Manage voice agents and calls
Build and operate Bland v2 voice agents from your AI. Design agent scenarios, review or update saved configurations, inspect environments and release checks, test behavior, and manage authorized deployments. More
MCP connector Server URL https://api.bland.ai/v1/mcpCopy Login OAuth Tools Reads data113 export_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ get_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ list_ query_ query_ rebase_ search_ search_ validate_ wait_
Changes data56 add_ archive_ attach_ attach_ clone_ create_ create_ create_ create_ create_ create_ create_ create_ create_ create_ create_ create_ create_ create_ create_ create_ create_ create_ create_ create_ detach_ estimate_ estimate_ find_ hand_ link_ mark_ publish_ publish_ publish_ rename_ resolve_ resume_ save_ set_ set_ submit_ update_ update_ update_ update_ update_ update_ update_ update_ update_ update_ update_ update_ update_ upload_
Deletes or can't be undone43 batch_ cancel_ create_ create_ create_ create_ delete_ delete_ delete_ delete_ delete_ delete_ delete_ delete_ delete_ delete_ delete_ delete_ delete_ delete_ delete_ detach_ disable_ merge_ merge_ promote_ publish_ publish_ rebase_ remove_ remove_ remove_ remove_ reset_ rollback_ run_ save_ send_ set_ set_ start_ stop_ stop_
Skills 26 in the ChatGPT listing
Agent channels Use when someone wants to "embed chat on my site" or set up a "web chat widget", use the "chat API" or "turn on API chat", "text my customers", "reply to this thread", "take over this conversation" or hand it back, set up "WhatsApp", "A2P", "branded caller ID" or an "RCS contact card", or asks "why isn't the agent answering texts" or "which version answers texts", on Channels (Web chat, API chat, Messages), Identity or the Conversations page. Agent environments Use when working on an agent's Environments or Variables page ("set an env variable", "add {{env.KEY}} for production", "AGENT_VARIABLE_UNRESOLVED", "set up checks before promote", "why is promote blocked", "test my branch", "merge the branch", "the branch is behind"), covering the unresolved-variable pre-flight, the staging check policy and verdicts, and branch, rebase and merge. Agent launch Use when someone wants to take a v2 agent from built to live end to end ("launch my agent", "go live", "set it up so it answers calls", "put it in production", "what's left before launch", "onboarding checklist"), including getting a number, teaching it from documents, wiring tools, picking a voice, testing, promoting and turning on chat or texting. analytics Query Bland call volume, completion rates, and other aggregate metrics over a specified period using the named analytics tool. Analytics authoring Use when building, editing, simplifying or explaining an agent's analytics Sankey ("Where conversations go"): waypoint stages and groups, parallel branches, the unclassified share, what the Simplified and Detailed views draw, or why an end state cannot be split. Analytics cohorts Use when building a list or cohort of conversations that match a condition ("show me the calls where…", "which calls…", "how many calls…"), counting calls by lake columns, filtering by outcome, call length, metadata, tools, webhooks or disposition values, opening a filtered list on the Conversations page, feeding a set of calls to a judge run, reading, adding, editing, deleting or backfilling a headline metric tile on the Analytics page ("add a metric", "what's my containment", "how is the agent doing", also from an Overview page), or explaining why a condition cannot be expressed. Api workspace Use when working with the Bland API from a coding workspace, finding product documentation, using named endpoint tools because bland_api_get or call_bland_api is unavailable, editing agent JSON instead of clicking the dashboard, reconciling unsaved page context with saved versions, or debugging a failed Bland operation. automations Use when someone says "when a call ends, do X", "when a CRM record changes, call them", "set up a trigger", "test my automation", "how do automations work", "why didn't my automation fire", "pause the automation", "alert me when", "resend the webhook" or asks about the "post-call webhook", on the Triggers or Alerts pages or an agent's Automations or Alerts tab. Call analysis Use when finding, reviewing, categorizing, or analyzing Bland calls or conversations in bulk ("show me this agent's calls", "read the texts", "web chat threads", "export the calls"), including hundreds or thousands of recent calls, an agent's SMS, iMessage, WhatsApp, RCS, API chat or web chat threads, reproducing a specific call failure (tool logs, knowledge retrievals, SIP ladder, call scores), or producing a reproducible workspace report with evidence; also for calls in progress ("who is on a call right now", "end this live call", "transfer this call", "can you listen in") and contacts ("everything about this caller", "merge these contacts", "why is this caller not a contact") on the Active calls and Contacts tabs. dispositions Use when working on an agent's Dispositions page or post-call outputs ("set up dispositions", "track call outcomes", "extract fields after each call", "classify every call", "send the outcome to my webhook", "why is this disposition blank", "the