However, over the last 12 months or so, I have started to change my mind about this.

Clearly SaaS doesn't mean no APIs, and these would be essential in many areas. There could also be data protection issues with some data being held by 3rd parties (and other legal issues – for example if data was being held in a different country etc.)

However, I believe that we are starting to see a change. I asked a question about University Libraries using a hosted model for SFX at a recent meeting, and was suprised to find that Cambridge have opted for this model.

I'd be very interested in trying a SaaS in a low risk arena for the library, and seeing how it worked.

It looks like several vendors are offering their next gen search products as SaaS. We saw III yesterday and talked about this, but in this case the current lack of API to flexibly reuse data was offputting.

Lorcan — I was thinking about asking that question, but the Enterprise session was running waaay over time and lunch was awaiting!

The first Enterprise release will be SaaS only, and it would certainly make sense for SD to aggregate user supplied reviews, comments, tags, etc. Whether or not they've considered that, I don't know.

My main concern with SaaS (which was echoed by every academic library admin I spoke to) is that it divorces you from your data and limits your ability to customise the product in ways SD either don't want you to (for support reasons) or in ways they haven't thought of. We've investigate SaaS style models for some of the other systems at Huddersfield (e.g. the VLE) and rejected it.

I don't remember Talin mentioning that SaaS Enterprise would have an API. Without that, how do you even start to integrate a SaaS system with your internal systems (e.g. finance, student records, etc)?

Try https://ibistro.cambridgeshire.gov.uk or http://ilink.gcal.ac.uk/ the book title does the same thing as the Details button.

Subtle, but functional?

I am interested to know whether there is any attempt here to build higher level services over the 'serviced' instances. For example, to aggregate circulation data for recommendation along the lines you have done locally.