SWFT
Compare 24 documented systems across software, finance and other industries. Find how each works, who reviews the result, and what is publicly available.
Search for a company, system or task, or browse by sector. Reported use describes the scope of the workflow. Access tells you what is public, commercial or internal.
Company results are first-party reports unless otherwise labeled; unknowns are written out. Source dates describe the evidence window, not a promise that every system still operates unchanged.
24 company cases · alphabetical, not ranked · source checks range from September 1, 2026 to September 8, 2026 by Codex (AI agents); each row retains its own date
24 of 24 cases
Scroll sideways to compare all columns. Tab to the comparison, then use Left and Right Arrow. Home and End move to the first and last columns.
| Company and system | Main job | Reported use | Human checkpoint | Access | Sources checked |
|---|---|---|---|---|---|
| AirbnbLLM migration pipelineTransport and travel | Update nearly 3,500 React test files and check each result. | Single-task production lineOne completed production migration program. | People improved the workflow and finished the final 3% of files. | Not publicly availableThe migration system is described publicly; its code is internal. | First-party report |
| AmazonQ Developer code transformationCommerce | Upgrade large portfolios of production Java applications. | Single-task production lineTens of thousands of migrations; AWS is retiring the product. | Developers review and accept each proposed upgrade. | Existing customersEligible existing customers retain transformation access through April 30, 2027; AWS directs new work to AWS Transform. | First-party report |
| AnthropicClaude Code and Claude TagSoftware and AI | Support coding, review, and incident response with shared instructions and checks. | Product with internal useCommercial products used throughout Anthropic's own engineering work. | People set review, release, and incident permissions according to risk. | MixedClaude products are public; Anthropic's internal Tag and integrations are private. | First-party report |
| BlockBuilderbot, Goose, and BuzzFinance and trading | Turn Slack requests and tickets into tested code changes for review. | Internal operationBuilderbot is used internally; Buzz remains in early internal testing. | People steer work, review changes, and approve consequential actions. | MixedGoose, Builderbot, and Buzz have public repositories; internal services differ. | First-party reportObserved artifact |
| Capital OneContext Specs, VulnHunter, and DataAgentsFinance and trading | Plan changes, repair security findings, and generate data rules with different checks for each task. | Single-task production lineInternal security and data projects are reported alongside a public planning framework. | Engineers review repairs and validate domain rules; agreement among agents is only one check. | MixedContext Specs and VulnHunter are public; the bank's underlying systems and DataAgents project are internal. | First-party reportObserved artifact |
| CloudflareInternal AI engineering stackSoftware and AI | Manage agent access, shared information, tools, spending, and code review. | Internal operationCompany-wide internal system across thousands of repositories. | People approve changes; automated rules can block serious findings. | InternalThe internal system uses some services available as public products. | First-party report |
| CognitionDevinSoftware and AI | Turn bug reports, design checks, and routine tasks into proposed code changes. | Product with internal useCognition uses Devin to develop its own product. | People write task instructions, review changes, and control releases. | CommercialDevin is commercial; the internal code and operating data are private. | First-party report |
| CursorCloud agents and security agentsSoftware and AI | Prepare code changes in parallel with evidence for reviewers. | Product with internal useCommercial agents plus internal review and security workflows. | People assess findings, review changes, and decide what ships. | MixedCursor is commercial; a security-automation reference implementation is public. | First-party reportObserved artifact |
| DropboxNova and DeflakerSoftware and AI | Run coding sessions and repeated repair tasks through one shared service. | Internal operationInternal use for failed builds, intermittent test failures, and migrations. | Task owners define checks; workflow code controls publication and retry limits. | Not publicly availableNova and Deflaker are internal systems described in a public engineering article. | First-party report |
| GitHubCopilot coding agentSoftware and AI | Turn assigned issues into draft code changes for review on GitHub. | Product with internal useCommercial product used in the repository that builds github.com. | Another person approves the work; existing code and test rules still apply. | CommercialCopilot coding agent is a commercial GitHub capability. | First-party report |
| ItaúDevinFinance and trading | Connect codebase knowledge with repeated software maintenance and review. | Internal operationBroad adoption reported in a vendor-hosted customer account. | Engineers review patches and generated tests. | CommercialDevin is commercial; bank systems remain private. | First-party report |
| Jane StreetAIDEFinance and trading | Build internal tools with agent-readable tests and a custom coding environment. | Internal operationInternal use is documented; adoption and concurrent session counts are undisclosed. | Engineers assess behavior the tests do not capture; release authority is not publicly detailed. | MixedAIDE is internal; Bonsai_term and example applications are public. | First-party report |
| LinearCoding SessionsSoftware and AI | Assign and review coding-agent tasks from issues and team chat. | Product with internal useCommercial product with reported internal design and issue-triage use. | People inspect changes and test evidence, then decide whether to merge. | CommercialCoding Sessions is a commercial Linear capability; the implementation is private. | First-party report |
