Build your own tool only if its upkeep will cost less than the subscription. AI makes the first version cheap, but maintenance is usually most of what software costs over its life. For a 50-person startup, a homemade Figma has to need less than about two hours of engineering time a week to beat buying it.

Why building your own tools looks cheap now

AI has made the first version of almost any internal tool quick to build. A team that used to wait months for engineering time can now prompt a working dashboard, review space or workflow in a few days or even hours.

Companies are acting on it.

…32% of respondents said their organization had decided against buying...

In McKinsey's 2026 survey, 32% of respondents said their organization had decided against buying at least one software product or feature because they could build it in-house with agentic coding tools [4].

Retool's survey of 817 builders found that 35% of teams have already replaced at least one SaaS tool with something they built, and 78% plan to build more this year [7] (Retool sells a platform for building internal tools, so treat its numbers with a grain of salt).

The motivation is easy to understand.

Zylo puts the median SaaS spend at $9,455 per employee a year [13]. When every tool is billed per seat, building your own looks like a way out of a bill that grows with every hire.

Building is the cheap part

Most of the cost comes after the tool works. AI makes each fix faster to write, but someone still has to notice the problem, decide what to change, check it works and handle requests from colleagues. That time adds up, and it usually lands on one person.

And that person is easy to lose.

…when the builder changes teams or leaves, looking after the tool becomes nobody's job.

In Retool’s survey, 60% of builders said they had shipped tools outside IT oversight in the past year [8]. However, when the builder changes teams or leaves, looking after the tool becomes nobody's job. You also lose the person who knew how it was meant to be used. Then everyone uses it however they like, and it gets messy. Fast. Think of most company Notion workspaces.

The hours have a second price. At a startup, every hour an engineer spends on an internal tool is an hour not spent on the product your customers pay for. That cost doesn’t show up on any invoice, so it’s easy to ignore.

If the tool touches customer data, it also has to meet GDPR and your own security rules, which bought tools usually come ready for.

Buying isn’t free either

Subscriptions get more expensive every year. Vertice, which tracks software spend for its customers, measured software prices rising 14.3% year on year in September 2026 [12]. Per-seat plans also grow with every hire, so a bill that’s fine at 20 people can look very different at 60.

More tools now charge for usage instead of seats. In Kyle Poyar’s 2026 survey of about 230 software companies, 37% used hybrid pricing, a seat fee plus usage or AI credits, up from 25% a year earlier [3]. That makes the bill harder to predict, because a busy month costs more.

How much upkeep can you afford?

The simplest way to decide is to turn the subscription into working days. Divide what you pay each month by what an hour of engineering time costs you. Then divide by 8 to get working days. The result is the upkeep budget your own version has to stay under. I use $100 an hour below, within the $70–120 range that senior developers bill in the US [11].

Three worked examples for a startup of about 50 people:

  • Zapier at heavy use. Around 500,000 automated tasks a month on the Team plan costs about $2,699 a month [6]. That's about 3-4 working days of upkeep a month ($2,699 ÷ $100 = 27 hours).

  • Salesforce for 10 sellers. The Enterprise plan is $175 per user per month [9], so $1,750 a month. That's about 2 working days of upkeep a month ($1,750 ÷ $100 = 17½ hours).

  • Figma for a 50-person team. On the Organization plan, 5 designers on Full seats ($55), 15 developers on Dev seats ($25) and 30 others on Collab seats ($5) come to $800 a month [1]. That's 1 working day of upkeep a month ($800 ÷ $100 = 8 hours), or about two hours a week.

These are list prices, and your own mix of seats and usage will differ.

The point is the order of magnitude, not the exact number. If your homemade version needs more than about 2 hours a week of someone’s time, buying ready-made software like Figma is cheaper.

Also, one engineer has about 20 working days a month. If every homemade tool needed 1 day of upkeep, like the Figma example, one person could look after about 20 tools. A company with 50 or fewer employees uses around 36 apps, according to Okta [5]. Rebuilding them all would keep almost 2 engineers busy full-time, just keeping the lights on.

Chart: working days of engineering time a month that cost the same as the subscription, at $100 an hour. Zapier at heavy use: 3-4 days a month. Salesforce for 10 sellers: about 2 days a month. Figma for a 50-person team: 1 day a month.

The days alone don't decide it. The real question is how they compare with the upkeep the tool will actually need, and that depends on how complex the tool is.

If your homemade version needs more than about 2 hours a week of someone's time, buying ready-made software like Figma is cheaper.

