Case study
How a SaaS company got an internal AI research tool in 1 week instead of hiring 2 engineers
A SaaS company was about to hire 2 engineers to build an internal AI research tool. We built it in 1 week instead. It took 20 hours a week of manual research off the team, with no new hires, and the company owns everything that was built.
- Trigger
- AI step
- Person decides
- Your systems
- TriggerResearch question
- AI stepInformation gathered
- AI stepFirst-pass summary written
- Person decidesPeople review the output
- to build an internal AI research tool instead of hiring 2 engineers
- 1 week
- of manual research saved
- 20 hrs/week
At a glance
- Client
- A B2B SaaS company (anonymized)
- Problem
- Manual research was taking 20 hours a week; the alternative was hiring 2 engineers to build a tool
- What we built
- An internal AI research tool
- Results
- 20 hours a week saved; built in 1 week instead of hiring 2 engineers
- Time to live
- 1 week
The client is anonymized. The figures are the real before-and-after numbers from the project, reported as-is and not rounded up.
What was the problem?
The team was spending 20 hours a week on manual research, and the plan on the table was to hire 2 engineers to build an internal tool for it.
What did we build?
An internal AI research tool that takes over the gathering and first-pass summarizing the team had been doing by hand, with people reviewing the output. It was handed over with documentation so the company’s own engineers can maintain and extend it.
What changed?
| Measure | Before | After |
|---|---|---|
| Manual research time | 20 hours a week | 20 hours a week saved |
| How it got built | Plan to hire 2 engineers | Built in 1 week, no hires |
- to build an internal AI research tool instead of hiring 2 engineers1 weekSaaS case study: to build an internal AI research tool instead of hiring 2 engineers
- of manual research saved20 hrs/weekSaaS case study: of manual research saved
What should you know before building something similar?
- Decide what a good research output looks like, with two or three real examples, before any code is written.
- Make sources visible in every answer so people can check the work quickly.
- Hand over to a named owner on your team, even if you keep support. Internal tools die without one.
Could this work for you?
If your team does similar work by hand, probably. Request a free workflow audit and you get a one-page plan: the 3 tasks worth automating first, hours saved, and what a build would cost. See also AI automation for SaaS and Custom AI agents.