About Mercor
The human expertise behind today’s most advanced AI systems
Mercor operates an expansive AI talent network that connects experts with frontier AI labs across the globe. From lawyers, cardiologists, and researchers to software engineers, pop culture video creators, and astronomers, Mercor enables these professionals to test, review, and refine model outputs, bringing the human judgment needed for AI systems to perform better in the real world.
“Slack helps us coordinate people and agents at a scale and speed nothing else can match. It's how we go from client request to fully staffed workspace in a matter of hours.”
The challenge
Coordinating a rapidly growing network of experts required speed, scale, and secure access across every project
In less than two years, Mercor grew from $1 million to $2B in annualized revenue run rate, all while building a network of more than 60,000 experts working every week.
That growth introduced a new level of complexity. Recent projects have brought together up to 23,000 experts in a single environment, from lawyers reviewing legal outputs to researchers assessing reasoning. In many cases, clients need thousands of experts ready within hours, making speed a critical part of Mercor’s ability to deliver.
Before Mercor supported any automation, those environments had to be created manually. Project leads had to set up Slack workspaces and add participants one by one — slowing down project launch and making it harder to respond quickly to client needs.
“Provisioning was the thorn in every project lead’s side,” said Aaron Langerman, Strategic Operations Lead at Mercor. “Getting people into the right Slack workspace was the biggest challenge.”
Manual provisioning complicated access management, requiring teams to remove experts as
projects ended to maintain proper permissions and protect sensitive data. As projects scaled and tens of thousands of experts asked questions and shared insights, communication volume grew nearly tenfold in less than a year — from around 20,000 daily messages to over 180,000.
Some support requests still came through email, and without the context of a centralized environment, it was difficult for Mercor teams to keep conversations organized and work moving forward — highlighting an opportunity to bring more of these interactions directly into Slack.
“If Slack were to disappear tomorrow, it would be chaos. All of our operations are powered by this platform.”
How Mercor works better with Slack
Orchestration that coordinates people and workflows to keep projects moving at scale
Slack is at the core of Mercor’s operations. On Slack’s Enterprise plan, Mercor manages thousands of project-specific workspaces as a single system, enabling internal teams, project leads, and experts to work together across projects, requests, and support in channels.
Launch in hours, run on autopilot
Each client project now runs in its own dedicated Slack workspace, automatically provisioned by Mercor’s Team Platform — an HRIS purpose-built for the unique needs of Mercor’s talent operations. When a client request comes in, Slack’s APIs are used to create a new workspace with relevant channels. Mercor’s systems then identify best-fit experts, extend contracts, and grant access to channels based on role, expertise, and history.
“A project brief in the morning can become a fully staffed workspace by the afternoon,” said Anomitro Paul, Software Engineer at Mercor. “What used to take days of manual coordination now happens automatically in near real time. Our ops teams have come to expect that as the baseline.”
The system scales across thousands of active workspaces, allowing Mercor to manage projects as a coordinated ecosystem rather than a collection of individual instances. What once required project leads to grant access to one person at a time now happens automatically, enabling teams to rapidly launch new projects.
“Provisioning was one of the most difficult problems — getting everyone to the right place and making it automatic and seamless,” said Langerman. “Now, that’s been solved.”
The system also manages offboarding. When a contract ends, access is instantly revoked, reducing the risk that someone might retain access to confidential project information.
The integration runs on two layers of Slack’s APIs. Slack’s SCIM API keeps users and group membership continuously in sync with Mercor’s Team Platform. From there, Slack’s Admin APIs are used to create workspaces, provision channels, map groups to workspaces, and map groups to channels — letting Mercor stand up a fully configured project environment in a single automated flow.
These APIs are also used continuously throughout a project to adapt its environment as needs evolve, such as adjusting channel membership as roles shift within a project, or moving experts across workspaces as they take on new projects.
With Slack, Mercor can now programmatically create, manage, and scale thousands of workspaces as projects launch and evolve over time — turning provisioning into a repeatable, automated process that gets experts contributing within minutes of accepting a contract.
