Enterprise AI Assistants: What Makes the Offering Stand Out

A promise of consolidation can carry the risk of lock-in.

Gartner is accustomed to making this remark in its Magic Quadrant—especially as it focuses its analyses on platforms. The one dedicated to enterprise AI assistants is no exception.

The firm groups in this category “AI-first applications powered by at least one generative model.” The main ones are ChatGPT Enterprise, Claude Enterprise, Copilot and Gemini Enterprise. Their respective providers (OpenAI, Anthropic, Microsoft and Google) are nonetheless classified as “Leaders” in this market.

Almost all present their offering as a “one-stop shop” for employees. Gartner invites readers to weigh the pros and cons of this approach by examining, beyond the lock-in phenomenon, the risk of overlap with existing tools and the efforts required for change management.

The choice of an enterprise AI assistant also depends heavily on how closely it fits a productivity suite, and on the willingness—whether or not—to delegate agent management to it, typically offered as an add-on module.

In any case, committing to multi-year engagements is not easy, given the difficulty in predicting costs, rapid functional evolution and the varying maturity levels of providers.

Enterprise AI Assistants: 10 Vendors, 4 “Leaders”

On the functional side, the main mandatory elements to feature in the Magic Quadrant for enterprise AI assistants were:

• Native orchestration of models, tooling and data
• Security and compliance building blocks
• Centralized administration and an adoption-tracking dashboard

Many elements were optional, increasing the number of vendors classified. For example:

• Agentic actions (in addition to recommendations)
• Context engineering
• Collaborative knowledge management
• Workspaces
• Agent store and no-code agent design
• Data residency controls
• Bring your own model

The Magic Quadrant is structured along two axes. One focuses on “execution,” i.e., the ability to meet demand (customer experience, pricing, quality of products/services…). The other focuses on “vision”—that is, the strategies (sales, innovation, marketing…).

The situation on the “execution” axis:

The situation on the “vision” axis:

From MCP to Cowork, Anthropic laid the groundwork…

In addition to developing its LLMs, Anthropic has established, with Claude Cowork, a standard for “agentic knowledge work.” It is also behind MCP. Gartner praises the notoriety of its Claude Enterprise offering and its capabilities both in software development and collaborative work. It also highlights its capacities for long-horizon tasks, the level of integration with Microsoft applications, and the emergence of sector-specific variants.

… but it struggles to keep up with demand

Sales and support activities struggle to meet demand. As a result, concerns have arisen about SLAs. Data governance still largely relies on integrations—for audit and DLP, among others. Also worth noting is hosting, which is limited to the United States, unless one goes through third-party providers—giving up some Claude Enterprise features. Gartner also notes that Anthropic is one of the rare providers here that charges both per-seat licenses and consumption-based pricing. That choice, at scale, can complicate budgeting.

Google, ahead on integration…

With Gemini Enterprise, Google has managed to convey a unified-platform message. It opens to Microsoft 365 data and builds a catalog of connectors. Underneath, there is a level of integration (chips, infrastructure, models, tools) that few competitors match. There is also a pricing model Gartner calls “relatively simple,” with license pooling into a global quota and the inclusion of elements like search in the base SKU. Governance of agents without extra cost is another differentiator.

… but lagging on collaboration and computer use

Despite its openness, Gemini Enterprise offers more productivity features with Google Workspace. Beyond that, there are no native meeting summaries, for example. Google also lags behind the other “Leaders” in both computer use and collaborative work. Antigravity has only recently extended beyond its original scope, which was coding. There is also lag in the offer’s verticalization: few native capabilities.

Microsoft can capitalize on its footprint…

With about 30 million paid seats, Copilot is the most deployed enterprise AI assistant on the market. Microsoft can capitalize on its installed base and its partner network. It can also leverage integration into Office apps. And on context connectivity via the Work IQ layer. Gartner also notes that investments in OpenAI enabled Microsoft to quickly establish a market presence. Meanwhile, the partnership with Anthropic has enabled bridges to Copilot Cowork.

… but it lacks a referential LLM

The unique dependency among the market’s Leaders: a Microsoft 365 subscription is required to use Copilot. Microsoft also, relative to its main competitors, has not yet developed a foundation model that is “fully capable of carrying its offering,” to borrow Gartner’s wording. Vigilance is also warranted on licensing, increasingly complex. The coupling with Cowork helps, as does the gateway to Agent 365. The E7 package streamlines things but increases the lock-in risk.

OpenAI, a coherent alternative to productivity suites…

Beyond ChatGPT’s user familiarity, OpenAI demonstrates growing maturity in B2B. It has formed hosting partnerships (notably with Microsoft and Amazon), while building a base of consultants and integrators. Gartner also praises a “coherent” offering judged sufficiently exhaustive to constitute an alternative to productivity suites, without the lock-in risks that come with them.

