The first build-versus-buy problem I remember
Long before I worked with enterprise technology, I wanted a Monopoly set. A cousin had one, and my parents told me it was too expensive for us.
So I built my own, with paper, glue and improvised pieces. I changed the board and added rules. My younger brother and I played it for hours, and he still remembers it.
That memory is one reason the changing build-versus-buy question interests me. Making something yourself can create access and fit that buying does not. It also makes you responsible for whatever breaks. A handmade board was easy to repair. An enterprise workflow comes with customers, data, support and obligations that cannot be solved with another piece of paper.
AI-assisted development can make a narrow internal tool easier to create. The executive decision is whether that advantage survives deployment and years of ownership. A convincing first version is the beginning of that comparison.
Start at the renewal, where the decision is real
A subscription provides more than screens and features. Depending on the product, the company is buying security maintenance, integration, service continuity, assurance and a support organisation. It may also be paying for unused seats, bundled functionality or a workflow that never fitted the business particularly well.
Those are questions about a specific contract. Broad claims that SaaS is dying obscure the useful distinction between a replaceable internal utility and a deeply embedded system of record. Feature resemblance does not establish replacement equivalence.
I would begin with an upcoming renewal and the work the software supports. Who uses it? What outcome does it enable? Which integrations and records depend on it? What does the business actually spend to run the workflow, including effort outside the product?
That exercise can reveal a better procurement option before any new code is written. Removing unused licences, simplifying the process or changing the contract may produce value with less risk than rebuilding.
A quick build and a durable service are different commitments
The AI Model Benchmarking Dashboard is an example of a decision-tool project in my portfolio. It is useful for exploring how AI-assisted development changes prototyping. It should not be treated as an audited comparison with the delivery cost of a production enterprise platform.
The Helion operational intelligence concept and Pipesignal project concept similarly explore possible interfaces and workflows. They illustrate design possibilities. They do not establish measured savings or revenue from replacing a SaaS service.
Before an internal build becomes an operating service, someone must own permissions, data changes, failure handling, support and the next release. The cost of those responsibilities belongs in the comparison. A tool dependent on one enthusiastic employee is exposed even if its first version was inexpensive.
AI can help with tests and documentation as well as code. The business still needs evidence that the resulting service behaves correctly under its real workload, including uncommon cases. That is where the apparent build advantage is either confirmed or reduced.
Put the costs on the same horizon
Consider a hypothetical internal workflow with a three-year renewal decision. Use the following illustrative figures to compare the options.
Assume the subscription costs CHF 120,000 annually, or CHF 360,000 over three years. An internal option is estimated at CHF 150,000 to build and CHF 60,000 a year to run and maintain. That gives CHF 330,000 before transition costs. Add CHF 50,000 for migration, testing and retirement of the old service, and the total becomes CHF 380,000.
The apparently cheaper build is now more expensive on those assumptions. It could still be preferable if it solves an important fit problem, improves an outcome or reduces dependency. Those benefits need their own evidence. Conversely, maintenance could prove higher or the vendor could offer a better renewal price.
The sponsor should see ranges, timing and the conditions that reverse the decision. Include internal staff time, security review, incident support, model consumption and the cost of leaving either option. Avoid comparing a vendor's full-service price with an internal prototype's creation cost.
Four useful responses
I use four responses to organise the renewal conversation. They are practical categories, not a validated scoring model.
Rent where the service is effective, defensible and worth its full cost. The decision can include fewer seats or different terms. Keeping a good platform is an active allocation choice.
Replace a bounded workflow where the organisation can demonstrate better fit or economics and sustain the service. Start with a limited test while the existing process remains available. Approval to test is smaller than approval to retire.
Wrap a system when its records and integrations are valuable but the interaction or work around it needs improvement. An agentic layer adds its own permissions, monitoring and exception handling. Keeping the underlying platform does not remove those responsibilities.
Defend capability that would be costly or consequential to disrupt. Protect critical records, integration knowledge and service continuity while reviewing whether the surrounding commercial terms still make sense.
The Build vs. Buy vs. Partner Framework can structure those alternatives. The AI ROI Calculator helps expose assumptions, but the accountable owner has to validate the inputs and the route to realised benefit.
A different billing unit still needs scrutiny
Suppliers may price by seats, usage, actions or completed work. Each unit distributes cost and risk differently. Outcome pricing sounds attractive until the buyer asks what counts as a resolved case, who validates it and whether corrections are included.
A low price per action can become expensive when a workflow needs retries, extensive context and human review. A seat price can be efficient when an expert uses a mature tool heavily. The evaluation should follow completed, acceptable work and total operating cost rather than the fashionable billing label.
Ask for auditability and an exit route alongside the price. If the supplier changes the model or service, the business needs to know whether quality, cost or responsibilities change. Contract flexibility has value, but it should be weighed against discounts, service commitments and switching effort.
The useful question before signing
Bring one renewal to a joint discussion between the business owner, technology and finance. Put the current workflow, viable alternatives and three-year assumptions on the same page. Name the service owner for an internal option and the evidence needed to retire the existing process.
The decision might be to keep the subscription. It might be to test a replacement or wrap the platform. What matters is that the choice has been made against a credible alternative, with the work of ownership included. Cheaper creation changes the options. Leadership determines which option the business can live with.