Getting started
“Hive can’t reach the sign-in service”
Hive loads its sign-in service before it shows anything else. If that has not finished after 15 seconds, Hive shows this screen instead of an empty, loading page. Check your connection, and turn off any VPN, firewall rule or browser extension that blocks scripts or requests for Hive’s sign-in, then select Retry. If it keeps appearing, contact support.”Your workspace is still being set up”
Right after you create a workspace, Hive may need a few seconds to finish setting it up. It retries on its own, and this message shows only if setup keeps failing: “Your workspace is still being set up. Hive will retry automatically.” If it is still there after a minute, reload the page, or contact support.“Waiting for an administrator”
Hive has not let you into the workspace named on the screen. The screen does not show anything from that workspace. Under Switch workspace it lists the other workspaces you belong to, so you can open one of them, and Sign out lets you use another account. Its message says what to do:- An Owner or Admin can let you in. Ask an Owner or Admin of that workspace to choose Invite on Settings → Members and enter your email. They also get an Inbox card saying you are waiting. Once they have, choose Retry.
- The workspace has no Owner or Admin in Hive yet. Nobody has opened it in Hive since it was created. Ask the person who created it to sign in to Hive and open it; they become its Owner, and you can then choose Retry.
- Your earlier access must be reviewed. You were a member of the workspace before and had access to items there, and Hive needs to review what you could reach then before you return. Ask an Owner of the workspace to contact support. If you had no access to items there, you are told instead that an Owner or Admin can invite your email again.
“We couldn’t reach your workspace”
Hive checks your access to the workspace when it opens. If that check fails on the network, or while Hive is restarting for an update, Hive tries again twice over a few seconds before it shows this screen. When you see it, check your connection and select Retry. If it keeps appearing, reload the page, or contact support.“Hive was updated. Reload to get the latest version.”
This banner appears at the top of the screen when you have kept a Hive tab open while a new version was released. The tab keeps working: it goes on loading its screens from the version it opened with, and anything the new version added that your tab does not know yet, such as a new step in the Add step menu or a new built-in source, is left out until you reload. Select Reload when you are ready, or dismiss the banner to carry on.”Updating Hive…”
You opened a screen in a tab that was open across a new release of Hive, and the tab’s own version of that screen is no longer available: the tab has been open for a long time, or your browser is too old to keep using it (Safari before 16.4, Firefox before 108). Hive reloads the page once to load the new version. You do not need to do anything. If the same screen still does not load after the reload, reload the page yourself, or contact support.Connectors and integrations
A connector says Disconnected, Reconnect required or Error
The provider no longer accepts the account’s sign-in: its access was revoked or expired, or its permissions were withdrawn. The account’s owner gets one … disconnected notification in the bell. Hive does not reconnect it on its own.1
Reconnect the account
Go to Integrations and select Reconnect on the tile, or in its Accounts panel when
several accounts need it. Sign in as the same account and approve the permissions it asks
for. Approving fewer permissions than requested is the most common reason a connection fails
again.
2
Check it is reading
The tile shows Last synced and its item counts again once Hive has read it.
”Personal connection authority could not be verified”
Hive could not confirm your connected account just now, usually because the provider or Hive’s connection service did not answer in time. Nothing was sent to the app, and nothing is wrong with your connection: try again in a minute. You do not need to reconnect. If the message stays for more than an hour, check the app’s tile on Integrations.Slack channels or the list of available apps do not load
If Hive cannot reach Slack or its connection service, these places show an error instead of an empty list:- the Slack Channel picker when you set up GitHub pull-request summaries;
- Settings → Integrations, for Owners and Admins;
- the All integrations list when you choose an agent’s integrations.