… but limited on governance and integrations

OpenAI’s scalability remains to be proven, with hyperscalers still its main customers. Its sectoral approach remains clearly focused on IT workloads. As for integrations, they primarily target productivity solutions. Support for ERP systems such as SAP, ServiceNow and Workday is more limited. Gartner also notes limits on DLP and agent governance, including execution.

AWS is functionally exhaustive…

AWS stands among the “challengers” in this Magic Quadrant.

Its Amazon Quick offering has succeeded Amazon Quick Sight and Amazon Q Business. Gartner praises its exhaustiveness (chat, enterprise search, automated workflows, BI, orchestration and agent development…). It also notes the choice of models and data-residency/processing options and the integrations with other AWS services, as well as the price/performance of the Nova models for small workloads.

… but must prove its UX and productivity

In the enterprise AI assistants segment, AWS does not enjoy the same brand recognition as the main providers. That may imply a need to invest in user training and change management. Collaborative features remain in development: you may find better in competitors more tightly integrated with productivity suites. AWS must also demonstrate the ability to deliver a coherent UX, with Quick resulting from bundling multiple products.

Glean, ahead on the knowledge graph…

Glean is also classified among the “challengers.”

It differentiates itself on the knowledge graph (it originated as an enterprise search specialist). This bedrock allows it to serve as a contextual layer for competing solutions. Gartner praises deployment flexibility (on-premises, in public cloud, or in managed private cloud on AWS, Azure or GCP). It also notes broad adoption among many clients, who tend to commit long-term.

… but behind on agentic capabilities

Glean trails the competition on both agentic capabilities and on computer use and collaboration. Many of its connectors (SAP, ServiceNow, Workday…) are read-only. Overall, it remains a “small” vendor. This shows in its partner ecosystem, and in its customer base, concentrated largely in North America. Elsewhere, local support resources can be scarce.

ClickUp, praised for pricing and UX…

In the Magic Quadrant for enterprise AI assistants, ClickUp is categorized among the “niche players.”

It also stands out for its knowledge graph, which is ideal for those already using ClickUp. The UX is another strong point, with a no-code agent-design interface that Gartner finds easy to use. ClickUp also offers straightforward pricing and attractive discounts, even for small deployments.

… but its scalability remains to be proven

Even though Brain is an autonomous SKU, it remains tightly bound to the rest of ClickUp. The latter does not have the scale of the major providers, especially since it recently cut staff as part of a restructuring. As for the solution’s scalability, it remains to be demonstrated: few clients commit long-term, and most deployments are small-scale. Gartner also notes the absence of sovereign hosting, despite some tenancy options at the workspace level.

A back-office specialism at ServiceNow…

ServiceNow also figures among the “niche players.”

It differentiated its EmployeeWorks offering by bridging with its HR and ITSM solutions, focusing its strategy on integrating ERP systems and back-office workflows rather than productivity apps. Like others, it can capitalize on its installed base and partner network. Another advantage is the ability to integrate with a wide range of tools (communication, intranets, self-service portals…) to reach the user wherever they are.

… but a roadmap that needs clarification

Beyond the IT/HR/support focus, which sidelines “agentic knowledge work” relative to the competition, EmployeeWorks lacks collaborative AI capabilities. The options to create and work on documents, presentations and videos are limited. The product, stemming from the Moveworks acquisition, overall lacks a distinct roadmap that would address, in particular, coexistence with Otto, ServiceNow’s other AI assistant.

Superhuman, extensible and “proactive”…

Another niche player: Superhuman, formerly Grammarly.

Its Superhuman Go offering stands out by covering both web and desktop, reducing context switching. Gartner appreciates the use of inline triggers to provide feedback to users during content creation. It also praises the extensibility options, strengthened by dedicated support for key connectors.

… but not at the level of governance and cost management

Superhuman is not at the governance and cost-management level of its rivals (no DLP or agent lifecycle management) and has limited cost controls (no granular per-group controls). Its business is largely self-service sales, with most contracts under one year. Few customers deploy it extensively. There is only one hosting option: AWS in North America.

Zoom has its edge in communication…

Zoom is also a niche player in this Magic Quadrant.

As part of its ZoomMate offering, it has leveraged its strength in meetings and collaboration. It can thus incorporate communication signals that many rivals cannot access natively. Gartner also notes the comprehensiveness of its catalog of connectors for ERP systems.

… but it does not stand out in other areas

While it differentiates in the “conversation graph,” Zoom does not stand out elsewhere. One might be tempted to look elsewhere for the agentic component or for broader ecosystem integrations. Also beware of hosting limited to the United States and official English-language support only.

Mistral AI noted for… hosting Vibe in Europe

Gartner says it includes only the vendors most relevant for its clients in its analysis, but it does mention five others. Namely, Alibaba (QwenWork), Tencent (WorkBuddy), Perplexity, SpaceXAI… and Mistral, whose Vibe offering’s first distinguishing element is hosting options in Europe…