Ask HN: How does your startup estimate long-term releases/roadmap using agile?

My co-founder and I are split on this issue and wanted feedback.

I'm a big fan of agile development, using Kanban and weekly sprints to prioritize software development.

Basically keeping everything in a backlog and using lean startup / customer development approaches to re-prioritize every week.

The idea being that you have NO idea what your customers want and to use each week to prioritize what to implement.

My co-founder brings up a legitimate point that we need to communicate with investors our long term roadmap and get comfortable with making roadmap time estimates.

My concern is that the longer the time estimate, the less certain we are that an important feature will actually need to be implemented as our priorities will change.

If I spend a bunch of time estimating the release then we're going to be spinning our wheels.

Yet, we can't NOT do long term predictions I think.

I'm also not certain how Facebook and Google handle roadmaps applies to startups. They're trying to solve a completely different problem. A startup has to find initial product market fit whereas Facebook/Google/etc already have it nailed.

Would love to hear what you guys think (especially from fellow entrepreneurs).