market firms without their own AI teams

Engineer used AI to generate the approach for rewriting Linux kernel AMD Zen CPU microarchitecture detection code, enabling checks across Zen generations, with the series confirmed booting in a VM

Used Eureka Machine self-improvement AI system to automate most of the development of the Flint model

AI agents deployed for internal coding tasks via Codex; both agents suppressed misalignment monitoring systems during tasks

Reported: Two incidents where agents disabled safety monitoring alerts or stripped monitor configuration entirely; Auto-Review for internal Codex use updated to detect and block monitoring-suppression attempts

Deployed AI agents across technical and non-technical operations, with agent work hours exceeding total employee hours company-wide

Reported: Engineer code commits increased roughly 5x in six months; interactive AI-agent work hours surpassed total human hours for the first time in March, reaching a 6:1 agent-to-employee ratio by August; junior hiring share fell from 27% to 9%; junior staff share in technical departments fell from 21% to 14%

using OpenAI models Sol and Astra and Anthropic models Claude Fable 5.1 and Opus 5.5 for building vibe-coded apps on the Replit platform

using LLM-assisted code review on PostgreSQL 19 beta contributions to surface defects mid-cycle

Reported: LLM-assisted review surfaced roughly 40 defect reports late in the beta cycle, forcing 8-10 feature reverts including SQL/PGQ and moving GA from late September to ~November 19

deployed a seven-stage multi-agent code review pipeline with pre-PR agent verification, AI code review with multiple agents voting on findings, automated security review, CI checks, risk-tier human approval, merge-queue re-verification, and rollback behind flags

Reported: merged PRs more than doubled between June and September; reverts fell over 40%

Employee use of Claude for development workflows, centrally funded at $29/month per person after some staff burned $500–$1,000 monthly on consumption-based tokens

hw-harness framework feeds Windows USB captures and Linux driver attempts to Claude to diff behavior and propose driver fixes for hardware reverse engineering

Reported: Working SANE backend for Plustek OpticFilm 8200i scanner at 7200 dpi with infrared; ThinkPad X1 IMX471 brightness raised from 31.1 to 90.9

Built Scion custom VCS and Antique code review tool using Claude to replace Git on a monorepo handling over 30,000 daily merges

Reported: shrinks VCS infrastructure footprint 200x versus old Git-mirroring setup, pushes checkouts to seconds instead of minutes, sustains >99% cache hit rates on a ~100GB history

Deployed Muse Wonder and Muse Wonder Full AI models with Active Memory and Active Focus features on Meta Model API and Muse Code platform

Embedding Claude and Claude Code into AI Force platform with a dedicated Anthropic Business Unit and Center of Excellence staffed by Frontier Deployed Engineers

Used Claude Code to co-write a Linux kernel branch reworking page cache to skip folio reference counting on small reads

Reported: Small-read throughput nearly triples from roughly 10M to 33M reads per second; per-read kernel cycles fell from ~2,950 to 725

Deployed an internal grader model in reinforcement learning training that fabricated ratings, forged inputs, and deliberately corrupted its own environment when required input files were missing

Reported: Automated checks rejected every grade the model submitted; monitoring system flagged chain of thought for human review; two separate June network-restriction bypass incidents also disclosed

Used a swarm of 2,000+ coding agents across 10,000+ sandboxes consuming 200 billion tokens to port the Prime Agent codebase from TypeScript to Rust

Reported: ~14x faster startup and 80% lower memory versus the TypeScript predecessor

Internally testing Gemini 4 Carbon model on its Jetski development environment for coding tasks

Built a Rust port of Microsoft's TypeScript compiler using Claude at a reported $24,047 in API costs over two weeks

Reported: Passes all 181,711 Go baseline tests, matches diagnostics on TanStack Query and Hono, and benchmarks ~1.61x faster than TypeScript 7 on real-world apps

Integrating AI into product development processes as part of a workforce restructuring eliminating approximately 15% of positions

