Gainsight AI: Apps, Agents and Workflow Automation
Understand Gainsight Copilot and Staircase AI, verify the APIs, and build governed account reviews on Major with cross-system evidence, durable state, and human approval.

Gainsight AI helps customer success teams understand account health, summarize customer context, and manage follow-up inside Gainsight. This guide covers its native AI, the Gainsight API, and how Major adds governed apps, agents, and workflow automations across customer systems.
Major agents build deterministic apps for repeatable work. Those apps hold the account evidence and review history, while agents assess ambiguous signals and workflows route approved actions.
What Gainsight is and who relies on it
Gainsight CS, also called Gainsight NXT, brings customer records, health context, Calls to Action, and Success Plans into customer success operations. Its API overview exposes those objects for integration with external systems.
Customer success managers use CTAs to organize follow-up and Success Plans to track account objectives. CS operations teams maintain account mappings, templates, ownership, and the data feeding those processes. Leaders need consistent account evidence before prioritizing an escalation or renewal intervention.
The operational difficulty is often outside Gainsight. A product analytics decline, a growing support backlog, and a renewal conversation may describe the same account without sharing a reliable identifier or review history.
Major fits at that boundary. An agent can build an app that joins approved account identifiers and preserves evidence windows, then run that software on subsequent reviews. The agent spends its judgment on whether the combined evidence warrants action. The CSM keeps a review queue rather than receiving another unexplained alert.
Gainsight AI: native capabilities and where Major adds value
Gainsight Copilot answers natural-language questions using Company, Relationship, Timeline, Cockpit, Success Plans, Scorecards, and other supported sources. It can summarize account context and help users create or update supported records through conversation. Admin enablement and permissions matter.
Copilot is useful when a CSM needs account context where the account work already lives. Its documented boundaries also matter: Timeline analysis uses up to 50 recent entries per account, and text analysis of CTAs, tasks, and Success Plans has record and time-window restrictions.
Staircase AI adds named capabilities: Ask Staircase answers questions using customer communications, Risk Analyst identifies potential account risks, and Expansion Analyst surfaces possible growth opportunities. These are native Gainsight capabilities, not Major features.
Major adds a different execution layer: cross-system evidence collection, managed state, explicit approval, and recorded outcomes. For example, an agent can assess whether Amplitude usage and Zendesk support evidence agree, while an app preserves the comparison and a workflow waits for the account owner. Native AI remains useful inside Gainsight; Major coordinates the broader process.
How Gainsight connects to Major
Gainsight connects through Major's Pipedream connector, app pgainsightnxt. There is no native Gainsight connector. Major orchestrates Gainsight; it never replaces it.
IT sets up the connector once, binds it to the correct Gainsight tenant, and stores credentials centrally in the connector vault. Apps, agents, and workflows use scoped access rather than copying keys into code or prompts. Major SSO, RBAC, and audit logs govern access to the resulting software.
The Company API documentation shows the example host https://companyapi.gainsightcloud.com. Configure the tenant's actual API base URL rather than treating that example as a universal endpoint. The documented API uses an access key supplied in the accesskey header.
Connector access does not mean every documented API operation is automatically exposed as a ready-made action. Confirm the required operation and payload support before implementation. Keep the first version read-only, test account matching, and grant write access only to the workflow steps that need it. Connector scoping is an additional control, not evidence that Gainsight provides granular OAuth scopes for these endpoints.
Gainsight API guide for apps, agents, and workflows
Authentication, pagination, limits, and testing
The Company API documents an admin-provided access key, JSON requests, and keys that do not expire. No named OAuth permission scopes are documented for the object operations below. Protect and rotate the key under your organization's policy.
Company queries use limit and offset; CTA and Success Plan retrieval use pageSize and pageNumber. The latter APIs return up to 1,000 records per request. Keep each API family's pagination separate.
Published limits conflict. The overview recommends 100 synchronous calls per minute and 100,000 daily, while the CTA reference lists 50,000 daily. Confirm the tenant allowance and budget conservatively. Rate-limit response-header names are not documented publicly in these references; do not assume a remaining-quota header exists.
For events, External Actions lets Rules Engine or Journey Orchestrator call an external API. Retries default to three and can be configured up to five. The guide documents configurable headers, but no standard HMAC signature or retry timing. Authenticate the receiving endpoint with an approved configured mechanism; do not invent signature verification.