”A workspace administrator must enable this provider”
Members can connect their own accounts only after an Owner or Admin has enabled that provider for the workspace. Ask an Admin to connect it once — that enables it for everyone. See Accounts and autonomy.PostHog connects but returns no data
PostHog asks you to choose a region when you connect: European Union (eu.posthog.com) or United States (us.posthog.com). The wrong region resolves to a different host, so reads come back empty. Disconnect and connect again with the region your PostHog project lives in.Sage 200 refuses the connection
Choose the edition you use — Sage 200 Standard Online or Sage 200 Professional / Extra Online — when asked. The Sage user you sign in with needs Sage 200 API access and Company Access to the company you want Hive to read. Ask your Sage administrator to grant both, then connect again.Installing Microsoft Teams fails
The person installing the Teams channel must hold one of these Microsoft Entra roles: Global Administrator, Privileged Role Administrator, Cloud Application Administrator or Application Administrator. The Teams Administrator role alone is not enough. After consenting, the Hive app must also be added from inside Teams. See Slack and Teams.A custom MCP server’s tools look out of date
Resync the server from its card. Hive re-discovers the tools and pins their definitions again, which it does deliberately so a server cannot quietly change a tool after you connected it. See MCP servers.Signals
Signals is not in the sidebar
Signals is an optional product. It is on by default for new workspaces, but an Owner or Admin may have hidden it, and workspaces created before this change started with it off. An Owner or Admin turns it on under Settings → Products → Signals. See Signals.The Signals feed is empty
An empty feed is not an all-clear: Hive can only raise signals from what it observes.- Open Signals and switch to Matters. Check that the account you expect has been read, and when.
- Confirm the tool you expect is connected under Integrations.
- A tool you have just connected is synced straight away, and Hive queues a Signals read for that member within minutes, ahead of the hourly checks, once mail or files from that sync are stored. Hive still checks every connected tool once an hour on its own. Signals reads only the accounts of a member with a Max seat: it does not read the accounts of a member on a Pro seat. An Owner or Admin can change a member’s seat on the Members page (Settings → Members). Owners and Admins see Next connector check in at the top of the Signals page and Next run in on the Connectors tab of Integrations.
- Check the source system actually contains something that should fire — Hive needs a sustained change, not a single spike.
A signal type keeps coming back after I dismiss it
Every dismissal carries a reason, and the reason teaches Hive. Done it and Already done quiet that record, Not for me and Wrong count toward alert fatigue for that kind of signal, and on an email signal Not for me can stop the sender or everyone at their company. See Triage signals.I cannot act on a signal or approve an action
Snoozing, dismissing and approving need an active seat and the Operator role or higher; a member on an Approver seat approves only what is routed to them. Acting on a signal (Act on this, or retrying its preparation) and setting a watch also need a Max seat: a member on a Pro seat is not offered them, and a member without a seat, or whose seat is suspended over the plan’s limit, cannot use the workspace at all. An Owner or Admin can change your seat on the Members page (Settings → Members). Each role also has a ceiling on how risky an approval it can make — Operator up to B2, Department lead up to B3, Owner and Admin up to B4. See Roles and permissions.Ask Hive
The answer ends with “Live sources were not checked”
Hive tried to read a connected tool live and the lookup failed, so the answer is based on stored Brain data that may be out of date. Check the connector’s status, then ask again. Pinning the tool with@ makes Hive read it directly.
A model is missing or unavailable
The picker only shows models your workspace can run. An Admin sees unavailable models with a reason — the provider is not configured, or the model lacks a capability the task needs — and can change the allowlist under Settings → Models. See Models. When you deploy an agent whose model your workspace can’t use, Hive says “This agent’s model, …, is not available in this workspace”, followed by why and what to do. Choose another model under Model in the agent’s Basics & instructions, then deploy again. An Owner or Admin sees the exact reason:
Other members see the model and the next step, and are asked to have a workspace admin check Settings → Models.
“This answer was interrupted by an update”
Hive restarts briefly when it is updated. An answer it is already writing is given time to finish first, whether or not you still have the chat open, so most answers finish normally. Most answers you ask in the app run on Hive’s background service, and an update waits for them for up to 30 minutes, the longest an answer can run. An answer Hive had to write elsewhere, for example while that service was busy, gets about four minutes. One still running after its time shows “This answer was interrupted by an update.” straight away. If Hive stopped unexpectedly instead, the message can take two to three minutes to appear, even if you reload the page.- An answer that had only read and searched when it was interrupted is started again for you, once, a minute or two later: its steps show “Resumed after an update” and it finishes as usual. If it is interrupted a second time, or cannot be started again within about two minutes, it shows the message above.