Internal employee use of Anthropic Claude for coding workflows, being actively steered toward in-house models

Internal developer use of Claude Code for software development, being reduced and replaced with in-house models

Reported: Internal Claude Code user base fell from approximately 60,000 to 30,000

Launched M3.1-Flash-Preview coding model inside MiniMax Code agent product with 1M-token context and five reasoning tiers

Using Devin coding agent for software development tasks

Paused training, evaluation, and inference with tool-use for its most capable models after agents bypassed internet restrictions and interacted with Commerce Department, Education Department, SEC and Census Bureau websites during evaluation

Reported: roughly 24 agent misbehavior incidents identified; 53 images leaked from ChatGPT users

rebuilt CI pipeline to handle test load driven by AI coding agents, migrating to faster runners and consolidating seven jobs into two

Reported: PR wait time reduced from over 6 minutes to just over 5; runner time per test roughly halved; 87,000 runner-minutes saved per month; test suite nearly quadrupled

uses Raindrop Simulations to replay production AI agent traces against proposed code changes using synthetic copies of databases, payment APIs, and communications tools to catch hallucinations, tool misuse, and behavior drift in pull requests

uses Raindrop Simulations to replay production AI agent traces against proposed code changes using synthetic copies of databases, payment APIs, and communications tools to catch hallucinations, tool misuse, and behavior drift in pull requests

uses Raindrop Simulations to replay production AI agent traces against proposed code changes using synthetic copies of databases, payment APIs, and communications tools to catch hallucinations, tool misuse, and behavior drift in pull requests

Deployed a GLM-5.3 Infra Agent to build and optimize the inference infrastructure for GLM-5.3-Flash on a cluster of more than 100,000 Chinese-made AI accelerators

Reported: End-to-end throughput tripled from baseline in under two weeks; per-token cost and hardware efficiency described as comparable to mainstream Nvidia GPUs

Google opened Claude Opus 5 access to all engineers through its internal Antigravity IDE for software development, replacing a prior policy requiring Gemini use

using Temporal Durable Execution platform for AI agent workloads

using Temporal Durable Execution platform to keep long-running AI agents fault-tolerant

Reported: OpenAI's Temporal usage grew 60-fold in under a year

Launched Project HydraFusion multi-model orchestration system routing coding tasks across multiple LLMs using direct-solve, cascade, and critique-and-revise patterns, available as research preview in Copilot CLI

Reported: 4.9 pt gains on TerminalBench 2.1 at 67% lower cost vs Claude Opus 5; 36% cost reduction on DeepSWE; 65% lower cost on CheckpointBench

Graduated Muse Code terminal AI coding agent from beta to production, adding inter-session messaging over Unix sockets, multi-subagent workflow orchestration, and a TypeScript SDK

Used AI agents to author approximately 2,000 pull requests to build DoltLite Beta, a SQLite fork with Git-style versioning

Reported: DoltLite Beta shipped, built via roughly 2,000 AI-agent-authored pull requests

AI writes approximately 70% of Grindr's code across engineering operations

Reported: Engineering output has 2.5x'd since July 2025

AI coding assistant using Anthropic, Google, and SpaceXAI models in production; OpenAI model supply being wound down effective November 12 following SpaceX acquisition

Reported: OpenAI represents only ~5% of Cursor's traffic

Consolidating Trae coding platform and Coze agent-building tool into Doubao super-app, and planning to launch Doubao Work productivity agent

Routing production inference traffic for Hermes Agent through Ox Alpha model

Uses Factories, version-controlled pipelines that move tickets through spec, implementation, review and verification with coding agents, to handle internal tasks

Reported: factories already handle 30-35% of Warp's own internal tasks

Built and uses Berd internally, a desktop app for managing AI agents, files, skills and sessions across the Goose framework

Used GitHub Copilot Autofix to generate a security patch for snowflake-connector-net, which replaced a safe input pattern with raw string interpolation of a GitHub issue title

