A business owner calls and says they need a new app. Often what they need is a reliable bridge between two tools that already exist.
I have seen this at ecommerce merchants, enterprise teams, and small offices with six employees. The pain is real. The greenfield project is not always the honest answer.
Integration is a product decision
Connecting Store A to Warehouse B is not "just plumbing." It decides what syncs in real time, what waits for a nightly job, who gets alerted when a row fails, and what happens when someone edits the same record in two places.
Skip those questions and you get a script that works until volume doubles, or until a password rotates, or until someone adds a required field nobody documented.
Good integration work starts with a map: systems, owners, failure modes, and the one report everyone still exports to CSV because they do not trust the sync yet.
APIs are contracts, not magic strings
REST, webhooks, vendor SDKs - they are all the same idea at the desk: two sides agree on shape and timing. My job is to make that contract visible. What fields move. What errors mean. How to replay a bad day without corrupting production.
At TPIsoftware I worked around API platforms and enterprise products where those contracts were long-lived. At Lucid I wired storefronts to external services merchants depended on. The same discipline applies whenever a team outgrows manual copy-paste between tabs.
The stack changes. The discipline does not.
When a new app actually makes sense
Sometimes integration is a band-aid on a broken process. If the workflow itself is wrong - too many manual approvals, no source of truth, permissions that make no sense - a new system can be the right call.
I try to say that plainly in discovery. Building is what I do for a living. I still lose work by recommending the smaller path when it is the truthful one. That is cheaper for the owner and better for the relationship.
A practical test
Before you fund a new build, ask:
- Do we have two systems that already hold the data we need?
- Is the pain mostly movement and timing, not missing capability?
- Can we ship a thin integration in weeks and learn before we commit to a platform?
If yes, yes, yes - you probably do not need a slide deck about your "digital transformation journey." You need a reliable pipe and someone who will answer when it clogs.