Back to vendors
W

Wrike

Also known as: wrike.com, Wrike AI Agents, Wrike Copilot, Wrike MCP Server

Visit site
Entry priceFree $0; Team $10 per user a month; Business $25 per user a month, billed annually; Pinnacle and Apex quotedFull pricing detail

Wrike is an enterprise work management platform for projects, requests, resources and portfolios, and its AI agents act on that work without being asked. Space admins build an agent from a template or a chat description, give it a scheduled, event, date, API or automation trigger, and let it post comments, change fields, statuses, dates and assignees, create items from blueprints, start approval flows, and email or message people through Gmail, Outlook, Slack and Teams, with every action logged with its reasoning.

Wrike sells a work management platform used by marketing, operations, PMO and professional services teams for projects, request intake, Gantt planning, resource and capacity management, budgeting, dashboards and reporting, with Whiteboard, Datahub and Wrike Integrate as add ons. More than 20,000 organizations run work in it, among them NVIDIA, Walmart, Nickelodeon and Syneos Health. Wrike, Inc. is based in San Diego and owned by Symphony Technology Group, offers plans from a free tier to the enterprise Pinnacle and Apex editions, and Wrike Copilot answers questions about work on demand.

AI agents are the part that acts on its own. On Business, Pinnacle and Apex, space admins create agents under Space Settings, either from prebuilt templates such as intake, triage and risk status reporting or by describing the agent in a chat builder that sets its scope, triggers and actions, with a manual form for fine tuning and a Playground for testing without spending the AI allowance. Triggers run on a schedule, on events such as a new item or a field or status change, relative to a date field, when changes arrive through the Wrike API, or from Wrike Automations, which work in both directions.

Each agent reads the item and its surroundings and runs one or more actions in order: posting comments, changing custom fields, statuses, names and dates, assigning work, creating tasks, projects or folders from blueprints, moving items between folders, picking Datahub records, starting or managing approval flows, emailing through a connected Gmail or Outlook account, and notifying a Slack or Teams channel. Agents respect workflow transition rules and cannot skip required statuses. An activity view lists every agent's recent actions with timestamps, and each log entry shows the item, the trigger, the action and the full reasoning. A Wrike MCP Server connects Claude, ChatGPT, Microsoft Copilot and Gemini to the account within each user's permissions.

Team is $10 per user a month and Business $25, billed annually, with Business including a starter pack of AI Elite actions; Pinnacle and Apex are quoted with three and ten times that allowance, and extra action packs are sold separately. Wrike holds SOC 2 Type II, ISO 27001, ISO 27018 and CSA STAR, and hosts accounts in Paris for Europe or Santa Clara for elsewhere, with the EU data center selectable.

It fits operations, marketing and PMO teams already running work in Wrike who want intake, triage, assignment, status hygiene and approvals handled by agents on rules they set. It fits less well for teams that want to choose the model behind each agent, agents that work mainly in other systems, or a scored evaluation of agent decisions.

Vendor details

Canonical URL

https://www.wrike.com

Category

Enterprise operations agent

Subcategory

Enterprise work management platform whose AI agents watch projects and act on triggers to triage, assign, update, approve and notify

Funding status

Private company. Wrike, Inc., based in San Diego, has been owned by Symphony Technology Group since July 2023, when STG took full ownership from Vista Equity Partners and Elliott Investment Management.

Company status

independent

Use cases & customers

Primary use cases

request intake review and triageautomatic assignment and routing of workrisk and overdue work flagging with status reportsapproval flows started by agentsemail, Slack and Teams notifications on project changes

Target customers

PMO and project operations teamsmarketing and creative operationsprofessional services organizationsenterprises standardizing work intake and approvals

Deployment options

SaaS hosted in Paris for the EU or Santa Clara, CaliforniaEU data center selectable

Integrations

Agents act in Wrike and send email through connected Gmail or Outlook accounts and notifications to Slack channels and Microsoft Teams channels or chats. The Wrike REST API can trigger agents, Wrike Automations trigger and are triggered by them, and Wrike Integrate and Two Way Sync connect other systems. The Wrike MCP Server connects Claude, ChatGPT, Microsoft Copilot and Google Gemini through OAuth or an API key.

In practice

A marketing operations team sets an intake agent on its request form. Each new request is checked for missing details, returned to the requester if incomplete, scored, given a project from the campaign blueprint and assigned to the right designer by workload.