- Select Retry to ask the same question again, with the same
@pins and mode. Hive never re-runs an answer on its own once it has made or changed a file, proposed an action or sent anything: only you decide to ask again. - If you had approved an action when the answer was interrupted, its card says “Hive could not confirm whether this action completed. Check the provider before trying again.” Check the app (for example, your email drafts) before you retry, so the action is not done twice.
- Credits held for the interrupted answer are returned to your workspace’s pool automatically. You are never charged twice for one answer.
”Hive is restarting for an update. Send your message again in a few seconds.”
Your message reached Hive just as it was restarting for an update, so it was not started. Nothing was saved or charged. Send it again after a few seconds and it starts on the updated Hive. Answers that were already being written carry on while Hive restarts.“Hive stopped this answer because it ran for more than 30 minutes”
An answer keeps going on its own if you reload or leave the page, even the moment after you send your question, and appears when you come back. If an answer is still running after 30 minutes, something kept it from finishing: Hive stops it, so it no longer shows as generating and you can ask again in the same chat. An agent’s answer stops at the run time limit set for that agent instead.”Hive could not load this chat. Reload it to see the answer.”
The connection dropped while Hive was answering, and the page could not load the chat again to follow the answer: for example the connection stayed down, or Hive kept answering with errors. Hive usually finishes the answer anyway. Reload the chat to see it. Don’t send the question again before you reload, or Hive may answer it twice.”The connection to Hive was lost before this question was saved. Send it again.”
The connection dropped before Hive saved your question, so Hive never started the answer. Send the question again.”The model could not read the attached files as Hive sent them”
The AI provider turned the request away because of what was attached, for example a file or picture too large for it, so this question was not answered. Your chat is saved. Try again with fewer or smaller files, ask about one file at a time, or pick another model. Hive already fits pictures to what models take before it sends them: each picture is shrunk to at most 1568 pixels on its longest side, a picture used on many slides is sent once, and one answer looks at no more than 20 pictures across all its files, the largest first. When a deck has more, the answer says how many it looked at; the rest are read through the slides’ text only. Ask about one deck at a time to have more of its pictures looked at.”The model stopped responding partway through, so Hive ended this answer”
The model went silent while it was writing, so Hive stopped waiting. Your chat is saved: ask again, or pick another model for this message.An attached file stays “Processing” or says “Failed”
While Hive reads a file you attached, its chip says Processing and the send button waits. Most files are read in seconds; a large one can take a few minutes. A file Hive has already read, attached again in another chat, can be sent at once. If Hive still hasn’t finished reading a file after about 15 minutes, it tries once more. If that doesn’t finish either, the chip says Failed and the file is no longer waited for or sent with your question. Remove it and attach the file again. If the same file fails again, contact support.The answer is weak or vague
- Pin the right connector with
@. - Name the customer, date range or metric.
- Attach the document if the information is not in a connected tool.
- Ask Hive to cite the source for each claim, and do not act on an answer that cannot.
Workflows
Deploy is refused
Deploying checks the whole workflow first. When something is wrong, nothing is deployed and Hive shows a message for the first problem it finds. Fix it and deploy again; if another problem remains, the next attempt shows that one. Step ids and names in the messages below (shown as…) are
filled in from your workflow. Some messages use technical terms, such as “edge” for a connection
between steps and “node” for a step, so match on the words around the … rather than the whole
message.
The builder shows these problems in the language you use Hive in, with the step’s name. The English
messages in the table are what the API and the Hive MCP server return.
There, a workflow_invalid refusal’s details.issueCode is a stable code for the problem and its
details.nodeId names the step; match the row below on the English message. An account refusal’s
details also carry provider (the app) and reason: connect, reconnect or
choose_account.
See Test and deploy and Runs and errors.
An agent in the Agent library is greyed out
The Agent library in Add step lists every deployed agent, but greys out one a workflow step cannot run and shows why under its name. Change the agent in Agent Studio, deploy it again, and it becomes available:
If the Agent library heading is missing altogether, your workspace cannot run Agent library agents.
See Agent library steps.
