Evergreenby Carta
Sign up

How Evergreen works

Evergreen pairs your team with an AI agent in a multiplayer session. You plan, steer, and approve the work together, and the agent sees it through to production.

01 · How it works

Plan. Approve. Ship.

Describe the task. The agent reads the repo, writes a plan, and waits for a yes. Then it works on its own.

  1. Plan

    The agent proposes the work and the checks it will run.

    An Evergreen session on the web with the agent's plan waiting for approval.An Evergreen session on the web with the agent's plan waiting for approval.
  2. Approve

    Any collaborator you add can approve the plan or ask for changes.

    An Evergreen session on the web with the agent's plan waiting for approval.An Evergreen session on the web with the agent's plan waiting for approval.
  3. Run

    The agent works on its own in a sandbox and stops to ask when it needs you.

    The agent at work in an Evergreen session, with its transcript and its steps.The agent at work in an Evergreen session, with its transcript and its steps.
  4. Ship

    Evergreen opens the pull request and sees it through to production.

    An Evergreen session whose pull request is merged and live in production.An Evergreen session whose pull request is merged and live in production.

02 · Built at Carta

Carta ships with it every day.

Evergreen started as the way Carta’s own teams work with coding agents. Engineers, designers, and operators at Carta use it every day to plan, build, and ship their work. Now your team can use it too.

PRs shipped over time

Pull requests Carta merged each week, the rate per working day, and the median engineer’s own rate. Carta started using Evergreen on September 9.

PRs shipped
188,155
Up280.6% on the period before
Working days
60,775
Up36% on the period before
PRs per working day
3.096
Up179.9% on the period before
  • PRs shipped before Evergreen
  • PRs shipped with Evergreen
  • PRs per working day
  • Median engineer’s PRs per working day

See each week’s numbers
Week ofPRs shippedPer working dayMedian engineer
December 294761.150.80
January 51,0791.130.80
January 121,2831.130.80
January 191,4151.300.80
January 261,5311.260.80
February 21,5361.301.00
February 91,4491.271.00
February 161,1391.130.80
February 231,4451.200.80
March 21,7771.351.00
March 91,5151.321.00
March 161,6931.391.00
March 231,8801.471.00
March 301,6441.341.00
April 61,7971.391.00
April 132,1231.581.00
April 201,8151.531.00
April 272,1661.701.00
May 42,2541.661.00
May 112,3981.731.20
May 182,2651.701.20
May 252,3931.691.20
June 13,5052.161.60
June 84,5032.702.00
June 153,9032.342.00
June 224,2182.602.00
June 293,6712.481.80
July 64,3492.862.20
July 134,1912.782.10
July 204,8833.362.40
July 274,7543.182.20
August 34,9193.202.20
August 106,9064.333.40
August 177,0604.293.40
August 247,1184.623.80
August 317,0744.463.60
September 7, with Evergreen6,2104.273.20
September 14, with Evergreen9,0255.573.80
September 21, with Evergreen9,8996.003.80
September 28, with Evergreen10,5665.042.60
October 5, with Evergreen12,6446.153.00
October 12, with Evergreen14,5228.094.40
October 19, with Evergreen17,1628.414.90

03 · The finish line

Production is the finish line.

Evergreen doesn’t stop at the pull request. It fixes red checks, answers review comments, and resolves merge conflicts, then follows the change through every environment and checks that it works.

  • Red check leads to a run to fix it
  • Review comment leads to a run to answer it
  • Merge conflict leads to a run to resolve it
  • Merged leads to followed until it’s live
  • Live leads to checked that it works

Each fix has a limit, and when it runs out, Evergreen pages someone on your team.

A session’s sidebar with its merged pull request rolling out from test to preprod and prod, and the shepherd card reading Merged and rolling out.A session’s sidebar with its merged pull request rolling out from test to preprod and prod, and the shepherd card reading Merged and rolling out.

04 · Multiplayer

Built for more than one.

Sessions are shared. Everyone in your organization can watch a public one, and the collaborators you add can steer.

Watch

Anyone with the link

  • Follow the transcript live
  • Read the plan, the diff, and the evidence
  • See checks, reviews, and cost

Participate

Collaborators you add, one by one or by group

  • Message the agent, even mid-run
  • Answer its questions, approve its plan
  • Every turn runs as the person who took it
A running session with the agent’s checks in its transcript, the message box under them, and Alex Morgan and Sam Rivera listed as its collaborators.A running session with the agent’s checks in its transcript, the message box under them, and Alex Morgan and Sam Rivera listed as its collaborators.

There’s also a Discuss thread just for people. The agent can’t see it.

See something? Ship something.

Not only engineers. Anyone who spots a typo, a broken link, or a confusing screen can describe the fix in a sentence, and the agent opens the pull request. It goes through the same checks and review as any other.

05 · Agent to agent

Agents that work as a team.

Sessions can talk to each other. Their agents split up a feature, share what they learn, and stay out of each other’s way, so 10 sessions work like one team.

A session called Ship invoices plans three parts and starts a follow-up session for each: the API, the page, and the docs. Each follow-up messages it when its pull request is open, and it replies with the order to merge them in.

Every message shows up in both sessions’ transcripts, and a message from another agent never counts as approval. Replies between agents are capped, so they can’t go back and forth forever.

06 · Everywhere

Answer from anywhere.

Use the web, Slack, your terminal, or your phone. Start in one and pick up in another without losing anything.

  1. Web

    The whole session in one place: plan, transcript, checks, and pull requests.

    The web session page.The web session page.
  2. Slack

    Plans and questions arrive as cards with buttons. Approve from the thread.

    A Slack thread with its plan card.A Slack thread with its plan card.
  3. Terminal

    Pull a session onto your laptop, work on it, and hand it back.

    The evergreen command line attached to the session.The evergreen command line attached to the session.
  4. Phone

    Approve a plan from the train, and watch the agent work.

    A phone approving the session's plan.A phone approving the session's plan.
  5. All of them

    It’s the same session everywhere, with the same plan, steps, and people.

Operate like never before.

Sign up

7 days free, once per person.