West Monroe Built 8 Tools Instead of Buying Them
Ricardo Argüello, October 8, 2026
CEO & Founder
General summary
West Monroe, a Chicago consultancy of about 2,000 people, used ChatGPT and Codex to build a payroll error checker and a manager insights tool that a vendor would have charged more than $300,000 a year for. It now has eight tools built this way. The saving is real because each one went through months of testing and gets reviewed every few months.
- Tanya Moore, West Monroe's chief people officer, put the avoided vendor cost for the two capabilities at more than $300,000 a year
- A McKinsey survey of 1,719 executives and managers, cited by Business Insider, found about a third of organizations skipped buying at least one product or feature because they could build it with AI
- What companies are building are narrow add-ons on top of bought platforms, like Spotify's HR Bot and Twilio's Jarvis, not replacements for the system of record
- Workday CTO Gabe Monroy warned that an agent that skips a payroll compliance step leads to an audit, a regulator or a lawsuit
- Steve Wade removed Backstage from three companies; one of them saved £300K a year by replacing it with well-organized GitHub repos
Think of a building's backup generator. It costs little next to what it saves you during a blackout. But if nobody starts it once a month to test it, it won't turn over on the night you need it. An internal tool built with AI is the same deal: the saving only exists if someone owns the monthly test.
AI-generated summary
The line that stuck with me in Sarah E. Needleman’s Business Insider piece is near the bottom, after all the savings numbers.
West Monroe has built eight tools in place of vendor products. Each one went through several months of testing before launch. And they get reviewed every few months to make sure they still work and are up to date.
Everybody will quote the $300,000. I’d quote the review schedule, because that is the part doing the work.
What West Monroe actually built
The Chicago consultancy has roughly 2,000 employees. It wanted one program to check payroll for errors and another to give managers useful insight about their people. Tanya Moore, its chief people officer, estimated a vendor would have charged more than $300,000 a year for those two capabilities. So employees built them earlier this year with ChatGPT and Codex.
Notice the size of what they built. A payroll checker checks payroll. It does not run it. The system of record stays bought.
The rest of the article has the same shape. Spotify, about 7,000 employees, built HR Bot, an HR help desk that answers from the employee handbook. Twilio built Jarvis, which listens to sales calls and coaches each rep. A Twilio spokesman says about 80% of its sales teams use it, and that users close deals in a cycle 54% shorter than non-users.
And a McKinsey survey of 1,719 executives and managers, cited in the piece, found about a third of organizations decided against buying at least one product or feature because they could build it with AI.
So most companies are not ripping out Workday. Business Insider says as much: the damage lands on the incremental revenue vendors make from add-on features. That is where the build-versus-buy line moved.
I’ll admit this is a shift from what I wrote in March, when I argued that AI companies keep buying SaaS and so should you. I still believe it for the CRM. I built one at Word Magic in the 90s that ended up needing a person whose whole job was working around it. What changed is how small the buildable piece can be and still pay.
The other column of the spreadsheet
The same article lists what the saving costs. Someone has to fix bugs, ship updates, handle feature requests and keep the tool working as more people lean on it. There is also the risk that an AI tool leaks staff data.
Gabe Monroy, Workday’s CTO, put the payroll case bluntly in a June blog post: if an agent skips a compliance step or approves pay outside your grid, you are dealing with an audit, a regulator or a lawsuit.
Yes, Workday sells exactly what West Monroe stopped buying. He is still right about payroll.
Moore says it herself. Other companies may not have the expertise, or the stomach, to go down this path. That sentence is the whole story. If you can’t name who reviews the tool next quarter, you didn’t save $300,000. You deferred a bill to a date you don’t know.
Sometimes the math says delete
Steve Wade posted the same calculation from the other side. He has ripped Backstage, Spotify’s open-source developer portal, out of three companies, and each shipped faster afterward. Every service in Backstage needs its own catalog-info.yaml. Miss an update and the catalog quietly drifts. One company replaced the whole thing with well-organized GitHub repos and decent README files, and saved £300K a year.
His line: “Spotify has 4,000 engineers. You have 50.”
Spotify shows up twice here, and both times it built for its own size. Backstage for thousands of engineers. HR Bot for thousands of employees. A tool nobody owns costs money whether you built it, bought it or adopted it for free.
Before you cancel the renewal
Four questions I’d want answered first. Who owns it, by name? How many months of testing before it reaches the whole company? Does it touch money, payroll or personal data? Which exact vendor feature does it replace, and which ones do you still need from them?
That is the same list we walked through for Harvey and its open-source rival MikeOSS, just applied to any add-on. If you ask us to build one of these through IQ Source custom software development, the owner and the review date come up in the first meeting, before the model does.
Got an add-on renewal coming due this quarter? Send us the quote before you sign it, and we’ll run the numbers with you.
Frequently Asked Questions
West Monroe built a program that checks payroll for errors and another that surfaces useful insights for managers about its roughly 2,000 employees. Chief people officer Tanya Moore estimated buying them would have cost more than $300,000 a year. According to Business Insider, the firm has built eight tools this way so far.
Build an internal tool with AI when the tool is narrow, sits on top of a system you keep buying, and has a named internal owner with a review date. West Monroe tested each of its tools for months and reviews them every few months. Without that owner, the license saving becomes a maintenance cost with no date attached.
An AI-built internal tool that touches payroll can leak employees' private information or approve pay outside the salary grid. Workday CTO Gabe Monroy wrote that such an error is not treated as a minor hallucination but as an audit, a regulator or a lawsuit. That is why payroll tools need long testing and periodic review.
Steve Wade removed Backstage, the developer portal Spotify built and open-sourced, from three companies and says all three shipped faster afterward. Backstage needs a catalog-info.yaml file per service that drifts when nobody maintains it. One of those companies replaced it with organized GitHub repos and README files and saved £300K a year.
Related Articles
Anthropic Uses Salesforce. Why Don't You?
The most advanced AI companies buy SaaS instead of building it. A framework for deciding when to build, when to buy, and why the default just flipped.
Harvey vs. MikeOSS: the code is free, the risk isn't
MikeOSS matches most of what Harvey does, for free. But someone still has to carry the operational and contractual risk that Harvey takes off a client's plate.
Your SaaS Moat Is a Trench They Cross in One Quarter
Jensen, Chamath, and the products themselves all say the same thing: agents don't need your dashboard. Your per-seat pricing model just expired.
Claude Cowork: AI on Every Employee's Desktop
Anthropic launched Claude Cowork with plugins for operations, finance, and HR. What it means for B2B companies and how to get ready.
Porsche Sold Its Consulting Arm, Then Rented It Back
Porsche sold MHP to TCS for 320 million euros and signed a five-year partnership worth 1.25 billion. The accounting works. The learning walks out.
Bristol Myers Squibb Is Building Its Own AI Factory, Not Renting It
BMS is the third pharma company in nine months to build its own AI supercomputer with NVIDIA. What it rents instead is a different layer entirely.