Skip to content
AI & Automation

Vercel Board explained: what it changes for small teams

Nov 19, 2026

Vercel Board in brief

Vercel launched Vercel Board on November 18, 2026, as an experimental primitive that gives AI agents a shared message store for claiming tasks, posting updates, and recording results (launch post). For a small B2B software team it changes where agent coordination lives, not what you ship, and Vercel has not published a price for it, so this month the right move is a bounded pilot on one workflow, not a budget line.

Vercel Board is a headless message store that agents read from and write to through a CLI or JavaScript SDK, organized into channels, threads, and labels.

TL;DR: Board competes for the job your agents already do through a Markdown file, a tracker, and a chat thread. Pilot it if several agents hand work to each other. Skip it if one assistant and your current tracker already cover the work.

Checked November 18, 2026: the Vercel launch post, Linear pricing and Slack pricing were re-opened on that date.

Key Takeaways

  • Board is labeled by Vercel as an experimental agentic primitive, and the launch post gives no general-availability date.

  • Vercel has not published Board pricing or which plans include it, and its pricing page does not list Board.

  • Linear is the closest rival for agent work. Its delegation model keeps a human as the responsible assignee, while Board adds a claim-and-lease queue.

  • Board's distinct pieces are task claims with leases, scheduled follow-ups, schema-validated messages, and per-board grants with an audit trail.

  • Market data argues for a pilot with a written success test, not a company-wide standard.

  • The launch post uses vercel board in most examples and vb in the sandbox example, so check the command names before you script anything.

How we evaluated these tools

We weighted six criteria for a small B2B team deciding this month. The weights reflect what a founder or engineering lead must settle before a pilot: whether the product does the coordination job, whether humans stay in control, whether cost can be predicted, and how much integration work it creates. These weights are our judgment for this reader, not vendor data. A team with no agents running yet would weight coordination fit much lower.

We did not score the products numerically, because Vercel publishes no Board price or maturity limits and the rival's pricing is per seat. The feature matrix instead records what each vendor's own pages state, and it marks any item a page does not cover. For another weighted vendor comparison built only from public prices, see this restaurant purchasing comparison.

CriterionWeightWhy it matters for this decision
Task coordination fit (claim, lease, status, threads)25%Decides whether agents can take work without duplicating it
Human ownership and access control20%Decides whether a person stays accountable and can limit agent scope
Fit with where the team already talks (Slack, GitHub)15%Decides how much migration a pilot needs
Published price and predictable cost15%Decides whether a budget line can be written this month
Stated maturity and support limits15%Experimental products carry more risk of change
Setup prerequisites (hosting, SDK, OAuth scopes)10%Decides who must build and maintain the integration

Total weight: 100%.

What Vercel shipped on November 18, 2026

Board is a headless communication primitive and message store for agents across tools and environments. Agents work with it through the CLI or the JavaScript SDK, and the Board skill is installed with npx skills add vercel/board. Each board is organized into channels and threads, and labels record details such as the linked pull request or current status.

In the launch post's example, an agent claims a ready bug task, posts progress, renews its lease when a task runs longer than expected, and completes the task with a link to a pull request for review. A watcher can run agent code when a new reply is posted or a label changes, and on Vercel, matching events wake the deployment so no process has to stay running between events. Agents can also schedule a follow-up, such as checking CI after opening a pull request.

Board syncs two ways with Slack and GitHub through its SDK, mirroring replies, reactions, edits, and deletions. Structured messages are validated against a schema when they are posted. Each board has its own access grants, and each board's events record who performed each action. The sandbox example gives an agent board access without a credential inside the sandbox, limited to chosen channels and permissions.

Vercel says it built Board for its own security-agent and software-factory work, and it invites developers to test it and give feedback. The launch post also names harnesses that it says can collaborate through Board, including Claude Code, Codex, OpenCode, Mastra, and LangChain. We did not verify those compatibility claims.

Vendor claim: v0 factory merges over 50% of PRs according to Vercel (2026). The post does not describe how that figure was measured, so treat it as the vendor's claim.

On the pages we opened, Board is not a renamed, merged, or discontinued product. Vercel's earlier agent work includes Open Agents, which InfoQ reported on April 30, 2026 as a reference implementation for background coding agents, not a finished product, according to InfoQ (2026).

The launch post does not state the following, and Vercel has not published them: a Board price, the plans that include it, a general-availability date, service levels, data retention, limits on channels or agents, or whether Board requires a Vercel-hosted deployment outside the watcher feature.

Feature matrix: Board, Linear, and the status quo

This matrix compares six things a small team decides on. Each cell reflects a page we opened. "Not covered" means the page we read does not describe that item.

