Looply
Keep control of what ships — even as AI writes most of it.
Looply is the shared workflow for people and AI agents. Your team sets the standard once; every result comes back checked against it, with the evidence to accept it or send it back. Starting with software teams.
Free for individuals · December 2026 Team $30/seat/mo · March 2027 Enterprise · custom
01 · Why now
AI writes faster than anyone can check. Attention is the bottleneck now.
Run several at once and assumptions pass silently between them, parallel tasks collide, and decisions get lost between sessions.
02 · How it works
Set the standard once. Looply carries it across every handoff.
Agree
Connect your repo. Looply proposes a standard from your code, tests, and docs — your team confirms what “good” means and who’s accountable.
Work
People and agents do the work. Looply carries the intent, decisions, and context across every handoff — nothing gets re-explained.
Judge
Every result comes back checked against the standard, with evidence: what passed, what needs a fix, what needs your call.
Improve
Looply learns from every result you accept or send back, so the next task starts closer to what your team meant.
03 · Where we’re going
Not another AI tool — the workflow your team and its agents run inside.
New ways of working need new tools. Agile gave teams Jira; human–agent teams need a workflow of their own. We start with software — the work agents already do best — and grow into everything else they’ll do by writing code.
The shared workflow for AI engineering
Standards, memory, and evidence across every person and agent shipping code.
Any work agents can do
The same workflow for work agents do by writing code — orders, features, revenue analysis — with a person judging the result.
Runtime-neutral, and it compounds
Works with the models, agents, and IDEs you already run. The standard and the memory belong to your team — and sharpen with every result.
Read the full vision — why now, the market, the trajectory →
A principle, not a stopgap
A person is always accountable for the result.
People set the intent and the standard. Agents coordinate and do the work. The person accountable accepts the result — with the evidence to stand behind it. That line doesn’t move as AI does more.
04 · Who we are
We’ve spent our careers on one question.
How do you trust what an AI system produces? Looply’s founders have chased that question from AR hardware and global e-commerce to self-driving perception — most recently building Genia, a generative-AI platform for professional engineers, where they lived this exact problem.
Zhihao Zhao
Founding head of product at XREAL · grew Amazon Global Store 3× in GMV · founded Genia (a16z-backed).
Has spent his career making new technology fit the way people already work.
Houtao Wang
Built a 300+ person engineering org · scaled Haoqiao to $500M GMV · CTO at Genia.
Ran large engineering orgs where standards slipped the moment AI sped everyone up.
Mukun Guo
Safety-critical perception at Waymo & Black Sesame · Genia’s first employee, VP of AI.
Comes from autonomous driving, where you must state in advance what an output has to satisfy.
FAQ
Questions, answered.
What is Looply?
Looply is the shared workflow for people and AI agents. Your team sets its standard once; that standard and the decisions behind it travel with the work across every handoff; and each result comes back checked against it — with the evidence for the accountable leader to accept it or send it back. It starts with software teams.
What problem does Looply solve?
AI produces work faster than people can review it, and teams now run several agents at once. Assumptions pass silently between them, parallel tasks collide, and decisions get lost between sessions — so leaders either slow down or approve work they haven't really checked. Looply gives people and agents one shared standard, one memory, and a record of evidence to work from.
How does Looply work?
Connect a repository and Looply proposes a standard from your code, tests, and docs; your team confirms it. From then on, people and agents do the work while Looply carries the context across every handoff and checks each result against the standard — passed, needs a fix, or needs a human's call — learning from what you accept or send back.
How is Looply different from the review in AI coding tools, or planning tools like Linear?
Coding tools review their own output — the author grading its own work — and planning tools coordinate people but not agents. Looply is the layer above both: independent of whatever wrote the code, one standard your whole team owns, shared memory across people and agents, and evidence instead of a scroll of comments. And it improves how your team works over time, not just the task in front of it.
Is Looply free?
Free for individual developers at launch in December 2026. The Team plan is $30 per active engineer per month from March 2027 — team standards, shared memory, coordinated assignments, and evidence. Enterprise is custom, adding customer-hosted models and administration across teams.
What happens to my code and data?
Looply reads only the repositories, standards, and context you connect. Bring your own runtime and the model call happens in your provider account; with Enterprise, models run inside your own infrastructure, so nothing reaches a third party.
Who's accountable for the result?
A person, always. People set the intent and the standard; agents coordinate and do the work; the accountable leader accepts the result — on payments, on core systems, everywhere — with the evidence to stand behind it.
Join the waitlist
If this is how you want your team to work with agents, come build it with us.
Free for individuals at launch, per-seat for teams from March 2027. We want the people who share the conviction early.
launching December 2026 · looplydev.com