The Copilot API guide confirms instance-specific keys and matching sandbox or production URLs. Use your available sandbox with separate credentials and synthetic accounts, not production writes disguised as tests.
Query Company records: POST /v1/data/objects/query/Company
Retrieves selected Company fields using filters and pagination. This is the account-matching starting point, not permission to copy the whole customer database.
- App: Store approved external-account-to-Gainsight identifier mappings and their validation status.
- Agent: Read company context before judging whether an apparent usage decline belongs to the correct account.
- Workflow: Fetch the eligible portfolio on schedule and checkpoint completed pages before continuing.
Update Company records: PUT /v1/data/objects/Company?keys=fieldName1,fieldName2
Updates Company records using configured matching keys. The field names shown are placeholders, not fields to send unchanged; the documentation supports up to three keys.
- App: Display the current field value beside the proposed approved change.
- Agent: Draft a reason for changing an agreed customer-risk indicator when evidence warrants it.
- Workflow: Validate matching keys and permitted fields, then execute only the approved update and store its result.
Retrieve CTAs: POST /v2/cockpit/cta/list
Fetches Calls to Action. A POST here retrieves records; it does not create a CTA.
- App: Show existing open follow-up beside each proposed alert so reviewers can spot duplicate work.
- Agent: Compare a new signal with existing CTA context and judge whether it adds a distinct concern.
- Workflow: Page through the relevant records before proposing a create or update operation.
Create a CTA: POST /v2/cockpit/cta/
Creates a Call to Action using the tenant's valid configuration and required record fields.
- App: Hold the proposed owner, rationale, evidence window, and approval record before submission.
- Agent: Draft a concise explanation of the account risk and suggested next step.
- Workflow: Check approval and duplicate status, create the CTA, and save the returned identifier or failure details.
Update a CTA: PUT /v2/cockpit/cta/
Changes an existing CTA rather than opening a second alert for the same concern.
- App: Present the previous state and proposed change for review.
- Agent: Assess whether fresh support evidence strengthens the existing concern or changes its explanation.
- Workflow: Revalidate the target record and approved fields before writing, then retain the response in the action ledger.
Retrieve Success Plans: POST v2/successPlan/list/
Retrieves Success Plans with pagination. The reference shows this relative route without a leading slash; resolve it against the configured tenant base URL.
- App: Associate active plan identifiers and objectives with the account review.
- Agent: Read plan context before deciding whether a ticket trend threatens an agreed objective.
- Workflow: Refresh plan context before review and mark stale or unavailable data explicitly.
Create a Success Plan: POST v2/successPlan/
Creates one Success Plan per request. The reference requires company, plan type, name, owner, and an external reference identifier.
- App: Store an approved plan proposal and the external reference used to track it.
- Agent: Draft proposed objectives from reviewed account evidence, without inventing customer commitments.
- Workflow: Validate required fields and owner approval, then create the plan and record the outcome.
Describe object fields: POST /v1/meta/services/objects/describe
Returns field metadata for specified objects, helping integrations validate the tenant's actual schema.
- App: Maintain a reviewed field-mapping configuration instead of hard-coding assumptions about custom fields.
- Agent: Use approved field descriptions to interpret context, not to authorize its own schema changes.
- Workflow: Detect missing or changed fields before execution and stop incompatible writes for administrator review.
Gainsight automation use cases on Major
Usage-drop risk review with Gainsight and Amplitude
This extends the published churn-alert process linked below without duplicating its build instructions.
- App: Track account mappings, baseline windows, missing-data flags, and previous risk reviews.
- Agent: Distinguish sustained adoption loss from seasonal behavior, incomplete ingestion, or a changed account structure.
- Workflow: Schedule retrieval, query Company records, retrieve existing CTAs, and create a CTA only after review.
- Endpoints: Company query, CTA list, and CTA create from the linked API guide above.
Support escalation review with Gainsight and Zendesk
Rising ticket counts deserve context before they change an account assessment.
- App: Store weekly ticket counts and severity alongside active Success Plans and account identifiers.
- Agent: Judge whether the trend threatens a plan objective, and explain contrary evidence.
- Workflow: Retrieve Company and Success Plan context, route the recommendation to a CSM, then perform an approved Company or CTA update.
