Build vs. Rent: When Off-the-Shelf Software Doesn't Fit How You Work

The default advice is correct: rent software, don't build it.

If your workflow matches what the tool expects, renting is cheaper, faster, and maintained by someone else. Most workflows match. Sales teams sell similarly enough that a CRM covers them. Support teams ticket similarly enough that Zendesk works. Hiring teams hire similarly enough that an ATS handles it.

The advice stops applying when the match isn't there. Not "we'd prefer it slightly different." Structurally isn't there. The tool assumes a sequence your process doesn't follow. You spend more time configuring around the tool's assumptions than doing the work.

Jemini hit this with recruiting.

What didn't fit

Jemini is a consulting firm. They place specialized consultants on complex engagements for large clients. Their recruiting process differs from standard hiring in specific, structural ways:

• Evaluation periods run long. Weeks or months, not days.
• Multiple stakeholders weigh in at overlapping (not sequential) stages.
• Candidates apply, get waitlisted, and reapply or are re-invited 18 months later. The system needs to remember them.
• Interview processes vary by role type and seniority level. There's no single pipeline.
• Shortlisting is collaborative and asynchronous. Hiring managers debate candidates over days, not in a single meeting.

They looked at ATS vendors. Quotes ranged from $5,000 to $15,000 per year, plus per-user fees.

The price wasn't the main issue. The fit was. Every platform they evaluated assumed one of two models:

High-volume, standardized hiring. Designed for companies processing hundreds of applicants per role with automated screening. Think retail, logistics, large enterprise.

Lightweight startup tracking. Pipeline view, basic status updates, minimal automation. Fine for a 10-person company hiring occasionally.

Consulting recruitment fits neither. Dan, Jemini's recruitment lead, described the difference:

"There are a lot of vendors who will sell an applicant tracking system and then a talent CRM... we've kind of woven them together, which I think is an advantage."

That weaving (tracking a person across years and multiple applications while maintaining full context) is exactly what standardized tools don't support well.

The workflow that doesn't exist as a product

One part of Jemini's process had no off-the-shelf equivalent at all.

Before interviews, the hiring team reviews candidates collaboratively. Three or four people look at a pool of 40 applicants and narrow it to 8 or 10. They leave notes. They disagree. They revisit. It takes days, not a single meeting.

In Monday, the team handled this by duplicating candidate rows into separate groups and dragging them around. It worked mechanically. Every duplicated row meant the same person existed in multiple places with conflicting status labels. Data integrity eroded over time.

No ATS we evaluated supported this. Pipeline tools track linear progression: applied, screened, interviewed, offered. They don't support a lateral process where multiple reviewers curate overlapping lists from the same candidate pool simultaneously.

We built a dedicated shortlisting tool in Zite. It reads from the Airtable candidate database but gives each reviewer their own workspace. Multiple people can maintain separate shortlists for the same role. Bulk selection replaces one-at-a-time dragging.

The critical design decision: notes written during shortlisting stay at the shortlist level. They don't attach to the candidate's permanent record. Dan described the logic:

"The shortlist is a playground... We want the notes in a candidate record to be regarding our assessment of a live candidate for a role."

When shortlisting ends, the decisions (who made the cut) flow back to the main system. The working notes (half-formed opinions, comparisons between candidates, "maybe if nobody better applies") stay in the playground.

Eric reinforced the boundary:

"I don't see the need for having shortlist notes go to Airtable. It's going to be so messy. You're creating so much data for, I think, not a lot of return."

That separation between working space and system of record is something we haven't seen in a commercial ATS. The market for "collaborative asynchronous candidate shortlisting with ephemeral notes" is too narrow to build a SaaS product around. For Jemini, it changed how hiring decisions get made. It had to be built because no vendor would ship it.

What "custom" costs in 2026

The system runs on four tools:

• Airtable: relational database, automation engine, admin interface
• Zite: custom interfaces for everyone who shouldn't touch the database
• Fillout: application forms, scheduling, feedback collection
• n8n: automations and integrations, self-hosted on Elest.io for $16/month

Monthly infrastructure cost: under $200. Airtable seats are limited to the 3–5 people who manage the backend (recruiters, ops lead). Everyone else accesses the system through Zite, which charges no per-user fee. Twenty interviewers, five hiring managers, and the CEO all use Zite interfaces without additional licensing.

Compare:

By year two, the custom build runs at a fraction of the SaaS subscription. The gap widens each year after that.