disposition is wrong", "it didn't run", "backfill", "score past calls", "roll back the disposition", "migrate my citations"), covering definitions, extractors, test runs, corrections and align, estimates and publish, rollback, backfills and moving v1 citations and outcomes. docs Find and explain official Bland product documentation, including features that the ChatGPT plugin cannot execute, using the named documentation search and retrieval tools. evaluations Use when creating or calibrating Bland evaluation judges ("align this judge", "correct this verdict"), configuring test cases and attached judges, building judge workbench panels, running or interpreting evaluations and simulations, applying verdicts as call tags, or investigating failed, partial, or inconclusive scores. Knowledge sources Use when someone wants to "teach the agent", "add this doc / FAQ / website / file", says "the agent doesn't know", asks "what couldn't it answer", "what is it weakest on" or "what did it look up on this call", wants to "answer these questions", "test this source", "update the source", "re-sync" or "delete this source", asks how knowledge works, on the Knowledge pages (Sources, Questions, Query logs, Source testing, Knowledge map). Outbound calls Use when placing or scheduling outbound calls with a v2 agent ("call this number with my agent", "place a call", "schedule a call"), running a campaign ("run a campaign", "call this list", "batch"), stopping one ("stop the batch"), or checking recipients ("is this number on the DNC"), including from the Send call and Batches pages. pathways Inspect or validate a user-supplied native Bland pathway graph with the pathway schema, context, and compiler tools. Phone numbers Use when someone asks "who answers this number", wants to "put my agent on a number" or "point +1… at" an agent, "buy a number", "block this caller", a "fallback number" or transfer on a number, a "SIP trunk", or asks "why is my number not answering", on the Phone numbers page, the SIP pages, or an agent's Calls channel. setup Connect Bland to ChatGPT with OAuth, verify the connection, or troubleshoot sign-in, organization access, and authorization errors. Tool library Use when someone wants to "create a tool", "call my API from the agent", "wire up a webhook", says "the tool returns HTTP 0" or "why did the tool fail on this call", asks "how do tools work" or "what does the caller hear while it runs", wants to "update the shared tool", "delete this tool" or asks "which agents use this tool", wants to "add a secret", "connect HubSpot / Salesforce" or "disconnect" an integration, or wants "custom code" in a tool, on the Tools, Secrets or Connections pages. Triage issues Use when filing or working issues on the Issues page, org-wide or for one agent ("file an issue", "log this bug", "track this", "this call went wrong, record it", "attach this call to the issue", "close the issue", "link these issues", "what's open on this agent", "flag a capability gap"), including evidence, flags, assignment, alert bindings and billed eval analysis. V2 authoring Designing and building a NEW Bland v2 agent from scratch — scenario architecture (what scenarios represent and where boundaries belong), hub + entry authoring, step/edge/tool design, the standard conduct-rule set, and how v2 building differs from v1 pathway building. V2 lifecycle Use when shipping a Bland v2 agent version to real traffic or asking which version answers ("publish to staging", "promote to production", "roll back", "which version is this call running", "my fix didn't take"), running an A/B test or staged rollout on the Experiments page ("start an experiment", "ramp the new version", "finish the rollout", "pick the winner"), attaching inbound numbers, deleting an agent, setting an agent's voice and language ("change my agent's voice", "make it speak Spanish"), or on the Pathways page ("copy this pathway to another agent", "update the pathway in all my agents", "pathway versions"). V2 maintenance Use when updating, repairing, debugging, or managing an existing Bland v2 agent, including prompt edits, routing, tools, settings, versions, environment configuration, and regression testing. V2 migration The hand-migration doctrine for converting a Bland v1 pathway or persona (raw JSON export) into a v2 agent snapshot — architecture rules, verbatim-carriage discipline, the full trap catalog, and the verification bar. V2 runtime Use when predicting or debugging Bland v2 runtime behavior, explaining routing or interruptions, explaining which version or runtime answered a call, web chat, API chat or text, explaining how an attached or library tool runs, or reconciling edited snapshot JSON with the builder canvas, hidden routes, and call traces. V2 snapshot The Bland v2 agent snapshot dialect, the exact JSON shape of an agent version (behavior graph, hub, complex scenarios, flow steps, edges, tools, settings) and the rules for authoring one by hand. V2 testing Use when verifying Bland v2 agents with simulations, test chat, or speech-to-speech tests; testing a text or messaging reply ("preview the SMS reply", "test the API chat"); placing a real test call to the agent's number; measuring interruptions or response latency; grading engine traces; or deciding whether a test is safe. 