A PMO runs a Risk Status Reporter agent every Monday morning across its portfolio. It reads each project's dates, dependencies and comments, sets a risk field, posts a summary comment and messages the Teams channel for anything newly at risk.

A creative team's agent watches for tasks moving to Ready for Review, reads the brief, starts an approval flow with the creative lead and adds legal when the brief mentions regulated claims, then emails the client contact through Outlook once the approval completes.

Agentic Index coverage score

9.5 / 14 capabilities · 68%

Integrations & Tool Calling Full

Agents act beyond Wrike through connected accounts: they send email through a Gmail or Outlook account authorized with OAuth, deriving addresses from assignees, authors or field values, and post to a set Slack channel or to a Microsoft Teams channel, one to one chat or group chat. Changes arriving through the Wrike API can start an agent, which lets outside systems hand it work, and Wrike Integrate and Two Way Sync carry results onward.

Sourcehelp.wrike.com/hc/en-us/articles/37056703440658-ai-agent-actions-referenceread 2026-11-18

Workflow Orchestration Full

One trigger can run several actions in a set order, such as reading a brief, starting an approval with the right reviewers and emailing the client once it completes. Wrike Automations work in both directions with agents, so a rule can start an agent and an agent's field change can fire downstream rules, and teams pair two agents to wait on an approval before continuing. New items come from blueprints with their own task structure.

Sourcehelp.wrike.com/hc/en-us/articles/37056331565586-ai-agent-triggers-referenceread 2026-11-18

Knowledge Grounding & RAG Partial

Each run reads the work item that triggered it and its surroundings, including custom fields, the item's scope and approval details, and an agent picking a Datahub record can search the whole linked database by value. That context is assembled per run from Wrike's own records. Agents have no document knowledge base or retrieval index they consult, and capacity and workload data stay out of view.

Sourcehelp.wrike.com/hc/en-us/articles/37056703440658-ai-agent-actions-referenceread 2026-11-18

Human Oversight & Guardrails Partial

Agents can put people in the loop on the work, starting approval flows with chosen approvers and flagging exceptions for review, and they cannot skip required statuses in a workflow or create folders. Their own actions, from changing a field to sending an email, run as soon as the trigger fires, with no step where a person approves the agent's change before it applies.

Sourcehelp.wrike.com/hc/en-us/articles/37056703440658-ai-agent-actions-referenceread 2026-11-18

Security, Identity & Governance Full

Wrike holds SOC 2 Type II, ISO 27001 and ISO 27018 certification and a CSA STAR listing, with GDPR and CCPA compliance. Only space admins can create and configure agents, generative AI must be switched on by the account owner, Pinnacle adds extensive role management, and assistants connected through the MCP server act strictly within the signed in user's permissions.

Sourcewrike.com/priceread 2026-11-18

Observability & Auditability Full

An activity view lists every agent with its recent actions and timestamps, and each detailed log shows the item affected, the trigger event, the action performed and the full reasoning behind the decision. Status changes made by agents are logged with that reasoning, and admins can track AI usage by agent, user and feature. Any agent's past decisions can be reviewed from that log.

Sourcecommunity.wrike.com/discussion/2672/ai-agents-in-wrike-from-labs-to-general-availabilityread 2026-11-18

Memory & State Persistence Not documented

What persists between runs is the work itself: fields, statuses, comments and approvals the agent wrote on items, plus its activity log. Agents start each trigger fresh from the item in front of them and keep no memory of earlier runs, preferences or lessons with their own scope and retention. Continuity comes from the items and their history rather than from the agent.

Sourcehelp.wrike.com/hc/en-us/articles/37056331565586-ai-agent-triggers-referenceread 2026-11-18

Deployment & Data Residency Full

Accounts registered in Europe are hosted in a Paris data center and others in Santa Clara, California, and customers can select the EU data center, with later moves arranged through Wrike. Wrike runs as a hosted service only, so agents run where the account lives rather than in the customer's own infrastructure. Those two locations are the full choice.

Sourcewrike.com/priceread 2026-11-18

Prebuilt Agents / Templates / Packs Full

Prebuilt agents cover intake, triage and risk, including a Risk Status Reporter, and space admins can start any new agent from one of those templates. A public AI Agents Library on the Wrike Community holds full agent configurations with prompts, triggers and actions for anyone to copy, across content, project coordination, compliance and reporting jobs.