- Endpoints: Company query and update, Success Plan list, CTA list and update.
Renewal coordination with Gainsight, Salesforce, and Slack
This is a proposed cross-system pattern, not a claim about a published customer deployment.
- App: Hold the renewal review queue, ownership exceptions, evidence timestamps, and approval history.
- Agent: Reconcile the account narrative with renewal timing and draft an internal recommendation.
- Workflow: Refresh account context, request owner approval, and create a Success Plan only when the owner accepts the proposed work.
- Endpoints: Company query, CTA list, Success Plan list and create. External-system access must be configured separately.
Architecture: build the customer-risk control surface on Major

Figure 1. Account evidence becomes a reviewed Gainsight action, with state held in the Major app.
Build the app first around durable account reviews, not a chat transcript. Its managed database should hold account mappings, evidence references, review windows, proposed actions, approvals, execution attempts, and remote record identifiers. A CSM sees pending reviews and previous decisions through the same app the agent uses for execution.
Then separate the moving parts:
- Connectors fetch permitted fields from Gainsight and the other approved systems.
- Workflow validates freshness, checks identifiers, computes repeatable comparisons, and routes missing data.
- Agent weighs ambiguous evidence and drafts a recommendation with uncertainty and source references.
- Decision gate applies the approval policy before consequential external changes.
- App and output retain the decision, execute the approved action through the workflow, and record its receipt.
An agent can build the software for those repeatable comparisons once and run it again. The model still reasons about each genuinely ambiguous account, but pagination, deduplication, arithmetic, and execution tracking stay in code. Reason once. Run forever.
A worked example with illustrative numbers
For example, suppose a CS operations team reviews 200 accounts each week. Manual collection and comparison take 12 minutes per account, or 200 × 12 = 2,400 minutes, which is 40 hours. These are planning assumptions, not a customer result.
A Major workflow gathers the permitted usage and account context. The app validates identifiers and computes comparison windows. The agent assesses the exceptions and drafts the supporting rationale rather than asking a person to reconstruct every account from scratch.
Suppose 160 accounts need a one-minute check and 40 exceptions need eight minutes of review. Human review becomes 160 × 1 + 40 × 8 = 480 minutes, or eight hours. The illustrative difference is 32 hours before implementation, maintenance, API costs, and ongoing quality checks.
A CSM approves one risk CTA after reading the evidence. The workflow creates it and the app saves the response. If the write times out, the item becomes an execution exception, not a second automatic create. Measure actual review time, false positives, missed risks, and duplicate actions before expanding the portfolio.
Governance and failure handling

Figure 2. Internal drafts can proceed while consequential Gainsight changes require a CSM decision.
Define permission boundaries before rollout. An agent may analyze approved evidence and draft an internal recommendation. It must never independently promise commercial terms, contact a customer, change account ownership, or rewrite health indicators outside the agreed policy. For an initial rollout, require CSM approval for every external write.
Use SSO and RBAC to separate reviewers from connector administrators. Keep credentials in the vault and retain audit records connecting the evidence, agent recommendation, reviewer, approved payload, and execution result. Data visibility in the app should follow the team's account-access policy.
For throttling and transient failures, use bounded workflow retries with exponential backoff and jitter. Honor a server retry instruction when present, but do not depend on an undocumented Gainsight header. Authentication and schema errors need intervention, not repeated requests.
Duplicate events require durable keys based on the account, signal type, and evidence window. Ambiguous create responses require reconciliation before retry. External Actions can themselves retry, as their guide documents, so receiving the same call twice must not create two reviews or CTAs.
Gainsight AI Apps, Agents and Workflow Automations
- Review Amplitude usage declines before creating Gainsight churn-risk CTAs: This guide compares account-level usage with historical baselines, checks for incomplete data and seasonal changes, and routes an evidence-backed recommendation for human review. It covers duplicate prevention and an auditable task-creation process.
This is the current core step-by-step guide for Gainsight in the library. Use it when the immediate problem is adoption decline, rather than copying its implementation into a different risk model. A support escalation review needs different evidence windows and thresholds; a renewal review needs a different owner and approval policy.