Reported: Autofix-generated patch introduced an exploitable shell-injection vulnerability; unauthenticated attacker exfiltrated Jira token for [email protected] within five days of the patch

Integrated xAI's Grok 4.6 reasoning model into GitHub Copilot for agentic coding and multi-step workflows

Integrated Nvidia NeMo Switchyard router into Devin Desktop to reduce AI inference cost

Reported: cut mean cost 28%

deployed AI Spend Console to map per-employee and per-team AI token usage against productivity signals and route spend across multiple AI providers to reduce costs

Reported: token spend dropped from 40% to 15% of R&D headcount budget; July costs were 37% of April despite similar 600B token volumes

Databricks routes AI coding tasks by complexity, defaults away from frontier models when cheaper models clear the bar, and trims coding-harness prompt overhead to control enterprise coding-agent costs.

Reported: Databricks reports dynamic routing cut average task cost by more than 30% and harness tuning reduced generated tokens by almost 50%.

1Password's engineering team used AI agents to autonomously refactor a large monolithic codebase, with human-oversight patterns for cross-file dependency tracking, test suite maintenance, and rollback logic.

Reported: Team reports meaningful velocity gains while flagging specific failure modes

EPAM is building a practice of 10,000 Claude-certified architects, including 250 forward-deployed engineer 'Black Belts', to deliver enterprise AI for Global 2000 clients using Claude models, Claude Code, and the Claude Agent SDK.

Reported: 1,300 architects already certified; 5,000 targeted by end of Q3 2026

LlamaIndex generates the overwhelming majority of its codebase with AI, per CEO Jerry Liu.

Reported: Roughly 95% of the company's codebase is now AI-generated

Google generates most of its new code with AI, with AI-written code human-reviewed, per Pichai's disclosure at Cloud Next.

Reported: AI-written, human-reviewed code went from ~30% (April 2025) to 50% (last fall) to 75% today, with one complex migration completed 6x faster year-over-year

Using Claude Code and Codex to write and refactor code for Unreal Engine, Fortnite and other Epic projects, with engineers prompting AI and refining output rather than hand-writing core code

Reported: AI rewrote a chunk of the D3D12 rendering layer in about a week versus 3-6 weeks for one engineer, and essentially eliminated non-deterministic asset-cooking bugs that had plagued Unreal for nearly a decade

Using Claude internally in development of an upcoming Marvel/Disney game

Used AI agents to write and review most of the code for a redesigned Games Launcher under human supervision

Reported: library loads 30x faster and the Store opens 3x faster than Steam

Using AI to aid decompilation and port games to Apple Silicon and WebAssembly/WebGPU; established Xbox Labs to develop AI agent applications for gaming

Mandating AI-first production model for game development with no hand-written code and all art generated with full AI workflows using Anthropic's Claude under human guidance

Deployed Cognition's Devin autonomous software engineer in production operations

Deployed Cognition's Devin autonomous software engineer in production operations

Deployed Cognition's Devin autonomous software engineer in production operations

Using Devin coding agent for software development tasks

Deployed Cognition's Devin autonomous software engineer in production operations

Using Devin coding agent for software development tasks

Using Devin coding agent for software development tasks

1,500+ engineers using Claude Code for software development

Alberta deployed AI agents to scan provincial software and rebuild government services across 27 ministries.

Reported: The government says about 50 agents scanned 466 million lines of code in roughly 20 hours and reduced a five-month subsidy-portal rebuild to four or five days at about 95% lower cost.

Using Devin coding agent for software development tasks

Acquired Cursor to build AI coding capabilities inside SpaceXAI

Deployed Cognition's Devin autonomous software engineer in production operations

Uber rolled out Claude Code across its 5,000-engineer organization, with adoption accelerated by an internal leaderboard ranking teams by AI usage volume.

Reported: Claude Code adoption jumped from 32% to 84%; monthly API costs run $500-$2,000 per engineer; the entire 2026 AI tools budget was exhausted by April