“I’ve never worked with a company that’s been this efficient on the IT side,” Langerman said. “Since everything is set up in Slack, it happens quickly and in one place.”
Orchestrating the expert economy directly in Slack
Mercor supports experts directly within Slack, creating a shared environment where internal teams, external partners, and experts collaborate directly. In most cases, Mercor uses Slack Connect to work directly with external partners and customers inside shared Slack channels.
Because Slack provides identity and workspace context automatically, support teams can immediately see who the expert is and which project they’re working on. This helps teams respond faster and reduces the confusion that often comes with email-based support.
Deploying AI agents inside the same environment
As expert support volume scaled into thousands of conversations a day, Mercor extended the same Slack-native support model with AI — deploying Maven, an in-house AI agent, directly into expert workspaces. By deploying Maven directly into Slack, Mercor built a multiplayer environment where human teammates and AI agents collaborate in the same shared space.
Today Maven handles roughly 2,000 expert support threads a day, with human teammates stepping in only on the harder edge cases.
“Slack’s agent platform let us deploy Maven at scale, with no separate auth, no separate UI, no engineering scaffolding,” said Nikhil Gautam, Product Manager at Mercor. “Today Maven resolves over 85% of expert questions autonomously. That’s more than a full day of human work saved every single day, time our team now puts into building new capabilities for experts.”
Training teams to use Slack as a system
At Mercor, project leads are trained to treat Slack as the operational system that organizes communication, requests, and workflows across teams.
“One of the chapters in our onboarding textbook is about Slack. It’s called ‘Communication Infrastructure: Slack as a Knowledge Routing System,’” Langerman said. “The goal is to get people thinking about Slack as a system. It’s how teams run and manage their work.”
Project leads learn how to structure channels, build Slack workflows to route requests, and keep conversations in shared spaces so information stays visible across teams. Instead of relying on direct messages, teams use Slack to ensure knowledge is accessible and visible, and work can move forward without bottlenecks.
Governance and data protection at scale
With an ecosystem of thousands of workspaces and tens of thousands of experts, security is critical for Mercor.
Experts join project workspaces as multi-channel guests, scoped only to the channels they need for the work they’re contracted on. Direct messages follow the same rule: an expert can only DM users they share a channel with.
Together, these constraints keep each project environment cleanly isolated. An expert working on one client’s project can’t see, search, or accidentally carry information into another’s, even when both projects live inside the same Slack Enterprise Grid. As Mercor runs more concurrent client engagements, that isolation is what keeps each client’s confidential data from contaminating another’s, enabling scalability and strengthening security.
On top of that isolation, Mercor uses Slack Data Loss Prevention policies to continuously scan for sensitive data leaks across every workspace and DM, running in the background without manual oversight.
“DLP is a super underrated feature. It scans across workspaces and DMs for things like API keys that shouldn’t be shared,” Radik said.
“We have 60,000+ users across 2,000+ workspaces, all supported by just one admin. That's the kind of operational leverage Slack has enabled us to build.”
What’s next
Expanding automation and intelligence across Mercor’s operations
Automation has already reshaped how Mercor operates. By eliminating manual workspace setup and provisioning, the team estimates they’ve saved tens of thousands of hours of project lead time.
As the company continues to scale, Radik expects Slack to remain the central system where people, systems, and workflows come together. Mercor is working with Slack engineers to implement Slack’s Multichannel Context Plane, connecting information across internal systems like Snowflake and proprietary applications so teams can surface answers directly within Slack.
With the rollout of new AI capabilities, including natural language search in Slackbot, Mercor sees an opportunity to reduce time spent tracking down context and make knowledge instantly accessible.
For example, Slackbot could recognize when an expert asks a question, search across connected channels, systems and agents, and respond in real time with the answer before a team member needs to step in. As more of these workflows run in the background, Mercor’s team can spend less time on operational overhead and switching between screens — and more time connecting exceptional experts with the world’s most ambitious AI projects.