For broader planning, the B2B SaaS automation collection provides related operating context. The Amplitude hub covers the analytics side, while the Zendesk hub helps frame support-system integration. These are related resources, not additional Gainsight build guides. Choose the process first, then confirm the required records, access, and review controls before adding another integration to the workflow.
Frequently asked questions
What is Gainsight AI?
Gainsight AI includes native capabilities such as Copilot and Staircase AI features. Copilot answers questions using supported Gainsight context, while Ask Staircase, Risk Analyst, and Expansion Analyst analyze customer communications for insights, possible risks, and growth opportunities.
Does Major replace Gainsight Copilot?
No. Gainsight Copilot helps users work with supported Gainsight context. Major adds deterministic apps, agent judgment across connected systems, managed state, and workflow approvals around the broader customer success process.
Does Major have a native Gainsight connector?
No. Gainsight connects through Major's Pipedream connector, app pgainsightnxt. IT configures the connection once, keeps credentials in the vault, and limits which apps, agents, and workflows can use it.
How does Gainsight API authentication work?
The documented object APIs accept an admin-provided access key in the accesskey header. The references do not name OAuth scopes for these operations. Configure the correct tenant URL, protect the credential, and validate the permissions available to the integration.
Can Gainsight automation create CTAs and Success Plans?
Yes. Gainsight documents create APIs for both CTAs and Success Plans. A Major workflow can submit an approved proposal when its connector supports the operation, required fields are valid, and duplicate checks have passed.
What are Gainsight API rate limits?
The API overview recommends 100 synchronous calls per minute and 100,000 daily, while the CTA reference lists 50,000 daily. Confirm your tenant's allowance rather than assuming the larger number. Rate-limit response-header names are not documented in the cited references.
Sources
- Gainsight API documentation overview: supported API families, synchronous and bulk patterns, and recommended fixed-window limits.
- Company API documentation: tenant host examples, access-key authentication, Company querying, matching keys, and updates.
- Call to Action API documentation: CTA retrieval, creation, updates, pagination, and published request limits.
- Success Plan APIs: plan retrieval and creation, required fields, external references, and page-based fetching.
- Data Management APIs: object descriptions and field metadata for validating tenant-specific mappings before writes.
- Copilot overview: supported sources, conversational actions, access requirements, and documented analysis boundaries.
- AI features in Staircase AI: Ask Staircase, Risk Analyst, Expansion Analyst, and their communication sources.
- Configure External Action from Gainsight: outbound calls, configurable authentication and headers, and retry-count settings.
- Copilot API documentation: instance-specific access keys, matching sandbox and production URLs, and API permission considerations.
Start building on Major
Build a Gainsight account-review app that keeps evidence, agent recommendations, human approvals, and execution receipts together. Start with one read-only portfolio review, then add approved writes after testing.
About the author
Ritvij Gautam is Head of Growth at Major.
Last reviewed November 20, 2026 by Ritvij Gautam.
Frequently asked questions
- What is Gainsight AI?
- Gainsight AI includes native capabilities such as Copilot and Staircase AI features. Copilot answers questions using supported Gainsight context, while Ask Staircase, Risk Analyst, and Expansion Analyst analyze customer communications for insights, possible risks, and growth opportunities.
- Does Major replace Gainsight Copilot?
- No. Gainsight Copilot helps users work with supported Gainsight context. Major adds deterministic apps, agent judgment across connected systems, managed state, and workflow approvals around the broader customer success process.
- Does Major have a native Gainsight connector?
- No. Gainsight connects through Major's Pipedream connector, app p_gainsight_nxt. IT configures the connection once, keeps credentials in the vault, and limits which apps, agents, and workflows can use it.
- How does Gainsight API authentication work?
- The documented object APIs accept an admin-provided access key in the accesskey header. The references do not name OAuth scopes for these operations. Configure the correct tenant URL, protect the credential, and validate the permissions available to the integration.
- Can Gainsight automation create CTAs and Success Plans?
- Yes. Gainsight documents create APIs for both CTAs and Success Plans. A Major workflow can submit an approved proposal when its connector supports the operation, required fields are valid, and duplicate checks have passed.
- What are Gainsight API rate limits?
- The API overview recommends 100 synchronous calls per minute and 100,000 daily, while the CTA reference lists 50,000 daily. Confirm your tenant's allowance rather than assuming the larger number. Rate-limit response-header names are not documented in the cited references.