R II F + REFLEX OPS
People decide.
Machines act.
Every machine acts on your latest decision, and asks you only what is new.
Reflex Ops: people first
2:45Why work waits, why agents build on yesterday's decision, and how Reflex Ops keeps people in charge of their machines.
INSTALL REFLEX OPS
1. Open Terminal and paste:
$ curl -fsSL https://reflex.riif.com/install | bash2. In a new window, sign in with riif, once:
$ reflex loginMac, Linux and Windows. Free to install. Why this matters
WATCH AND READ
Study how it works.
Nothing acts on yesterday's decision
Every rule for AI says a person stays in control. At scale that breaks two ways. A story about an eighth server, and why we named it Reflex Ops.
Read →
Your agents are building last week's plan
A requirement changes on a call. The people on the call know. The agents do not. How to say a change once and have the work that rests on it wait or ask.
Read →
Clicking yes is not oversight
Agents ask about everything, so we stop reading. Meanwhile the one question that matters waits all night. Answer once, and let the real question through.
Read →
Out of Claude Code usage by Wednesday?
Make Claude Code the director, not the typist. It writes the brief and checks the result; the long work runs in Reflex Ops under a dollar cap.
Read →
THE PROBLEM
In control, on paper.
Every rule written for AI says the same thing: a person stays in control. As the number of machines grows, that breaks in two directions.
Our decisions reach machines late.
We change our minds. A client says no. A date moves. Some machines hear it, and the rest keep working on a decision we already took back.
Our approvals become a reflex.
Machines ask so often that we stop reading. A person who clicks yes all day is in the loop in name only.
1 AUGUST 2012
The eighth server.
$460M+
lost
45 min
after the market opened
1 of 8
servers kept the old code
The trading firm Knight Capital decided to replace old code. A technician copied the new code to seven of its eight servers. The eighth kept the old code, and nobody knew. Seven machines heard the decision. One did not. That one was enough.
IT HAPPENS AT A DESK TOO
Two sessions. One person. Two versions shipped.
One person gives the same instruction in two AI sessions, in slightly different words. Two agents do good work, and ship on two versions of one decision. Now picture a team of ten, a hundred agents, and machines that keep working at night while the people who decided are asleep.
ILLUSTRATION
At noon an operator takes back a rule: do not spray the east field, people are working there. A drone loaded its plan at eleven. In software, acting on an old decision costs a redo. With a body, it can cost an injury.
THE OTHER HALF
93 out of 100.
People using one popular coding agent approve around 93 percent of the permission prompts it shows them. That is not carelessness. It is what anyone does with a question asked hundreds of times a day.
WHY WE CALL IT REFLEX
Your body already solved this.
Your body runs countless actions without asking you. Reflexes act inside limits your brain has set. When your intention changes, a movement already under way is stopped. You decide, and your whole body works from that decision until you change it.
01
Every job carries the decisions it rests on.
In the job, before it moves. Not in a log written afterwards.
02
A change reaches the work.
Work resting on the old decision waits or asks you before it goes on.
03
A machine asks only what is new.
Your answer becomes a rule, and no machine asks it again.
Fewer questions, every one of them real. Nothing acts on yesterday's decision.
TWO HALVES OF ONE WHOLE
R II F and Reflex Ops.
WHERE YOU DECIDE
R II F
Your rules, your decisions, the questions waiting for you, and your team's record of every job: who asked, which agent did it, what it rested on, and whether it was checked. On the web and on the go.
WHERE YOUR MACHINES ACT
Reflex Ops
Runs your team's AI agents on your own computer, with your own keys. A job can run under a spend cap and shows what it cost. Risky steps wait for your yes. Private folders stay away from AI that trains on what it reads. Works with Claude Code.
TODAY, AND NOT YET
What you can use today.
- An agent stops and asks you. You answer once, and you can make the answer a rule the next agent follows.
- Change a decision once. Work not yet started on the old one waits or asks you first. A person’s task already done is marked as done on an old decision: keep it or redo it.
- Your team’s record of every job, kept and sealed.
- Not yet: stopping a job that is already running when a decision changes. We are building it next. Today this works for AI agents doing software and office work. Robots and drones are where it matters most, and where we are going.
ONE LINE
Put it on your computer.
INSTALL REFLEX OPS
1. Open Terminal and paste:
$ curl -fsSL https://reflex.riif.com/install | bash2. In a new window, sign in with riif, once:
$ reflex login3. Start:
$ reflexUsing Claude Code? In your project, run reflex kit and Claude Code can hand work to Reflex Ops under a spend cap. All steps
Sources: Knight Capital: U.S. Securities and Exchange Commission, press release 2013-222 and order 34-70694 (2013). Approval prompts: The Register, 6 August 2026, reporting Anthropic's figure for Claude Code users.