# Operating Rhythm

> A meeting-cadence planner for growing companies, at https://operating-rhythm.skillsafe.ai/. List up to twelve recurring meetings (or import them from a calendar .ics export) and the company's headcount. The page puts them on one week grid, gives each meeting a call from the five-level meeting architecture (keep, tighten, redesign, make it async, drop), finds the levels the headcount needs but lacks, and adds up the person-hours a week and the hours that come back. One metered run writes the fix and a timed agenda for every meeting and a decider for every slow decision.

URL: https://operating-rhythm.skillsafe.ai/
API tutorial: https://operating-rhythm.skillsafe.ai/api.html

## What it does

- Inputs: the headcount, an optional cost of a person-hour and currency, up to twelve meetings (name, level, how often, weekday, start time, length in minutes, people, whether it makes decisions, whether it is mostly status, whether it has a named decider, whether you attend) and up to six slow decisions. Meetings can be imported from a calendar .ics file, read in the browser.
- Free, in the browser: the size band (under 30, 30 to 100, 100 to 300, 300 or more) and the levels it needs; a call per meeting (drop when the level is not needed at this size; make it async when it rarely decides and is mostly status; redesign when it rarely decides, or decides only sometimes and is mostly status; tighten when it is over its level's length or people mark, has no named decider or decides only sometimes; keep otherwise); the length and people after each call; person-hours a week, your own hours before and after, and the hours that come back, with cost when given; clashes among your meetings; a verdict (overloaded, needs work, sound); a week grid with a quarter strip as SVG/PNG; the redesigned rhythm as an .ics calendar; Markdown, CSV and JSON downloads.
- Metered (one run on gpt-terra): a headline, a fix per meeting with a timed agenda whose minutes add up to the meeting's new length, a proposal for each missing level, a type (hard or easy to reverse), decider and speed for each slow decision, the first move and when to look again. The run never changes the calls, the lengths, the hours or the verdict.

## Contract

POST https://api.skillsafe.ai/v1/app-api/run with string fields: task = "rhythm", company, meetings (the names, in list order), decisions (or "none"), facts (the verdict, the company and load lines, one line per meeting, gap, clash and slow decision). The reply is one JSON object: {task, verdict, headline, meetings: [{id, call, fix, agenda: [{min, item}]}], gaps: [{level, add}], decisions: [{id, type, decider, why}], first_move, checkpoint}.

## Source

Built on the SkillSafe skill @github/gtm-operating-cadence (https://skillsafe.ai/skill/@github/gtm-operating-cadence, by Smit Patel, MIT): the five-level meeting architecture, the levels a company needs at each size, the rule that every meeting produces decisions or is cancelled, and Type 1 / Type 2 decision speed.