Grid ConsoleALPHA MISSION CONTROL FOR CODING AGENTS Grid Console runs a fleet of AI coding agents across all your projects at once, and puts your attention where it is needed. Every card summarises itself: the plan, what changed, and the one thing it needs from you. You decide in seconds instead of reading a transcript, and nothing ships past a gate without you. The way of working is yours to shape: the stages, what each one runs, which agent does the work. Start free Join the waitlist macOS, Linux, Windows via WSL 2 · your repos never leave your machine · sign in with GitHub

THE BOARD · LIVE4 agents running

DOING

Rate limiter on public API▣ building · tests greenvertex

Nightly digest email▣ writing the templateatlas-docs

REVIEW

Stripe webhook retriesplan ready · 3 files, 2 risksbilling▶ APPROVE

Fix flaky auth testdone · diff is 41 linesvertex▶ REVIEW

DELIVER

Dark-mode settings pageapproved · committinghelios

one card is asking: everything else is running on its own

SEE IT, LIVEan interactive preview: it walks a card through every stage by itself, click in whenever you want to drive

free to poke, a preview of how Grid works

WHO IT HELPS

SOLO BUILDERSShip like a crew of fiveKick off five cards before breakfast. Agents plan and build in parallel while you approve plans and review results: one person, a whole roadmap.

TECH LEADSKeep up with AI-speed codeAgents produce more code than anyone can read raw. Grid explains every file, its goal, what changed, why, and which card did it, so review keeps pace.

TEAMS & NON-ENGINEERSHand the keys safelyDescribe work in plain words; Grid plans, asks, builds. Autonomy sits on one dial per project, so a PM can drive real work and engineers keep the controls.

HOW IT HELPSthree views, one habit: go where you’re needed. The app opens with a guided tour of all three.

BOARD Run the fleet The board is where you run the fleet. Every task is a card, an agent picks it up, and the card moves through your SDLC on its own. You keep track of many agents in parallel from one place. When a card needs you, it lights up and pulses: a plan to approve, a question to answer, a review to give. BEST WHENmaximum throughputseveral projects in flight

OPEN IT LIVE →

the board, running · not a screenshot

SESSION Work per card Click a card and its session opens. This is where you talk to the agent directly: steer it, correct it, ask it questions. The right pane tells you exactly what you need to know at that moment: the plan, the changes so far, and what the agent is asking. It saves you scrolling back and rereading the whole card. The rail on the left queues who needs you next, questions first. BEST WHENkeeping track of the detailsstepping in mid-runseveral cards at once

OPEN IT LIVE →

a session, running · not a screenshot

CODE Take control back Sometimes the details matter. The code view shows, card by card, exactly what changed, and lets you undo what should not have been. Code alone tells you what was written, not why or when. So the viewer pairs the source with AI explanation and card context: the goal of the file above line one, notes between the lines, and history that gives reasons. You always know who changed this line, and why. BEST WHENfixing a complex AI mistakecorrecting feature creeplearning an unfamiliar project

OPEN IT LIVE →

the code view, running · not a screenshot

Other great features

Command historyAgents run commands. This shows exactly what ran on your computer, and undoes it where undo is possible. You never reconstruct a shell command the hard way.WHAT RAN · UNDO

Usage insightGrid tracks tokens per card and per request. When usage spikes, you see which action caused it and act on it.TOKENS PER CARD

Plugins marketplace COMING SOONBundled plugins ship in the alpha, writing your own is free on every tier, and the marketplace opens with the public beta: change the default SDLC or add a stage, read cards from GitHub, Jira or Azure DevOps, connect another agent or model.YOUR SDLC · YOUR SOURCES

TRY IT The tour runs inside Grid on your own projects, so the first thing you see is your own board. Free while the alpha lasts, and it takes about twenty seconds. Download Grid

WHAT YOU HAND ITthe everyday work that stops interrupting yours

▣User storiesDrop the story in as written, acceptance criteria and all. The agent plans against the criteria, and Review checks them off box by box.DEFAULT CARD · STORY IN, PR OUT

▣Change requestsPaste the request from a customer or a colleague. The agent scopes it, names the risks, and asks before touching anything shared.DEFAULT CARD · SCOPED BEFORE BUILT

✕Support ticketsForward the ticket. The agent reproduces the complaint, traces it to its cause, and files the fix with the ticket linked.BUG CARD · TICKET, CAUSE, FIX

✕Bug triagePaste the report. The agent reproduces, bisects and proposes the fix; you approve at Review.BUG CARD · REPRODUCE, BISECT, FIX

?Decisions to researchAsk the question instead of the change: which mail provider, where a limiter belongs, why a page got slow. It reads the repo and reports back, and writes no code.RESEARCH CARD · NO CODE, A VERDICT

≣Work too big for one cardHand over the whole job. A masterplan splits it into subcards, runs them, keeps their status, and only asks you what the plan itself needs.MASTERPLAN CARD · SPLITS AND TRACKS

↻Migrations and refactorsLong, mechanical, dangerous. Worktree isolation and stage gates make them boring: the agent grinds, you approve.CHORE CARD · ISOLATED WORKTREE

↻Docs and test coverageCards update the docs they touched at Close. Point it at coverage gaps and it writes the specs, Verify runs them, and the diff shows what is covered.CHORE CARD · VERIFY IS THE GATE

THE SDLC, BUILT INseven stages, each with a contract a card must fill before it moves on Agents write the code. You stay in command.

