Vitally

Breadth across classes with an API underneath rather than a fixed connector list. Vitally's own framing is that customer data does not live in one place and neither should the platform, with direct integration to CRM, billing systems, product analytics and the data warehouse to build a single real-time view.

Named surfaces include Salesforce and Zendesk, and Segment works in both directions, as a destination syncing groups, identifies and tracks into Vitally in real time, and as a source pushing Vitally's own customer insights back out to other tools, which is a two-way connector rather than an ingest. Calendar connection brings scheduling and attendance in.

Underneath, the REST API creates and updates Accounts, Users, Traits, Custom Objects, Health Indicators and Events, and pulls structured data into BI tools and warehouses.

Sourcevitally.io/features/vitally-api and docs.vitally.ioread 2026-09-04

The chain and the workspace it runs in are the same product, which is this platform's whole argument. Playbooks enrol accounts, organizations or users meeting defined conditions and execute a series of actions on a schedule: firing risk indicators, starting conversations, creating tasks, assigning CSMs, setting traits on accounts or Custom Objects, and emitting Segment track calls to downstream tools.

Around them Hubs organize teams and processes, Projects carry onboarding and success plans through milestones with task assignment and shared updates, and Docs hold the written work, all inside the platform so a lifecycle motion is triggered and completed without handing off to a separate work management tool. The API can enrol accounts in Playbooks from outside, so the orchestration is addressable programmatically as well as from the interface.

Sourcedocs.vitally.io automated playbooks and vitally.io homepageread 2026-09-04

The AI layer reads the customer's unstructured material, not only their metrics. Vitally states that it creates detailed customer summaries from unstructured data including transcripts, notes and tickets, and that a user can query about a customer, a meeting or a conversation to extract insights, identify opportunities and generate takeaways.

That is retrieval and synthesis over the customer's own content, sitting on top of the unified account data, Docs and Projects the platform already holds. No citation or source attribution on generated summaries is documented, so a reader cannot trace a claim back to the transcript or ticket it came from. The AI layer is also sold as a separate add-on rather than included, so the grounding capability is not present for every buyer of the platform.

Sourcevitally.io homepageread 2026-09-04

The human does the work and the platform assists, which is the product's stated premise rather than a gap in it. Vitally positions itself around the CSM being able to act without leaving the workspace, and the AI layer generates plays, drafts conversation replies and produces summaries for a person to use rather than dispatching them, so a human sits between every AI output and the customer by construction. Administrator-assigned permission levels bound what each user can reach, and Playbook actions are configured by the team that owns them.

None of that is a mechanism the customer configures over the AI itself. No approval queue, review step, confidence threshold that withholds an output, blocked-action policy or escalation rule is documented, and the oversight is structural rather than enforced, so it depends on the workflow being designed that way rather than on the platform preventing anything. The AI layer is sold as a separate add-on, so for buyers who do not take it the oversight question does not arise at all.

Sourcevitally.io homepage and vitally.io/securityread 2026-09-04

Certifications and access controls both appear on Vitally's own pages. Vitally is SOC 2 Type 2 tested and audited, with the report available on request from a named privacy address, and it offers GDPR compliance resources for customers and EU data privacy certification.

Access is governed by organization-level security in which administrators assign permission levels controlling access to settings and data, so what a user can reach is set by role rather than by convention; single sign-on is listed on every plan; and API keys are issued per workspace from the integrations settings with the secret token generated by an account administrator, which extends the access model past the console into the programmatic surface. No trust center or compliance portal is published, so the report is request-only rather than self-serve, and SAML and SCIM are not documented.

Sourcevitally.io/security and docs.vitally.ioread 2026-09-04

Visibility into the customer, nothing recorded about the automation. Health scoring and analytics show account state, dashboards report across the portfolio, and the workspace keeps the artifacts of work, tasks, Docs, Projects and conversations, visible against each account, so a manager can see what was done even if not why the system did it. Nothing traces automated behavior.

There is no log of which Playbook enrolled which account and on what condition, no record of what the AI read before producing a summary or a suggested play, no configuration change trail and no retention control over such a record. Less runs unattended here than on more autonomous products, so the gap matters less, but it is the same gap.

Sourcevitally.io homepage and docs.vitally.ioread 2026-09-04

The account workspace is the business record, not memory. Customer data, health history, Hubs, Projects, Docs, tasks, conversations and survey responses accumulate against each account, and Vitally AI summarizes and queries them. No memory the AI keeps and reads back as its own state, with a scope, lifetime or delete path, is documented.

