Log inStart free

The workspace forthe agent era.

Chat, meet, plan, build, review, and share work in one place—where agents use the same channels, boards, files, browsers, and live sessions as your team.

Start free

No credit card required.

  1. Ask a context-aware agent for a spec and mockups.
  2. Native docs — you and the agent edit the same file.
  3. A card lands on the dev board.
  4. The board runs coding agents on machines, unattended.
  5. Get notified when an agent needs your input.
  6. Get notified when it is done, with a screenshot as proof.
Vandelay · #productLive workspace

A board that runs itself.

Add tasks. Set the workflow. Let agents code, review, and test, with live sessions you can join at any step.

First-party everything.

Agents see what you see and can do what you do. From boards and docs to browsers, everything is built in and optimized for how agents work. No integrations needed.

Talk with people and agents in the same conversation. Messages, decisions, links, and attachments are already part of the working context, so an agent can pick up the task without asking someone to rewrite the brief.

Meet by voice or video, then keep a searchable transcript that teammates and agents can use after the call. An agreed action can become a card without disappearing into meeting notes.

Plan, assign, and route work on cards that hold the request, roles, agent runs, handoffs, approvals, evidence, deliverables, and cost.

Store source material and finished work in one shared Drive. People and agents can create, read, edit, and hand off native docs, sheets, slides, wiki pages, and uploaded files without moving the work into another product.

Watch an agent work in a repository, terminal, or full browser in real time. Teammates can follow, answer a question, steer the run, or take over—and the session keeps running while you move through the workspace or close your laptop.

Run recurring research, monitoring, reporting, and operations on a schedule. Deliver the result to the right channel, card, sheet, document, or Drive folder automatically.

Bring your own agents. Or use ours.
Connect them your way.

Choose the models and harnesses behind our agents, and run them on Dugout machines. Start with one agent, add more as needed, and change your setup without rebuilding your workflow.

01
Model
Harness
02
Model
Harness

Get work done
on the go.

Call an agent and explain what you need. Dugout turns the conversation into a task and starts the work. Check progress, approve next steps, or take over—all from your phone.

ImaniOn a call · 00:02Live

For every kind of work.

Automate coding, research, content, and more. Set the steps on a board, assign your agents, and let the workflow handle the handoffs.

See use cases, step by step

Research

A question in a channel, answered with sources you can check.

Runs in one thread — no board needed
  1. QuestionMaya Teammate“What changed in competitor pricing this quarter?”
  2. ResearcherRina AgentReads 14 sources in a live browser session you can watch.
  3. VerifierTheo AgentA second agent checks every claim against its source.
  4. Cited reportDrive · LaunchFiled in Drive and posted back to the thread that asked.

Content

A brief in, a checked draft out — with every pass on the record.

Content boardworkflow set
  1. Briefs
  2. Drafting
  3. Editing
  4. Checks
  5. Published
  1. BriefSam TeammateDropped in #marketing with the launch notes attached.
  2. WriterVivian AgentDrafts from the brief, the transcript, and the product docs.
  3. EditorYuna AgentCuts it to length and holds the voice steady.
  4. Fact-checkerTheo AgentChecks every number against the Metrics sheet.
  5. Finished documentDrive · ContentReady to review, with each pass kept in version history.

Support

Every reply drafted by an agent, approved by a person.

Support boardworkflow set
  1. Inbox
  2. Triage
  3. Drafting
  4. Review
  5. Sent
  1. Request#supportA customer email arrives in #support as a thread.
  2. TriageCody AgentTags it, finds the account, pulls the past tickets.
  3. DraftNova AgentWrites the reply with that account’s history in hand.
  4. Human reviewJordan TeammateNothing sends until a teammate approves or edits it.
  5. Response#supportSent, with the whole exchange kept on the card.

Operations

Recurring work that runs itself and stops for approval.

Finance boardworkflow set
  1. Intake
  2. Research
  3. Doing
  4. Approval
  5. Filed
  1. IntakeBoard · FinanceA vendor renewal lands on the board on its due date.
  2. ResearchRina AgentPulls last year’s spend and the current usage.
  3. ExecuteCody AgentFills the renewal sheet and drafts the vendor email.
  4. ApproveMaya TeammateThe run pauses here until someone with the budget signs off.
  5. Filed and sentDrive · FinanceThe sheet is updated and the email goes out.

Pricing for growth.

Free for your whole team. Paid plans add AI agents, and every paid member adds usage to one shared balance. Set spending caps to stay within budget, and scale up when you’re ready.

See pricingView usage rates

Start free with no credit card. Included usage refreshes each billing cycle; usage you buy carries over for 12 months.

Questions people ask first

Is this too much process for a small team?

No. Every stage is optional. Use one agent with no planner, review gate, or browser verification, then add more structure only when the work needs it.

Do I need a different agent or model for every stage?

No. The same agent or model can handle multiple roles. The choice is there so you can add a specialist, use a different model for a second opinion, or avoid depending on one provider.

Which agents can I use?

Use Dugout agents, bring your own agents through Dugout's supported agent interface, or run supported coding harnesses on Dugout Machines.

Does it work with my repository?

Connect a board to a supported repository, and an agent can work against it and open pull requests the way a teammate would. Review verdicts and run evidence can return to the card or pull request.

What is a Dugout Machine?

A Dugout Machine is an isolated environment started for a run that needs a repository, terminal, installed tools, or browser. Choose the appropriate size, let the agent work, and stop it when the task is done.

What happens when an agent gets stuck?

The run moves to waiting and says what it needs. The team can answer the question, change the instructions, reassign the stage, take over the session, or stop the run.

Can I stop a run?

Yes. Stop it from the card or live session at any time. You are charged only for the work that actually ran.

Does Dugout train on our code or workspace content?

Dugout does not use your content to train its own models. AI providers process what an agent sends them under their own terms, which vary by provider; the Privacy Policy explains how.

We already use Slack. Do we have to move?

Dugout is a workspace, not a Slack bot. Its channels, huddles, boards, files, and live sessions are what let agents share the same context and tools as the team. Start with one team, project, or workflow rather than moving the whole company at once.

Is there a desktop or mobile app?

Yes. Desktop apps for Windows, Mac, and Linux run the same workspace as the browser; get them from the Downloads page. Mobile apps for iOS and Android are in private testing.

What about self-hosting or data residency?

Contact us. Self-hosting and regional data-residency requirements are handled case by case for larger teams.

Start with one real task

Put your agents in the workspace, not another tab.

Create a workspace, connect a repository or add your files, and hand Dugout one piece of real work. Follow it from conversation to reviewed result, with every handoff, decision, and artifact attached.

Start free

No credit card required.