Why the math changed

The "rent, don't build" advice dates from a time when custom meant hiring developers for months. That tradeoff was real. Building was slow and expensive. Renting was fast and predictable.

No-code tools shifted the math. Airtable, Zite, Fillout, n8n: these aren't toys. They're production infrastructure. The Jemini ATS handles 700+ candidate records, automated interview scheduling, multi-stage feedback collection, and role-specific application routing. It was planned, built, tested and deployed in five months, not eighteen.

Vibe coding has pushed this further. AI can generate working Airtable scripts, n8n workflows, and interface logic faster than anyone could write them manually three years ago. The building is faster than it's ever been.

But building is maybe 30% of the job.

The other 70% is figuring out what to build. Requirements gathering. Process mapping. Deciding which tool handles which job. Designing interfaces that people will actually use. Making sure the recruiter, the hiring manager, and the CEO are all satisfied with what they see.

Most clients know how work flows across their organization. They're less clear on how it should flow, and through what tools. Half the companies we work with haven't done any process mapping before we start. The story mapping sessions we run before touching a single tool are where the real value gets created. Eric put it directly:

"The single fact that you viewed your process entirely from end to end and restructured it... that would give more gains than just implementing AI in your system."

Your off-the-shelf Airtable expert or vibe-coding enthusiast can handle the 30% (the building) well. The remaining 70% (requirements, process redesign, tool selection, stakeholder alignment) requires experience and structured processes. That's the gap.

Six questions to decide

These aren't rhetorical. If three or more apply, custom is worth scoping.

1. Does your process include steps no standard tool supports?

Does your actual workflow require stages, relationships, or handoffs that you cannot configure in the tool you're evaluating? Jemini needed to track a person across multiple applications over years, with full context carried forward. Most ATS platforms treat each application as a fresh record. And no commercial product offered the collaborative shortlisting workflow described above.

2. Are you paying per-seat for people who barely log in?

If 20 people need access and 3 use the system daily, per-seat pricing taxes organizational breadth. A Zite interface gives everyone access through a shared frontend. Paid database seats stay limited to the people who build and maintain the backend.

3. Does the tool make you track things you don't need?

Enterprise software ships with enterprise assumptions. Compliance fields you don't file. Approval chains you don't follow. Reporting dashboards for metrics nobody checks. Each unused field adds friction to the people doing real work.

4. Is the integration you need unavailable or overpriced?

Jemini needed Float synced to Google Calendar. They needed application forms that create records across multiple related tables simultaneously. Off-the-shelf tools either don't offer those integrations or lock them behind enterprise tiers.

5. Would adopting the tool require changing your process?

If the demo goes well but your first thought is "we'd need to change how we do X," that's the tool telling you it was designed for someone else. Dan's take:

"No plan survives contact with reality intact. That's why I'm eager just to be in it."

He said this about launching the custom system with a live Director search. The team wanted a tool that matched their process, not one that required them to match its assumptions.

6. Do you need to own and extend your data?

SaaS vendors control your export options, your integration capabilities, and your upgrade path. With Airtable and Zite, Jemini owns every table, field, and automation. They can modify the system tomorrow without submitting a feature request.

When renting is the right call

Custom is wrong more often than it's right. Rent when:

Your workflow is standard. If you hire like most companies, an ATS will serve you better and cost less than building.

Nobody will own the system internally. Custom tools need a person who understands the logic and can modify it when things change. Without that person, rent software with a support team.

You need it working this week. Custom takes weeks to months. SaaS takes days.

Your problem is volume, not logic. If the challenge is processing 10,000 applicants efficiently, that's an infrastructure problem. Enterprise tools handle it. Airtable does not.

The decision, compressed

Your workflow matches standard tools: rent.

Your workflow is structurally different and you have someone internal to own it: evaluate custom.

Your workflow is different but nobody will maintain a custom system: rent, and budget for the compromises.

We design for cost, adoption, and scale. Getting from $2M to $10M shouldn't require enterprise-grade software at enterprise-grade prices. We build incrementally, targeting the spots where ROI makes sense first, then expanding from there.

If you want to think through how no-code projects get scoped and built, we wrote about that process here.

If you want someone to evaluate your specific situation honestly (including telling you to rent something off the shelf), that's what the Blueprint session covers. We spend 4–8 hours on story mapping and process analysis before recommending anything. $299.

Book a Blueprint →