Sourcevitally.io and vitally.io/solution/operationsread 2026-10-01

The platform runs separate EU and US clusters, so an account sits in one region or the other rather than in a single global estate.

Vitally's REST API administration documentation, as cited in an integration guide, covers the separation of EU and US clusters in its endpoint management, with the US cluster the default and the EU cluster addressed explicitly, so a customer placed on the EU cluster has their data held there. EU privacy certification is a separate matter and is not residency.

The documentation does not say whether cluster placement is chosen by the customer at signup or assigned by contract. No self-hosted, private cloud, VPC or single-tenant route is documented.

Sourcedocs.vitally.io REST API admins and vitally.io/securityread 2026-09-04

Reusable structures the customer builds, not a collection the customer picks from. Hubs organize teams and processes, Projects carry repeatable onboarding and success-plan shapes, and Playbooks are configurable automations covering standard motions such as CSM assignment, risk response and task creation, all of which are reusable once made.

Vitally's own verb is build. Playbooks are authored from conditions and actions rather than selected from a supplied set, and no template gallery, play library, prebuilt program catalog or agent marketplace is published, so there is nothing to browse and adopt.

Sourcedocs.vitally.io automated playbooks and vitally.io homepageread 2026-09-04

Signals in, a schedule underneath, and customer-facing channels out. Playbooks enrol accounts, organizations or users that meet defined conditions and then perform a series of actions, and Vitally's documentation states each action runs on a defined schedule measured from when the Playbook was activated, so the automation has a clock as well as a condition. Health scores, product usage signals and trait changes supply the conditions, and Custom Object field changes can drive enrollment too.

Channels out reach the customer rather than only the CSM: Playbook actions start conversations, fire risk indicators, create tasks and set traits, custom branded surveys go out at lifecycle moments with responses syncing back, and Segment track calls can be emitted to downstream connected tools. Inbound, the MCP server makes the platform reachable from a customer's own AI tool.

Sourcedocs.vitally.io automated playbooks and vitally.io homepageread 2026-09-04

The customer does not choose the model and no provider is named. No model list, routing policy, per-workspace selection or bring-your-own-key option is published. The MCP server is a different thing: it lets a customer connect Vitally to their AI tool of choice, which is the customer picking the interface that reads Vitally, and the opposite of choosing what powers the product. The tool of choice is the client, not the model. The AI layer is a paid add-on, so a customer is asked to buy a capability whose underlying models are not disclosed.

Sourcevitally.io homepageread 2026-09-04

A REST API with its own product page, help-center documentation and an MCP server alongside it.

Vitally publishes a dedicated API page describing a REST interface that creates and updates Accounts, Users, Traits, Custom Objects, Health Indicators and Events in real time rather than by nightly batch, pulls structured data including health scores, segments and engagement history into BI tools and warehouses, and takes actions from outside: enrolling accounts in Playbooks, creating Tasks, logging Notes and updating lifecycle stages from product events, billing changes or external systems. That is write-capable rather than read-only.

Administration is documented in the help center at docs.vitally.io, with API keys issued per workspace from the integrations settings. Separately, Vitally states on its homepage that a customer can connect Vitally to their AI tool of choice through its MCP server. Any tier gating on API access limits who can use it, not what it does.

Sourcevitally.io/features/vitally-api, docs.vitally.io and vitally.io homepageread 2026-09-04

Configuration the customer tunes against outcomes, and no test of the AI. Health scoring models are configurable and can be reshaped as a team learns what predicts churn or expansion in its own book, and Playbook conditions and actions are adjusted the same way, with results visible through analytics. The tuning targets business results. Nothing evaluates the AI layer.

There is no evaluation harness, scored test set, benchmark or regression check when a Playbook or scoring model changes, and no sandbox or dry run before an automation acts on live accounts. That matters commercially as well as technically, since the AI is sold as a separate add-on, so a buyer deciding whether it earns its price has no supplied way to measure it.

Sourcevitally.io homepage and docs.vitally.ioread 2026-09-04

Everything the platform touches arrives through an authenticated interface.

Data comes in through CRM, billing, product analytics, warehouse and Segment connections and through the REST API; work goes out through Playbook actions, conversations, surveys and tasks inside the workspace, and through Segment track calls to downstream tools; and the MCP server exposes the platform to a customer's AI client.

The product is a workspace a CSM operates, not one that operates other software on their behalf. No browser extension, portal navigation, form filling, headless session, screen or click automation is published.

Sourcevitally.io homepage and vitally.io/features/vitally-apiread 2026-09-04