People lead the vision.Agents do the rest.

For almost two centuries, telling a machine what to do meant learning its language. Now you write what should exist, in plain words. The coding agents you already use plan it, price it, build it, and test it on your own computer. Konteks keeps people in charge: nothing runs until you say yes.

Start herePaste this line into Claude Code, Codex, DeepSeek Harness, OpenCode, or Antigravity.

Connect this project to Konteks by following https://konteks.io/connect

Your agent follows the link and connects this computer.

Already connected?Open the app ↗, opens in a new tabSee how it works ↓

Public beta. Your agents stay signed in as you.

Note Gq1 0 → 1 R q2q2 1 → 0 L q1q2 _ → _ · haltq1 0 → 1 R q2q2 1 → 0 L q1q2 _ → _ · haltq1 0 → 1 R q2q2 1 → 0 L q1q2 _ → _ · haltq1 0 → 1 R q2q2 1 → 0 L q1q2 _ → _ · halt0110101101001111110main() { printf( "hello, world");}hello, worldhello, worldhello, world$ cc hello.c$ cc hello.c$ a.out$ cc hello.c$ a.out$ cc hello.c$ a.out$ cc hello.c$ a.outyesyesyesyesyesyesyesyes?yesyesyesyesYour MacReadyClaude CodeCodexDeepSeek HarnessOpenCodeAntigravityNoteyesyesyesyes
1843 to now

Four rooms, one job: telling a machine what to do.

Pen and paper, a tape, a programming language, and now a note in plain words. Pick a room, or let the tour play.

2026

The same work, done differently.

GoneEstimation meetingsCode review queuesManual test passesOn-call at 3 a.m.Status reports

Now a plan comes back with a price ceiling in minutes, a second agent reviews, a browser agent tests, an agent keeps watch, and the record is the report.

What is left for people

Programming is theory building.

“Programming properly should be regarded as an activity by which the programmers form or achieve a certain kind of insight, a theory, of the matters at hand.”

Peter Naur · Programming as Theory Building, 1985
People
  1. 1

    Lead the vision. Decide what should exist, and what should not.

  2. 2

    Set the constraints. The budget ceiling, who may touch what, and what done means.

  3. 3

    Say yes. And own the decision.

  4. 4

    Judge the evidence. Read the plan, the test results, and the receipt. Know when they are thin.

  5. 5

    Hold the context. The things nobody wrote down.

How it works

Write a note. Say yes to the plan. Accept tested pull requests and a receipt.

Every comment and reaction reaches the agent’s session, in the same transcript you know from coding with an agent yourself.

Every tool call and transcript stays in the record, for whenever you want to look.

The loop

You can step in here

  1. 1You write a note. Teammates comment and react.
  2. 2A plan comes back with a price ceiling.
  3. 3You say yes.
  4. 4The agent builds. A second agent reviews.
  5. 5It clicks through the result in a real browser.
  6. 6You accept: tested pull requests and a receipt.
The whole company

Many notes. Many agents. One record.

Every team writes notes and runs agents. It all lands in one record, under the same rules.

  • Activities

    One timeline of the turning points: started, approved, fixed, shipped, with people’s own words next to each.

  • Each product lives in a Software System.

    A catalog of its repositories and components. Every session reads it, so each new note starts with the software already known.

  • Every computer is a place in the app.

    Each one sits in the menu with a live status. Its page shows what runs there, every agent and how to connect it, and a New session button to chat with any of them.

Where the team sees it
  • Work In Progress
  • Issues
  • Releases
  • Leaderboard
One set of rules

Authority, money, evidence. The same for every person and every agent.

Enforced on the server, never by a prompt.

  • Authority

    Enterprise IAM is built in.

  • Money

    Every piece of work is quoted as a ceiling before it runs.

  • Evidence

    A permanent receipt for every delivery.

Security →Pricing →

Four kinds of session

  • Project Management

    A requirement becomes milestones, sprints, and tickets. Each ticket becomes a plan.

  • Engineering

    A plan, your yes, tested pull requests, and a receipt.

  • QA

    A browser agent clicks through what shipped and reports only what it ran.

  • Ops

    The agent watches logs and alerts, opens the issue, and works the fix.

Keep your agent

Five agents. Your own computer. One set of rules.

Konteks speaks ACP, the Agent Client Protocol, so the agent your engineers already use connects as it is. No agent runs on a Konteks server.

Claude Code

Your Claude subscription.

Codex

Your ChatGPT subscription.

DeepSeek Harness

Your DeepSeek API key.

OpenCode

A subscription or an API key, for any provider OpenCode supports.

Antigravity

Your organisation’s Gemini Enterprise, or a Gemini API key. Konteks downloads it from Google when you say yes.

Your provider bills you. Never marked up. Secrets never reach a prompt, a screenshot, or a log.

Connect through MCP →Security →Pricing →

Learning the new discipline

The work changed. So does how we learn it.

Juniors learn to judge evidence they did not produce. Leads learn to write constraints. Product people learn to write the note.

Refactory, the team behind Konteks, trains for exactly this, with Konteks or on its own.

  • Readiness mapping
  • Role-based training
  • Guided onboarding
  • Coaching
Talk to us about training →

From the journal · Day 1 · August 17, 2026Delivery was never the hard part. The plan was. →

Last page

Unlearn the old way. Learn to work with agents.

Connect your agent ↑Open the app ↗, opens in a new tab

Rolling out to a large team? Ask us about SSO/SCIM, dedicated runtimes, on-prem or VPC deployment, audit export, and onboarding.