This was Thursday's keynote. I won't go into a long description of TMMi, as I think the links below will probably give more and better information than I can.
Erik started by emphasising that there's little purpose in pursuing TMMi if you don't have the fundamentals to support this - better to put your energy into pursuing those. Decent requirements and project planning, for example, without which it doesn't much matter how great your test plan is. You also need management support, and a proper budget for process improvement - you should be running this as a project in its own right, with clear business goals to achieve (otherwise why are you doing it?), timescales, and resources allocated. (Not a couple of hours every Friday which will unavoidably get eaten up by some other project's overflow...)
I found the talk very valuable - I had no idea about TMMi when I went in, and was thinking it might be a bit detached from real life and abstract. Erik was very focused on the practical side of implementing process improvement throughout - e.g. process improvement is only good if balanced by good people, and good infrastructure (tools and test environment). If you don't have those, why aren't you working on getting those first? It must be business driven rather than model driven. Flexible rather than detailed processes - you have good people (a prerequisite) so give them scope to do their best work.
Other interesting snippets:
About 90% of organisations are level 1, and have difficulty getting to level 2 (I'm assuming this is from Erik's experience of being brought in as a consultant). Level 2 might look like there's nothing new in there (and there isn't), but a lot of organisations have difficulty in getting there. I think Erik said it usually would take about 2 years.
There have been some talks at Eurostar which I've found interesting on a personal level, but not something I'd take back to work with me. This is one of the talks where I'll definitely be taking a lot of talking points back to work.