Instantly cut token spend in half: introducing Merge for Workforce

For the past few years, the mandate for enterprise IT was simple: get AI into everyone's hands, fast. Buy seats, deploy coding agents, let every department figure out what AI could do for them. Token spend was the cost of discovery, and it was worth paying.

That era is over. 

AI is now a top expense at every company, and it grows every quarter. When the CFO asks for a plan, most IT teams have exactly one lever: tell employees to use less AI. Cap seats, cut tools, and set usage limits. Every option claws back productivity from teams that rebuilt their work around AI, and stalls the adoption leadership demanded a quarter ago.

We're releasing Merge for Workforce today to help enterprises lower AI costs via intelligent model routing and enforcement. By ensuring that tasks are always sent to the most efficient model for the job, teams can rein in their token spend without forcing their employees to use less AI.

How Merge for Workforce helps teams control AI costs 

It starts with a desktop client that you deploy across employees’ devices via your existing MDM. The client reports what AI is running on each machine: AI tools and harnesses that haven’t been configured to run through Merge, unapproved MCP servers, and skills employees installed on their own. Admins get the first accurate inventory of how employees are using AI, and a picture of the entire organization’s usage patterns and behaviors.  

From there, you can set AI model access by group in the Merge for Workforce dashboard. Choose which models each team can use, and where their AI tools send requests. 

For example, you might provide your support team working through routine tickets with faster, more cost-effective models from the frontier providers, or open-source alternatives that deliver similar performance at a fraction of the cost. Engineering might retain access to frontier models. Anyone who needs temporary access to additional intelligence can request approval. 

This policy then gets enforced at the device level. The Merge desktop client configures the harnesses your employees already use, so tools like Claude Code and Cursor start up approved models and connectors without needing your employees to migrate everything manually. A request to a model an employee isn't approved for gets blocked on their laptop, before AI runs. 

From there you can track spend by team, tool, and model to see which parts of the business are driving the bill. Merge for Workforce also tells you where to look first by proactively surfacing insights like usage concentration and off-policy model spend. The former helps admins separate cost from usage volume to help admins identify the highest contributors to AI spend, while the latter finds any model spend that isn’t being governed by Merge’s routing policy. 

These insights let you move light users onto more cost-effective models when the work doesn't call for more expensive ones, without completely taking AI away from anyone.

Get started today with Merge for Workforce

Companies have made massive investments in AI over the past several years, and they should be able to keep what they got out of it while getting costs under control. 

Request a demo to see Merge for Workforce run against your own fleet, or read more about managing employee AI spend here.

‍