Sourcehelp.wrike.com/hc/en-us/articles/37087792729234-ai-agent-examples-and-use-casesread 2026-11-18

Triggers & Channel Coverage Full

Five trigger types start agents without a person: scheduled runs daily, weekly or on a custom interval; event triggers on a new item or a status or field change; date triggers a set number of days before a due date or when a start date arrives, counting working days if needed; API triggers when changes come in through the Wrike API; and Wrike Automations rules. Each trigger is scoped to the item itself or to all of its subitems.

Sourcehelp.wrike.com/hc/en-us/articles/37056331565586-ai-agent-triggers-referenceread 2026-11-18

Model Flexibility & Routing Not documented

The model and provider behind the agents stay with Wrike and go unnamed, and space admins configure prompts, triggers, scope and actions without any model setting. Teams that want a particular assistant work with Wrike from Claude, ChatGPT, Copilot or Gemini through the MCP server, which changes where they ask rather than what powers Wrike's agents, so the choice of model never reaches the customer.

Sourcehelp.wrike.com/hc/en-us/articles/37056703440658-ai-agent-actions-referenceread 2026-11-18

APIs / SDKs / MCP Extensibility Full

The Wrike REST API, served from the developer portal, reads and writes the platform's work data and can start agents when changes arrive through it. The Wrike MCP Server lets Claude, ChatGPT, Microsoft Copilot and Gemini create tasks, assign work, update project data and answer questions across spaces, connecting with OAuth or an API key, and the portal also hosts a Skills library for those assistants.

Sourcewrike.com/ai/mcpread 2026-11-18

Testing, Debugging & Optimization Partial

Before an agent goes live, admins run it in a Playground against real items to see what it would do, without spending the account's AI allowance, and refine it in chat or in the form. That is hands on checking of individual runs, and there are no scored test sets or measures of how often an agent's decisions were right. Accuracy is judged run by run from the activity log.

Sourcehelp.wrike.com/hc/en-us/articles/37055545461010-how-to-create-an-ai-agentread 2026-11-18

Browser / Computer-use Not documented

Every agent action runs through Wrike's own data model or a connected email and chat account. No agent drives a browser, operates another application's screens or controls a desktop session, so work in tools without an integration stays with people. A browser based step, such as filling a form on a supplier portal, is outside what any Wrike agent can do.

Sourcehelp.wrike.com/hc/en-us/articles/37056703440658-ai-agent-actions-referenceread 2026-11-18

The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Pricing

Free $0; Team $10 per user a month; Business $25 per user a month, billed annually; Pinnacle and Apex quoted

Per user plans billed annually, with AI Elite actions for agents included by tier and extra action packs sold separately

Free tierTrial available

Included quota

Business includes a starter pack of AI Elite actions, Pinnacle three times that allowance and Apex ten times. Items an agent's filters skip use no actions.

What is public

Wrike charges per user, billed annually. Free costs nothing, Team is $10 per user a month and Business is $25, and Pinnacle and Apex are quoted after a call with sales. AI agents come with Business, Pinnacle and Apex.

Billing mechanics

Each agent action spends AI Elite actions from the plan's allowance, and items an agent's filters skip spend none. Testing agents in the Playground uses none of the allowance.

Cost watchouts

AI agents need Business or higher, at $25 per user a month or more, and heavy agent use draws down the AI Elite allowance. Whiteboard is $15 per user a month and Integrate, Two Way Sync, Datahub and Lock are separate add ons.

Variable cost rationale

Seat prices are fixed, but agent activity spends a monthly allowance of AI Elite actions that teams top up with packs.

Additional watchouts

Team covers 2 to 15 users and Business 5 to 200.

Overage / add-ons

Extra AI Elite action packs are sold separately.

Sales call required

Mixed (some tiers require a call)

Free / trial

Free plan and a free trial

Lowest paid plan

Team, $10 per user a month billed annually, for 2 to 15 users; agents start on Business at $25

Commercial notes

Apex includes most add ons apart from Wrike Lock and the AI Elite action pack.

Missing data

Wrike gives no count of AI Elite actions per plan, only multiples of the Business pack, and no Pinnacle or Apex price.

Agentic Index verified 2026-11-18

Alternatives to Wrike

Other products that do the work Wrike does, and what each one is.

Contact us

Found a vendor we missed? Have feedback on the index? We'd love to hear from you.