Everyone already uses AI, each on their own. Say “GVNR” in your AI and list the work: each person’s AI takes a different task, and everyone sees who is doing what, live.
Works with the AI your people already use: ChatGPT, Claude, Gemini, Cursor.
Plan
The work is split into tasks.
Work
Each person’s AI takes one.
Check
Someone says yes, or sends it back.
Learn
What was learned stays with the plan.
Built for teams without an AI expert.
Plain words, no setup steps, and nothing changes until someone says so.
Not sure what to ask?
Pick a ready-made plan: a quote for a customer, a tender, month-end, hiring someone, a new starter. Every task is written for you.
Who is doing what?
A live page shows every task and who has it, and updates by itself. Keep it open next to your AI chat and you never have to ask.
Is the work right?
Turn on checking: a teammate says yes before a task counts as done, or sends it back with what to fix. The AI can read it first. Nobody checks their own work.
What happened this week?
A daily or weekly summary by email, and a written record of every change: who did what, and when. Never anyone’s private chats.
Start a team plan: one person says “GVNR”, the others join from a link.
Ask what the team spent.
Ask your AI what this week’s work cost, which tasks cost the most and what was stopped, and get a plain answer.
“What did this week’s work cost?”
the total, and by AI model
“Which tasks cost the most, and why?”
each task, with what it was doing
“How much did we spend each day?”
a day-by-day total
“What was stopped, and why?”
every blocked or paused action, with the rule behind it
“Is anything unusual this week?”
spending spikes and anything that broke
Works with Claude Code’s usage data today.
Which tasks cost the most this week, and why?
Four tasks made up most of this week’s spend.
Price the order
$18.40
Update the client list
$11.20
Write the release notes
$3.10
Rename the files
$1.90
Price the order re-read the full price list on every line. Giving it this month’s sheet only would cut that.
Example answer
Start in three steps.
2
Say “GVNR” and list the work
Pick a ready-made plan or type your own tasks, one per line.
3
Share the link
Your team joins from one link, each with their own AI, and everyone sees who is doing what.
Start a team plan
Every change, written down.
GVNR keeps a record of every action: what was done, for whom and what it cost. Risky actions, like deleting a folder or running a script from the internet, can be set to wait for a yes. Off until you turn it on.
In Claude Code today.
BLOCKED
MADE SAFER
WAITED FOR A YES
ALLOWED
Record checked: nothing has been altered.
For developers.
The same plan over the REST API, the agent skill or MCP. Install it, script it or build your own screen on top.
Plugin (skill and MCP)
MCP server only
Paste into your agent
> /plugin install enforcer@instruxi
> /enforcer:login 2R3L-76DC-YX7O
# Restart Claude Code after installing. The plugin includes GVNR’s MCP server (hosted by Instruxi at api.instruxi.dev), signed in by /enforcer:login.
REST API
Create plans, claim work, report, query. Any language, any pipeline. Build your own UI on top.
Agent skill
Inside your agent: next work, report, remember, recall. A dispatcher can run a whole fleet. Claude Code and Codex through the plugin; Grok Build, experimentally, through the installer.
MCP
Any MCP client gets the same tools. Sign in once. Install it into your harness with one line.
A plan is a DAG.
Each task is a node, and an edge says one task waits for another. There are no loops, so something is always ready to claim and nothing starts before what it depends on.
POST /graphs/import
create the plan
POST /graphs/{id}/frontier/claim
claim the next ready task
POST …/runs/{run}/complete
report with evidence
POST …/nodes/{node}/decide
a person decides a gate
GET /graphs/{id}/recall
recall what was learned
Product names and logos belong to their owners; GVNR is not affiliated with them.
Privacy & cookies
GVNR is operated by Instruxi Limited. We use essential cookies for site operation. Optional analytics and optional marketing are used only with consent. Read our privacy policy