diff --git a/lib/timeseries/hoarder/sql/manyagg.sql b/lib/timeseries/hoarder/sql/manyagg.sql
new file mode 100644
index 0000000..cf8287b
--- /dev/null
+++ b/lib/timeseries/hoarder/sql/manyagg.sql
@@ -0,0 +1,89 @@
+CREATE OR REPLACE FUNCTION public.aggregate_autoselect(
+	root_table character varying,
+	data_interval interval,
+	filter_query character varying,
+	agg_pairs jsonb,
+	metric_statement character varying DEFAULT 'tags',
+	agg_basename character varying DEFAULT NULL,
+	agg_schema character varying DEFAULT 'public',
+	time_column character varying DEFAULT 'time')
+    RETURNS SETOF RECORD
+    LANGUAGE 'plpgsql'
+    STABLE
+	PARALLEL SAFE
+AS $BODY$
+DECLARE
+	selected_table record;
+	
+	selected_column character varying;
+	selected_aggregator character varying;
+BEGIN
+	agg_basename := coalesce(agg_basename, root_table);
+
+    WITH table_selection AS (
+	  SELECT table_name, column_name, 
+		coalesce(substring(table_name, agg_basename || '_\"%\"', '\'), '0')::interval AS table_interval
+	  FROM information_schema.columns
+	  WHERE table_schema=agg_schema 
+		AND ((table_name = root_table) OR (table_name LIKE (agg_basename || '_%')))
+		AND agg_pairs?column_name
+	)
+	SELECT table_name, column_name
+	INTO selected_table
+	FROM table_selection
+	WHERE table_interval <= data_interval
+	ORDER BY table_interval DESC
+	LIMIT 1;
+	
+	IF NOT FOUND THEN
+	  RETURN;
+	END IF;
+	
+	selected_aggregator := agg_pairs->>selected_table.column_name;
+	
+	RAISE NOTICE 'Using table %, column %, aggreagtor %', selected_table.table_name, selected_table.column_name, selected_aggregator;
+	
+	RETURN QUERY EXECUTE format($qry$
+		SELECT time_bucket($1, %I::timestamptz) AS time, (%s)::text AS metric, (%s) AS "value"
+		FROM  %I.%I
+		WHERE %s
+		GROUP BY 1, (%s)::text
+		ORDER BY time
+	$qry$, time_column, metric_statement, selected_aggregator, 
+		   agg_schema, selected_table.table_name, 
+		   filter_query, 
+		   metric_statement)
+	USING data_interval;
+END;
+$BODY$;
+
+CREATE OR REPLACE FUNCTION public.percentile_autoselect(
+	root_table character varying,
+	data_interval interval,
+	percentiles double precision[],
+	filter_query text,
+	metric_statement character varying DEFAULT 'tags',
+	agg_basename character varying DEFAULT NULL,
+	agg_schema character varying DEFAULT 'public',
+	time_column character varying DEFAULT 'time')
+	RETURNS TABLE ("time" TIMESTAMPTZ, "metric" TEXT, "percentile" DOUBLE PRECISION, "value" DOUBLE PRECISION)
+	LANGUAGE 'plpgsql'
+	STABLE 
+	PARALLEL SAFE
+AS $body$
+DECLARE
+BEGIN
+
+	RETURN QUERY 
+		WITH ts_data AS (
+			SELECT * FROM aggregate_autoselect(root_table, data_interval, filter_query,
+											  $${"value": "tdigest(10, value)", "tdigest": "rollup(tdigest)"}$$::jsonb,
+											  metric_statement, agg_basename, agg_schema, time_column)
+				r("time" TIMESTAMPTZ, metric TEXT, value tdigest)
+		)
+		SELECT ts_data."time", ts_data.metric, up.percentile, approx_percentile(up.percentile, ts_data.value) AS "value"
+		FROM ts_data
+		CROSS JOIN unnest(percentiles) AS up(percentile)
+		ORDER BY ts_data."time";
+END;
+$body$;
\ No newline at end of file