Decision factorVercel BoardLinear (agent delegation)Slack (status quo channel)
Core jobAgent message store with task claimsIssue tracker that accepts agent delegationTeam chat; no task-claim primitive in the pages we read
Task claim and leaseYes: claim with a wait, renew with a leaseNot coveredNot covered
Human stays accountablePer-board grants; events record who actedHuman stays the assignee; agent is the delegateNot covered
Connection to other toolsTwo-way sync with Slack and GitHub via SDKAgents can be @mentioned, delegated issues, and reply in commentsWorkflow Builder listed as included on the Pro plan
Structured dataSchemas validated when postedNot coveredNot covered
Scheduled follow-upsSchedule command posts a later reminderNot coveredNot covered
Stated statusExperimentalNot coveredNot covered
Published priceNot published by VercelBasic and Business plans, billed yearlyPro and Business+ plans, billed monthly or yearly

Board covers the claim, lease, and schedule work that trackers and chat do not expose as primitives in the pages we read. Linear covers the assignment half well, and Linear's docs say the assignee "remains responsible for the work, while the agent contributes on their behalf" according to Linear. Slack covers the conversation half, and Board can sync into it. The practical question is therefore not Board versus Linear alone. It is whether your agents need a shared claim queue beyond what your tracker already offers.

Pricing and total cost

Vercel has not published a price for Board, and its pricing page does not list it, so the table below prices only the adjacent plans that are published.

Pro plan on Vercel: $20 per month according to Vercel (2026). That is a hosting plan with a $20 included credit, not a Board price, and the vendor has not published whether Board needs it. Vercel's Enterprise plan is quote-based.

Linear's Basic plan costs $10 per user per month, billed yearly, and its Business plan costs $16 per user per month, billed yearly, according to Linear (2026). Linear's Enterprise plan is quote-based. Slack's Pro plan costs $8.75 per user per month on monthly billing and $7.25 on annual billing, while Slack's Business+ plan costs $18 per user per month on monthly billing and $15 on annual billing, according to Slack (2026). Slack bills only members who take an action in Slack within a 28-day period, so the list price is a ceiling, not a fixed cost.

VendorPlanPublished priceBilling basisWhat the page says is included
Vercel BoardNot namedNot publishedNot publishedThe vendor has not published this
VercelPro$20 per monthMonthlyUsage credit; hosting features
VercelEnterpriseQuote-basedCustomCustom terms
LinearBasic$10 per userYearlyAgent platform and MCP access; cloud coding agent needs AI credits
LinearBusiness$16 per userYearlyAdds triage and code intelligence
LinearEnterpriseQuote-basedCustomSame agent features as Business
SlackPro$8.75 per user monthly, $7.25 annualMonthly or annualWorkflow Builder
SlackBusiness+$18 per user monthly, $15 annualMonthly or annualAI workflow generation and AI steps

Consider an illustrative 12-person team (a scenario, not a customer) that buys Linear's Business plan for tracking. At $16 per user per month billed yearly, the math is 12 users × $16 × 12 months = $2,304 per year before any Board cost. Board's cost is not published, so the team cannot model it from list prices. The launch post's grant example uses prj_security with read and post permissions only. A one-channel pilot scoped that narrowly is how the team gets a real number from a Vercel quote instead of a guess.

Linear planUsersPrice per user per monthAnnual cost
Linear Basic12$10$1,440
Linear Business12$16$2,304

Signal vs Speculation

Signal, confirmed on the pages we opened: the launch date, the experimental label, the CLI and SDK surface, two-way Slack and GitHub sync, per-board grants, and the absence of a published price. Linear's delegation model keeps a named human as assignee, and Linear's own docs list restrictions such as no default assignee for new issues.

Speculation, our read: Board will be adopted first by teams already building on Vercel, because the watcher feature depends on Vercel deployments. The vendor's own reference uses are internal, so a small team has no outside reference case to check against.

Market context points the same way. Gartner forecast: over 40% of agentic AI projects canceled by 2027 according to The Star (2025), citing "escalating costs and unclear business value." That forecast is about the category, not about Board, but it is a good reason to define a success metric before the pilot starts.

Vendor profiles

Vercel Board

Best fit: a team already deploying on Vercel that runs several agents touching the same work, and that needs claims, leases, and an audit trail. Limitations: the product is experimental, the price is unpublished, watcher events depend on a Vercel deployment, and the launch post does not state limits. Implementation: install the skill with npx skills add vercel/board, add a route file like the one in the launch post, configure Slack or GitHub sync if needed, and define schemas and grants before the first agent run. Primary evidence is the launch post. Choose Board if you have an engineer who can maintain the route and accept experimental risk. Disqualifiers: no engineer to own the integration, or a requirement for a published service level before purchase.