Three tools, three different answers

Heavy automation: building often wins. Workflow automations are the category teams replace most, named by 35% of Retool's respondents, followed by internal admin tools at 33% [7]. An automation that moves data from A to B is small, its rules rarely change, and when it fails, someone notices quickly. TTT Studio, an agency that builds and fixes internal tools, says AI-built tools hold up when a small team uses them, the data is structured, the rules are simple and errors are easy to spot [10]. A script like that rarely needs 3-4 days a month.

Figma: it depends on what you'd build. If your design system already lives in code and your designers prototype there, a review space with walkthroughs and comments can be built on top of it, and it can do things Figma can't, like run on your real product. Then you're paying for a better tool, not saving money. But if the goal is to copy Figma to cut the bill, two hours a week goes fast. I wrote more about this in Does your startup still need Figma?

Salesforce: buying is usually the safer choice. The budget looks bigger, but a CRM is far more than a list of deals. It has integrations with email and billing, permissions, reporting and an audit trail. Keeping all of that working is more than one person's side job. Klarna is a useful example. It made headlines in 2024 for dropping Salesforce and Workday, and many read it as "Klarna built its own". Klarna later told CX Today it had moved to a mix of other vendors' software and in-house tools [2]. Even the company that became the symbol of "just build it" still bought part of it.

The middle option: buy, then build on top. Many teams keep the bought tool and add their own small pieces: a script that connects Zapier to an internal database, or an agent that reads designs from Figma through its API. You pay for the hard part and build only what’s specific to you. Klarna’s mix of vendors and in-house tools is closer to this than to building everything.

A quick test before you build

Before anyone opens Cursor or Claude, go through these six steps. If you can’t answer one of them, that’s usually your answer.

  1. Check whether you need it at all. Some teams find that a free plan, or no tool at all, is enough.

  2. Turn the subscription into working days. Divide the monthly cost by your hourly engineering cost, then by 8. That's your upkeep budget in days per month.

  3. Estimate real upkeep honestly. Count bug fixes, updates, requests from colleagues and onboarding, not only the build. If you can't estimate it, assume more.

  4. Name the owner. Write down who maintains it and who takes over when they leave. If nobody comes to mind, buy.

  5. Check what you'd gain. Build when your version does something the bought tool can't. If it only saves money, the saving has to survive the upkeep.

  6. Start with the simple, low-risk tools. Automations and small internal dashboards are the safest place to build first. Systems of record, like your CRM or payroll, are the hardest to get right yourself.

When building is the smartest choice

Most software is built for the most common way of working, because that’s where most of the vendor’s customers are. If your way of working is different, building your own is now a real option. Think of pricing so unusual that even Salesforce CPQ would need custom code to handle it. You’ll pay for upkeep either way, but you get a tool that fits how you actually work. So it's not only about saving money.

A note on the numbers

The prices are public list prices in US dollars from late 2026. Annual contracts, discounts and different seat mixes will change your own result. The $100 hourly rate is an assumption; use your own. Nobody has good long-term data yet on how much upkeep AI-built tools need. Retool's survey covers people who already use a building platform, so it likely overstates how common building is. McKinsey's survey is broader but counts decisions to build, not whether those builds worked out. Vertice and Kyle Poyar's survey describe the software market overall, not small startups in particular.

Edita Dermontaité

Edita Dermontaité

Hands-on CPO and product leader who still builds (with AI)

I’ve often been the first: the first product hire at Leadcamp and Bizzy, where I led product as CPO for five years, and the first user researcher at Lighthouse, where I spent two years building the research function.


For the past year I’ve worked as a fractional CPO and hands-on product leader, helping B2B SaaS teams between traction and scale connect what engineering builds to what brings in revenue.


I started out as a graphic designer, so I still design and test things myself, these days mostly with AI: in Claude and Lovable, or straight in code with Cursor.


More about me →

Working through this at your company?

Book a free 30-minute call, or a coffee in Ghent. We’ll look at what’s slowing your product down and what to do about it. I work with teams fractionally or hands-on as an embedded product leader, 2–5 days a week, depending on your needs and my availability.

Product leadership for B2B SaaS between traction and scale, from a hands-on CPO and product leader who still builds (with AI). I take product decisions off the founder’s shoulders and get product, engineering and sales moving in one direction: revenue.

Edita Dermontaité speaking to an audience

Copyright © 2026 New North BV

Copyright © 2026 New North BV

Made in Belgium with ❤️

Made in Belgium with ❤️