I feel that complex workflows are best served by dedicated integrations, but it takes time for developers to build each integration natively into a product. We launched https://kloudless.com to simplify the process for app developers to natively integrate an entire category of SaaS services at once (I'm a co-founder).
Ignoring platforms like Salesforce, some of the "product" apps that we see adopting native integrations instead of directing users to Zapier include apps that require:
1) synchronous interactive integrations and not just background automation. e.g. a user pulling a lead from Salesforce into a marketing app.
2) complex/multi-step integrations that offer way better UX and much lower support costs to just build natively for users.
3) integrations as a competitive differentiator, or without requiring users to pay for third-party tools.
4) integrations without an API of their own to integrate to.
That said, there are always a long tail of apps and use cases so I'm certain we'll see more specialized tooling pop up to handle specific complex workflows.
Ignoring platforms like Salesforce, some of the "product" apps that we see adopting native integrations instead of directing users to Zapier include apps that require:
1) synchronous interactive integrations and not just background automation. e.g. a user pulling a lead from Salesforce into a marketing app.
2) complex/multi-step integrations that offer way better UX and much lower support costs to just build natively for users.
3) integrations as a competitive differentiator, or without requiring users to pay for third-party tools.
4) integrations without an API of their own to integrate to.
That said, there are always a long tail of apps and use cases so I'm certain we'll see more specialized tooling pop up to handle specific complex workflows.