diff --git a/TODO.md b/TODO.md
index f144b863542..aabf40669df 100644
--- a/TODO.md
+++ b/TODO.md
@@ -1,30 +1,39 @@
 # TODO
 List of TODO items for Panoramix
 
-## Improvments
-* Read dashboard filter from URL
-* Table description is markdown
-* Animated scatter plots
-* Filter widget
-* Default slice instead of default endpoint
-* dashboard controller + filters
-* Color hash in JS
-* Widget sets ()
+## Features
+* **Dashboard URL filters:** `{dash_url}#fltin__fieldname__value1,value2`
+* **Browser history in explore.html:** use location.hash to manage query history
+* **Default slice:** choose a default slice for the dataset instead of default endpoint
+* **Color hash in JS:** it'd be nice to use the same hash function for color attribution of series
+    on the js side as on the python side (`panoramix.utils.color`)
+* **Widget sets / chart grids:** a way to have all charts support making a series of charts and putting them in a grid.
+    the same way that you can groupby for series, you could chart by. The form fieldset would be common and use
+    a single field to "grid by", a limit number of chart as an N * N grid size.
+* **Free form SQL editor:** Having an Airpal-like easy SQL editor
+* **Advanced dashboard configuration:** define which slices are immune to which filters, how often widgets should refresh, 
+    maybe this should start as a json blob...
+* **Getting proper JS testing:** unit tests on the Python side are pretty solid, but now we need a test
+    suite for the JS part of the site, testing all the ajax-type calls
+* **Annotations layers:** allow for people to maintain data annotations, 
+    attached to a layer and time range. These layers can be added on top of some visualizations as annotations.
+    An example of a layer might be "holidays" or "site outages", ...
+* **Worth doing? User defined groups:** People could define mappings in the UI of say "Countries I follow" and apply it
+    to different datasets. For now, this is done by writing CASE-WHEN-type expression which is probably good enough.
+
+## Easy-ish fix
 * datasource in explore mode could be a dropdown
-* [sql] make "Test Connection" test further
+* Create a set of slices and dashboard on top of the World Bank dataset that ship with load_examples
+* [sql] make "Test Connection" test further, run an actual dummy query
 * [druid] Allow for post aggregations (ratios!)
 * in/notin filters autocomplete
 
-## Better Javascript enables
-* Async on Druidify! in exploration page
-* Configurable widget auto refresh in Dashboard view
-
-## New Features
-* Annotations layers
-
-## Low value
-* [sql] define column based grouping
+## New viz
+* Animated scatter plots
+* Horizon charts
+* Chord diagram
+* ...
 
 ## Community
-* Creat a proper doc
+* Creat a proper user documentation (started using Sphinx and boostrap...)
 * Usage vid