INBOXcaptured, not started

PREPAREwrites the plan, asks its questions, you refine

DOINGbuilds: the card is the live progress report

REVIEWread the diff, see what differed from the plan, test it locally, then approve or revise

DELIVERopens the PR / MR / push as configured, then watches the deploy

VERIFYmerged, deployed, nothing around it broke

CLOSEmemory written, docs updated, last conflict check

◂ A WAY BACK AT EVERY GATE · REQUEST CHANGES, REJECT, OR REJECT & CLOSE

Review your wayChange by change like a code review, approve, revise or rollback each file, or one approve for the lot. The app is already running locally, port on the card.

A way back at every gateRequest changes sends it back to Doing. Reject parks it in Inbox. Reject & close undoes every change. Nothing merges until Verify says the delivery landed.

Customize it the way you wantEvery stage lists the sections a card must carry: the plan it must promise, the progress it must report. Defaults out of the box; add your own in Settings.

THE ONE RULE Agents run on their own. Grid interrupts you only for real decisions.

✓Never ships behind your back. Schema changes, force pushes and production stay behind a tap, whatever the autonomy dial says. ✓Everything has a way back. Any card, any change, any conversation can be rewound to an earlier point. ✓It says which decision, and why. An ask arrives with the plan, the diff and the risk spelled out: never a bare “proceed?”.

PRICINGFree is free forever; paid tiers open with the beta

Freefree forever, for individuals and personal projects€0free forever✓Unlimited projects, all core features✓As many agents as your machine can run✓Board, cards, rewind: all kept locally, one machine✓Bundled plugins, plus the ability to download (free and paid) plugins from the marketplace or your own ↗✓Commercial use up to 60 daysStart here

Soloone person, commercial use€19/month billed annually · €24 monthly✓Everything in Free, commercial use for one person✓Multi-device sync (up to 3 devices)✓Phone app and push notifications✓Paid premium plugins included ↗Join the waitlist

Crewa team on shared boards€29/seat/month billed annually · €34 monthly · 2 seats minimum✓Shared projects, boards, sessions and provenance; private projects when you need them✓Roles, per-seat autonomy caps, free viewer seats✓Jira, GitLab, Slack and Teams integrations✓Shared, version-pinned plugins and workflows ↗✓Audit log, 90 days✓Try it free with your team for 60 daysJoin the waitlist

Fleetorg-wide, on your termsTalk to usannual✓SAML, SCIM, admin-locked safety rules and plugin policy✓One seat per person across all workspaces✓Self-hosted or air-gapped licence server✓Private model endpoints, SLA, priority support✓Private plugin registry, signed builds, extended audit retention ↗Contact sales

Solo is for one person. Organizations can evaluate for 60 days, then need Crew. Prices exclude VAT for businesses; consumers see VAT-inclusive prices at checkout.Which plugins ship with which release, and what each tier can install: see Bundled plugins by version.

SELF-HOSTEDCOMING SOON Grid itself is a locally run binary. For the cloud collaboration features of Solo, Crew and Fleet we provide a Docker image: run it on your hardware and no code and no data ever leaves your control. It powers card syncing across your devices, team collaboration (Crew and up), audit logging and the phone app. Don’t want to host it yourself? We can host it for you. docker run gridconsole/sync

PLUGINS Extensibility is part of the system. Every stage transition is a hook, every panel a slot: plugins decide what Prepare must answer, what command Doing sends, how Deliver ships, what Verify accepts. Add panels, wire other LLMs, extend the code view’s right-click menu, bind your own shortcuts. Bundled and marketplace plugins on every tier; Crew pins them per project, Fleet runs a private signed registry. grid plugin add ./my-gate

MARKETPLACE Plugins have a marketplace. Developers build plugins and customisations and release them to the marketplace; the alpha already ships with plugins bundled. Plugins can be free or paid. For paid plugins Grid Console keeps 15% + $0.50 per transaction, which covers payment processing, currency handling and VAT or sales-tax collection worldwide through our payment provider. grid plugin publish

SHIPPING IN THE OPENthe last releases, unspun

Legal 0.1.0First alpha: the board, sessions, the code view, stage contracts, plugins23 AUG 2026CHANGELOG →