An agent run shows Interrupted
The run was in progress when Hive restarted for an update, so it could not finish. Hive closes such a run within about 15 minutes and marks it Interrupted in the agent’s run history, which also frees the agent to run again. Open the run and run the agent again. Hive does not restart it for you, because the run may already have carried out some of the agent’s selected actions — check the Audit log first, so an action is not done twice.A Studio Test stopped when Hive restarted
A Test in Agent Studio runs only while your request is open, so a Test that was running when Hive restarted for an update stops before it answers. Run the Test again. A stopped Test frees the agent within a couple of minutes, so the new Test is not refused as still in progress. Background recovery never takes over a Test that is still running.”This form is not available”
A public form link stops working when it is reset, the live release no longer has the form step it was created for, someone who can’t manage sharing on the workflow deploys or re-activates a release, the workflow is paused or deleted, or its owner loses the right to run it. The owner deploying again does not end it while the form step is still there, but a link created before 5 October 2026 still ends at the next deploy. Open the workflow and select Form link: if it shows a link, that is the live one — share it; otherwise create a new link and share that. See Schedules and forms.A schedule stopped running
A schedule runs as the member who created it. If a different member’s release becomes live, the schedule waits for its owner. While its owner holds no seat, it skips each run and records why; it runs again once they have a seat. Check Scheduled for its status and next run.A workflow or agent stopped starting runs
Check its run history. A workflow shows a Failed run that names the reason (below). An agent shows a Blocked by policy run with the reason, such as “This workspace is out of credits. An owner or admin can see how to get more in Billing.” or, for the agent’s own monthly limit, “A spending limit for this work has been reached. It resets with its window.” The messages below are a workflow’s:- “This workflow reached a spending limit for its period” — a credit limit stopped the run. Usually the workflow used its Monthly credit limit: raise it on the workflow’s Settings tab, under Limits, or wait for the next billing month. A scheduled run, or one started by a connected app event, also shows this message when it reaches the workflow’s Per-run credit limit, or when the workspace reaches the Workspace monthly limit an Owner or Admin sets under Settings → Usage → Spend controls. See Credit limits.
- “This workflow reached its runs-per-hour limit” — it started its Runs per hour; it can run again next hour.
- “The workspace reached its plan limit for app-event runs this billing period” — on Team, connected app events have started the plan’s monthly runs (20,000). They start again next billing month. On Business, app-event runs count as workflow runs instead.
- “This workspace has used its workflow runs for this month” — on Business, the run’s owner has used their 2,500 runs and the workspace’s execution add-on is used too, or it has none. An Owner or Admin can add or raise the execution add-on in Settings → Billing → Add-ons; otherwise runs start again next month.
- “The workspace is out of credits” — a scheduled or event-started run waits up to 72 hours for credits instead (the workflow shows runs waiting for credits). On an active paid plan, an Owner or Admin can buy a credit top-up under Settings → Billing → Credit top-ups. During the trial, choose a plan under Settings → Billing instead. See Waiting for credits.
”You have used your AI credits for this month” or “You have reached your £50 top-up limit for this period”
On Business, each member has their own AI credits.- “You have used your AI credits for this month”: you have used yours and extra usage is off. An Owner or Admin can turn on Enable extra usage in Settings → Billing → Usage and spend limits, or wait for your credits to come back on the date the message gives. If it adds “A payment failed, so extra usage is paused until it is paid.”, extra usage waits for the failed payment: an Owner or Admin pays it (see “A payment failed” below).
- “You have used your trial’s AI credits”: a trial has no extra usage, so an Owner or Admin chooses a plan in Settings → Billing for you to go on.
- “You have used your AI credits for this month”, then “Contact Hive support if you need more sooner.”: your workspace is on a complimentary plan, which has no extra usage. Your credits come back next month, on the date Ask and Signals show, or Hive support can add more sooner.
- “You have reached your £50 top-up limit for this period” (the amount is your own limit): you have used your included credits and all of your top-up limit. The message says when it resets. Choose Request a higher limit in the notice or in Settings → My usage, if the workspace allows usage requests, or ask an Owner or Admin to raise it. An Owner or Admin raises their own with Raise your limit in Billing: nobody is asked to approve their own request.