| LinkedInCAPTMedia and networks | Give coding agents reusable company instructions and approved internal tools. | Internal operationInternal system reported in use by more than 1,000 engineers. | Specialists maintain the instructions; engineers remain responsible for results. | Not publicly availableCAPT is internal; it connects tools through the open Model Context Protocol. | First-party report |
| MetaCapacity Efficiency agentsMedia and networks | Investigate performance problems and propose code changes for review. | Internal operationInternal workflow for performance problems and improvement opportunities. | Engineers review, deploy, and remain responsible for the proposed fix. | Not publicly availableThe agent platform and its skills are internal. | First-party report |
| OpenAICodex and SymphonySoftware and AI | Take tasks from a project board, work in isolation, and review the results. | Internal operationInternal teams plus one fully agent-written experimental product repository. | People set priorities and acceptance rules; agents can review each other's changes. | MixedCodex CLI and Symphony are public; internal repositories and operating data are not. | First-party reportObserved artifact |
| RampInspectFinance and trading | Run coding tasks remotely with company information and prepared workspaces. | Internal operationInternal platform supporting several specialized engineering workflows. | Agents prepare changes; people review and merge them. | Not publicly availableInspect is internal; an independent inspired implementation is public. | First-party reportObserved artifact |
| ReplitInternal agent fleetsSoftware and AI | Coordinate parallel tasks across company systems and check the results. | Product with internal useCommercial product plus an internal system where agents coordinate other agents. | People set policy, review changes according to risk, and judge product outcomes. | CommercialReplit Agent is public and commercial; internal orchestration and data are private. | First-party report |
| ShopifyWorld, River, and AquiferCommerce | Work with employees in shared Slack threads and preserve task history across interruptions. | Internal operationRiver operates internally; Aquifer was being introduced beneath it in May 2026. | Employees join company-visible threads, redirect work, and review proposed changes. | MixedRoast is open source; River, Aquifer, and Dispatch are internal. | First-party report |
| SpotifyHonk and FleetshiftMedia and networks | Maintain code and migrate data across many repositories. | Internal operationInternal agent workflows use Spotify's existing tools for maintaining many services. | People start tasks, review and merge changes, and handle unresolved cases. | MixedHonk is internal; Fleetshift is available as a managed Backstage product. | First-party report |
| StripeMinionsFinance and trading | Complete coding tasks in the background and return checked changes for review. | Internal operationInternal system processing a large volume of background coding tasks. | The initiator inspects the result and another engineer reviews the pull request. | Not publicly availableMinions and its Stripe-specific infrastructure are internal. | First-party report |
| TempoCustom AgentsSoftware and AI | Turn prioritized product issues into parallel tasks and proposed changes. | Early rolloutInternal-use examples are public; company-wide scale is not stated. | People prioritize tasks and review proposed code or media edits. | CommercialTempo's agent products are commercial; internal operating details are limited. | First-party report |
| UberSoftware FactoryTransport and travel | Coordinate coding tasks, share relevant information, check results, and track cost. | Internal operationInternal platform spanning several stages of software development. | People choose useful work, assess quality, accept changes, and control release. | Not publicly availableUber's factory infrastructure is internal. | First-party report |
| VerceleveSoftware and AI | Run agents with saved progress, isolated workspaces, access rules, and checks. | Internal operationMore than 100 deployed internal agents reported; simultaneous worker count is unstated. | Engineers set checks and release controls and remain responsible for results. | MixedThe eve framework is open source; Vercel's internal agents and connections are private. | First-party reportObserved artifact |
LLM migration pipeline · Transport and travel
First-party report
Q Developer code transformation · Commerce
First-party report
Claude Code and Claude Tag · Software and AI
First-party report
Builderbot, Goose, and Buzz · Finance and trading
First-party reportObserved artifact
Context Specs, VulnHunter, and DataAgents · Finance and trading
First-party reportObserved artifact
Internal AI engineering stack · Software and AI
First-party report
Devin · Software and AI
First-party report
Cloud agents and security agents · Software and AI
First-party reportObserved artifact
Nova and Deflaker · Software and AI
First-party report
Copilot coding agent · Software and AI
First-party report
Devin · Finance and trading
First-party report
AIDE · Finance and trading
First-party report
Coding Sessions · Software and AI
First-party report
CAPT · Media and networks
First-party report
Capacity Efficiency agents · Media and networks
First-party report
Codex and Symphony · Software and AI
First-party reportObserved artifact
Inspect · Finance and trading
First-party reportObserved artifact
Internal agent fleets · Software and AI
First-party report
World, River, and Aquifer · Commerce
First-party report
Honk and Fleetshift · Media and networks
First-party report
Minions · Finance and trading
First-party report
Custom Agents · Software and AI
First-party report
Software Factory · Transport and travel
First-party report
eve · Software and AI
First-party reportObserved artifact