Linear

Best fit: a team whose issue tracker is the system of record, where humans must stay responsible while agents do the work. Limitations: Linear does not support a default assignee for new issues, agents need access to the issue's team, private team issues can only go to private team members, and some third-party agents need a teammate's connected account to start work. Implementation: assign an issue to an agent from the assignee field, or @mention the agent in a comment, as described in Linear's assignment docs. Installing an agent app with a workspace scope requires admin permissions. Choose Linear if the tracker already runs your work. Disqualifier: you need a claim-and-lease queue that several independent agents can draw from.

Slack and the status quo

Best fit: a small team where agent work is light, and coordination is conversational. Limitations: Slack is a chat system, and the pages we read do not describe a task-claim or lease primitive. Teams that use a Markdown file for each agent get no shared queue at all. Implementation: keep human conversation in Slack and let Board mirror threads through its Slack sync. Choose the status quo if agent work is occasional. Disqualifier: relying on chat threads as the audit record for agent actions.

When NOT to use US Tech Automations: a team with one agent and a tracker that already works should skip any orchestration layer. A team that cannot grant API access to its tracker or chat tool cannot run the proposed design below. A team that wants no custom code and no review step should stay with the native assignment features of its tracker.

Who this is for

This is for a founder or engineering lead at a small B2B software company that already runs two or more agents touching the same work, and that has one engineer willing to own an integration. If your agent model choice is still open, the GPT-5.6 alternatives guide covers that decision separately.

Red flags: no engineer to maintain a route handler; no agent work running yet; a need for a published price or service level this quarter.

Decision checklist

  • Name the three to five agent tasks you would move to a shared queue, and confirm each one currently has a human owner.

  • Confirm your tracker's delegation model can serve as the owner record before adding a second system.

  • Ask Vercel for a Board quote, the plans that include it, and its limits, since none are published.

  • Write the success test for the pilot: duplicate work avoided, or time from task to reviewed pull request.

  • Set a stop date for the pilot and a rollback path to the current tracker and chat setup.

Common mistakes

  • Treating the 50% pull request figure as an independent result rather than the vendor's own claim.

  • Mixing up Vercel's hosting plan prices with a Board price.

  • Giving agents broad grants during a pilot and reviewing access only after an incident.

  • Scripting vb commands from one example while the rest of the post uses vercel board.

  • Assuming Linear's agent delegation covers claim and lease behavior, which its docs do not describe.

Build it yourself, or configure a managed design?

Zapier, Make, and n8n can keep run histories, retries, error branches, and audit evidence when you configure them. You still design and own observability, idempotency, escalation, access controls, and maintenance. Most small teams underestimate the last of these.

A proposed US Tech Automations design would be configurable rather than built from scratch. Trigger: a new issue delegated to an agent in Linear, or a bug report posted in a Slack channel that is mirrored to Board. Action: the design creates or updates a Board task with a stable identifier so a retried event does not create a duplicate, then routes it to an agent role with read and post grants on one channel. Output: a Board thread linked to the Linear issue, and a summary posted back to Slack. Prerequisites are a Linear API or OAuth connection with the scopes it needs, a Board SDK route deployed by your engineer, and a Slack connection. For the Slack side, see how a monday.com to Slack automation is connected. This is a proposed, configurable design, not a deployment we have run for any customer.

Human review points would sit in three places. A person approves any grant change before it takes effect. A person approves any merge that an agent proposes. A person reviews the weekly list of tasks an agent claimed and did not complete, since a stale lease is the most likely silent failure. The design would also need its own escalation rule, for example routing any task that stays claimed past its lease window to the engineer who owns the integration.

Questions buyers ask

Is Vercel Board generally available?

No, the launch post calls Board an experimental agentic primitive, and the page does not give a general-availability date.

What does Vercel Board cost?

Vercel has not published a price for Board, and its pricing page does not list it, so the only prices you can budget from today are the adjacent plans in the table above.

Do I need Vercel hosting to use Board?

The vendor has not published whether Board requires Vercel hosting, but the launch post says matching watcher events wake a Vercel deployment.

How is Board different from Linear's agent delegation?

Linear keeps a human as the assignee and adds the agent as a delegate, while Board adds claim-and-lease task handling, scheduled follow-ups, and schema-validated messages that the Linear pages we read do not describe.

When should a small team skip Board?

Skip it if you have one assistant and a tracker that works, or if you need a published price or service level before the end of the quarter.

What to do this month

Run one bounded pilot on a single channel with one agent role, a written success test, and a stop date. Ask Vercel for a Board quote and its limits before you commit to anything beyond the pilot. If the tracker you already use can hold the ownership record, keep it, and treat Board as an add-on for claims and follow-ups only. For a team that wants the review gates and audit trail configured rather than assembled, see how US Tech Automations configures this.

Sources: Vercel launch post, Vercel pricing, Linear pricing, Linear assignment docs, Slack pricing, The Star, InfoQ

About the Author

Garrett Mullins
Garrett Mullins
Workflow Specialist

Helping businesses leverage automation for operational efficiency.