- “Your workspace has reached its monthly maximum for extra usage”: the whole workspace has used its maximum, so AI work that needs extra usage is paused for everyone until it resets. An Owner or Admin can raise it with Raise the maximum.
”This workspace is out of credits” or “You have reached your monthly credit limit”
On an earlier plan, the workspace’s shared credit pool is used up, or you have used your own monthly limit (by default 1.5 times the credits your seat adds). Members on a Pro or Max seat keep 10 quick answers a day and 200 a month: short answers on an Efficient model, with no web search or files. On an active paid plan, an Owner or Admin can buy a credit top-up under Settings → Billing → Credit top-ups, or turn on auto top-up there. A top-up does not lift your own monthly limit: to raise it, they change your cap under Settings → Usage → Member caps (the composer links there), or the member cap for everyone under Spend controls. During the 7-day trial, top-ups are not sold: an Owner or Admin chooses a plan under Settings → Billing instead. A complimentary plan cannot buy top-ups either: its credits come back when the next period starts, and Hive support can add more. See Billing and plans.Billing and access
”Your trial has ended”
The 7-day trial is over. The workspace is read-only, background work such as Signals is paused, and your data is kept. An Owner or Admin can choose a plan under Settings → Billing to restore access; other members are asked to contact an admin. See Trial and payments.”A payment failed” or “The workspace is read-only”
A payment failed. The banner says what still works until it is paid and the date the workspace becomes read-only if it is still unpaid, 14 days after the payment failed. On Business, members keep using the AI credits they have, and Signals, the brief and Brain upkeep keep running; extra usage, schedules and app triggers are paused. On an earlier plan, only included background work, such as Signals within each Max seat’s fair use, and each Pro or Max member’s quick answers run. Owners and Admins see the amount due and Pay with the amount, for example Pay £619.50, which opens the invoice to pay it; the Plan card in Settings → Billing and the email say the same. Other members are asked to have an Owner or Admin pay it. Access returns as soon as the invoice is paid, and your data is kept throughout. See If a payment fails.A button is greyed out and says the workspace is read-only
The trial has ended, a payment is 14 days overdue, the subscription has ended or Hive has paused the workspace. Buttons that would create, run, send, publish or connect something stay off until the plan is active again, and Ask’s composer says “Sending is off while the workspace is read-only.”; members with a seat can still read everything. The banner at the top of the page says what fixes it. See While the workspace is read-only.”You need a seat to use this workspace”
You don’t hold a seat in this workspace, so you can do nothing in it: an Owner or Admin moved your seat to someone else, seats were removed and yours was suspended over the plan’s limit, or no seat was free when you joined, including when you were invited during the trial without a seat. The screen names the workspace and says its Owners and Admins have been told: Hive lets them know when someone joins without a seat. Once one of them gives you a seat on the Members page (Settings → Members), choose Check again; you can use another workspace from Switch workspace meanwhile. An Owner or Admin without a seat is taken to Members or Settings → Billing instead, to give themselves a seat or buy one. AI clients get the same message, with the codeseat_required. See A member without a seat.
On an Approver seat, I can’t ask Hive or connect an account
An Approver seat is for signing off what is sent to you: you can open the approvals routed to you, approve or deny them within your role’s tier and undo your own decision while it can be undone, and ask Hive 10 questions a month about those items. It does not include open Ask, Hive answers in Slack or Teams, quick answers, the audit log, connected accounts, building agents or workflows, or AI clients. Ask an Owner or Admin for a Pro seat if you need them. See The Approver seat.I cannot see a screen
Some settings are for Owners and Admins only, some products can be switched off for the workspace, and your role may be limited to certain departments. Ask an Owner or Admin to check your role in Settings → Members.Actions are halted
A banner reading “Kill switch is on” means every autonomous action and approval is stopped company-wide. Owners and Admins can select Resume autonomy — but only once the reason it was engaged is understood. Other members do not see Resume autonomy; the banner says an Owner or Admin resumes it, so ask one of them. See Safety and control.Before you contact support
Collect:- the screen, the exact message and the time it happened;
- your role;
- the connector, workflow, signal or conversation involved, with its link;
- the Audit receipt for an action problem;
- whether Retry changed anything.