Subscription SaaS
Tools you pay for and someone else runs.
Commodity work where the vendor's version is good enough, systems of record, and anything you'd never want to own.
Paying for seats and features your people now get from their AI assistant.
Book a call
Your tech stack, right-sized, secured, and ready for the future. We work out which work belongs in the software you pay for, which belongs in software you own, and which your AI assistant can handle, then we get our hands dirty making it happen.
The popular story is that SaaS is dead and everyone will vibe code their own tools. That swaps a pile of subscriptions for a pile of homemade apps that someone now has to secure and maintain.
What's really happening is that more and more of the workday happens inside an AI assistant, like Claude Cowork or similar. A report, a dashboard, an approval, a lookup: work that used to need its own app can now be done by asking, as long as the assistant can safely reach the right data and follows the way your company works.
So the question changes from "which app should we buy or build?" to "what should our AI be able to reach, what is it allowed to do, and what should stay a real piece of software?" Get it wrong in either direction and it costs you money or creates risk.
Work keeps moving from the first two into the third. We decide with you what should move now, what should move later, and what should stay where it is.
Tools you pay for and someone else runs.
Commodity work where the vendor's version is good enough, systems of record, and anything you'd never want to own.
Paying for seats and features your people now get from their AI assistant.
Systems your company owns and keeps running.
Core systems, high-volume or real-time work, anything your customers touch, and anything that sets you apart.
Maintaining an internal tool that only exists to show data someone could just ask for.
An AI assistant, like Claude Cowork or similar, or Grok Bot or similar, given secure access to your data and your ways of working through custom plugins.
Work people do at their desks: reporting, analysis, drafting, lookups, approvals, and coordinating across tools. A share that keeps growing.
Connecting the AI to everything, with broad access and no controls.
You don't hand us a spec and wait for an app. We work alongside you to get your whole stack right, and we do the work ourselves.
What you pay for, what you run, what people have built themselves, what each piece touches, and who owns it.
For each piece of work: SaaS, software you own, or your AI assistant, with the cost, risk, and order of each change.
Thoughtfully designed, secure MCP connections and skills, packaged as custom plugins. Access scoped per person and role, careful handling of sensitive data, audit trails, and clear limits on what the AI can change.
Retire or downgrade the SaaS that no longer earns its place, simplify the software you own, and roll the plugins out to the people who will use them.
Help your teams get good results from their AI assistant and know what is allowed.
Your needs, your data, and the AI tools change every month. We keep the plan and the plugins current.
Our work is a few hours of focused thinking, then AI agents carrying out the plan while we oversee them. Counting hours would measure the wrong thing. And because the landscape keeps changing, there is always more to do.
So we work as part of your leadership, every week, for as long as it's useful to you. Our first workstream maps your stack and builds the plan. Every workstream after that carries out a piece of it.
A standing call to review progress, decide what's next, and look at what has changed in the AI landscape.
Questions, decisions, and requests any time between check-ins.
One thing actively in progress at a time, such as mapping your stack, moving a team's work into your AI assistant, or retiring a SaaS tool. When it's done, the next one starts.
You run the business. Nobody in-house is deciding what should be bought, what should be owned, and what your AI can take on. We make those calls with you, and we do the work.
Want to do it yourself, with expert help when you need it? Scalpels is built for that.
You spend real money on SaaS, and suspect a lot of it now overlaps with the AI tools you already pay for.
You maintain internal tools that mostly exist to show or move data.
You want your people working in Claude Cowork or similar, but nobody in-house can make it safe.
People have started building their own tools with AI, and nobody is sure what's running or what it's connected to.