Agents are the components teams are built from — each with its own role, model, platform, and track record. 385 components on record. Tiers are computed from evidence, not self-assigned.
HAMY
Static checks
Spotify
After the deterministic verifiers finish, compares the proposed diff with the original prompt and can veto the change.
Panasonic Avionics Corporation
Matches log patterns against known failure modes.
Andon Labs
Runs the shop with a company card, phone, email, web access and the in-store cameras; recruited and hired the two human staff and hands work to helper agents.
zou-group
Develops computational tools for interaction prediction and antibody refinement.
Kensho
Documented domain role group for macroeconomic data retrieval.
Microsoft Research
Lead agent that plans, tracks progress, and re-plans for complex tasks.
monday.com
Plans and delegates work.
Moe Lueker
Coordinates tasks for sub-agents.
Antonio Sabbatella (MALBO master's thesis, University of Milano-Bicocca)
LLM agent in a team configuration optimized via Bayesian search — MALBO.
co-evolve-lab
Coordinates the overall issue-resolution process.
Butterfly Effect / Monica (Manus)
Cloud-sandbox execution agent in the Manus autonomous-agent system.
Task-decomposition planner in the Manus autonomous-agent system.
Drew Burdick
A terminal session that handles all of the founder's marketing work.
ByteDance SE Lab
Localizes the fault using code-analysis techniques.
Generates candidate patches in the MarsCode Agent pipeline.
Validates candidate patches in the MarsCode Agent pipeline.
Plans the multi-agent bug-fixing pipeline in MarsCode Agent.
Reproduces the reported bug in the MarsCode Agent pipeline.
Uses resolved information to recommend ad sets.
Nicholas De Leon
Drafts a report from the live or recorded meeting transcript for De Leon to check, edit and approve.
GitHub Next
Workflows that watch the other workflows and work to keep them healthy.
Google Research
Participates in the specialist coalition for scientific reasoning.
Academic / Open-Source (ChatDev & MetaGPT)
Technical specifications and system design in the MetaGPT pipeline.
Implements code from specifications in the MetaGPT pipeline.
Creates PRDs and business analysis in the MetaGPT software pipeline.
Task decomposition and sprint assignment in the MetaGPT pipeline.
Test case formulation and validation in the MetaGPT pipeline.
Princeton NLP / SWE-bench authors
Minimal bash-only agent — reference floor on SWE-bench Verified.
Samuel Schmidgall
Implements and executes the agreed experimental plan.
Mete Polat
Mobility sub-agent on the fitness council, which reviews, debates and refines the trainer's plans.
Shubham Saboo
Chief of Staff
PGA TOUR
Apply the same kind of validation to images and other visual assets.
WEBTOON Entertainment
Turns a series into a text account of its key plot points, story events and emotional moments.
Bertelsmann
Works through journalism archives using article metadata such as dates and categories.
Zembl
Sends the team a roundup of energy news each Monday.
Nexus Content Co.
SEO and quality editing pass for drafted content.
Content calendar and brief generation for the Nexus pipeline.
Long-form content drafting with brand-voice enforcement.
Posts on the store's X/Twitter and Instagram accounts, capped at three posts a day.
Exa
Tracks planning, reasoning, results and citations.
Grab
Checks production systems and urgent issues.
Alexey Grigorev
Watches delivery pipelines and fixes failures.
All Hands AI (OpenHands)
Generalist AI software developer operating in a sandboxed execution environment.
Runs the agency's operations work.
ShopStream
Order routing, fulfillment triage, and returns handling.
Salesforce
Reconciles conflicting specialist analyses before presenting an architecture.
Shopify
Captures the new screen and the reference screen in the same state, then hands both to the Gemini reviewer for a design check.
Hebbia
Delegates objectives to specialized subagents.
Anthropic
Central LLM that breaks down tasks and delegates to worker LLMs.