diff --git a/src/jsonchain/__init__.py b/src/jsonchain/__init__.py
index a7ce613..f61f455 100644
--- a/src/jsonchain/__init__.py
+++ b/src/jsonchain/__init__.py
@@ -5,61 +5,11 @@
 __version__ = "0.1.0"
 
 
-import json
-import pathlib
-from typing import Any, Optional
-
-def load_json(filepath: str | pathlib.Path) -> dict | list:
-    """
-    Loads the JSON data at 'filepath' into a Python object.
-    """
-    with open(filepath, 'r') as file:
-        return json.load(file)
-    
-
-def dump_json(object: list | dict, filepath: str | pathlib.Path, indent=2) -> None:
-    """
-    Dumps the 'object' (which must be JSON-serializable) into a JSON
-    file at 'filepath'.
-    """
-    with open(filepath, 'w') as file:
-        json.dump(object, file, indent=indent)
-
-
-def extract_keys(
-        object: dict[str, Any], 
-        key_name: str,
-        include_startswith: Optional[str] = None,
-        exclude_startswith: Optional[str] = None,
-    ) -> list[dict[str, Any]]:
-    """
-    Returns a list of dicts where each dict has a key of 'key_name'
-    and a value of one of the keys of 'object'.
-
-    e.g.
-    object = {"key1": value, "key2": value, "key3": value}
-    key_name = "label"
-
-    extract_keys(object, key_name) # [{"label": "key1"}, {"label": "key2"}, {"label": "key3"}]
-
-    'include_startswith': If provided, will only include keys that start with this string.
-    'exclude_startswith': If provided, will exclude all keys that start with this string.
-
-    If both 'include_startswith' and 'exclude_startswith' are provided, exclude is executed
-    first.
-    """
-    shortlist = []
-    for key in object.keys():
-        if exclude_startswith is not None and key.startswith(exclude_startswith):
-            continue
-        else:
-            shortlist.append(key)
-
-    acc = []
-    for key in shortlist:
-        if include_startswith is not None and key.startswith(include_startswith):
-            acc.append({key_name: key})
-        elif include_startswith is None:
-            acc.append({key_name: key})
-
-    return acc
\ No newline at end of file
+from .io import (load_json, dump_json)
+from .envelope import envelope_tree
+from .tree import (
+    compare_tree_values,
+    extract_keys,
+    trim_branches,
+    retrieve_leaves
+)
\ No newline at end of file
diff --git a/src/jsonchain/envelope.py b/src/jsonchain/envelope.py
index ff33535..e4e7112 100644
--- a/src/jsonchain/envelope.py
+++ b/src/jsonchain/envelope.py
@@ -37,6 +37,14 @@ def envelope_tree(tree: dict | list, levels: list[Hashable | None], leaf: Hashab
         # Otherwise, pop the next level and dive into the tree on that branch
         level = levels[0]
         if level is not None:
+            try:
+                tree[level]
+            except KeyError:
+                raise KeyError(
+                    f"Key '{level}' does not exist at this level. Available keys: {list(tree.keys())}. "
+                    "Perhaps not all of your tree elements have the same keys. Try enveloping over trees "
+                    "that have the same branch structure and leaf names."               
+                )
             env_acc.update({level: envelope_tree(tree[level], levels[1:], leaf, agg_func, with_trace)})
             return env_acc
         else:
@@ -77,6 +85,4 @@ def min(x: list[float | int | None]) -> float | int | None:
     Returns the max value in 'x' while ignoring "None".
     If all values in 'x' are None, then None is returned.
     """
-    return PYMIN([y for y in x if y is not None])
-
-
+    return PYMIN([y for y in x if y is not None])
\ No newline at end of file
diff --git a/src/jsonchain/io.py b/src/jsonchain/io.py
new file mode 100644
index 0000000..4c67022
--- /dev/null
+++ b/src/jsonchain/io.py
@@ -0,0 +1,19 @@
+import json
+import pathlib
+from typing import Any, Optional
+
+def load_json(filepath: str | pathlib.Path) -> dict | list:
+    """
+    Loads the JSON data at 'filepath' into a Python object.
+    """
+    with open(filepath, 'r') as file:
+        return json.load(file)
+    
+
+def dump_json(object: list | dict, filepath: str | pathlib.Path, indent=2) -> None:
+    """
+    Dumps the 'object' (which must be JSON-serializable) into a JSON
+    file at 'filepath'.
+    """
+    with open(filepath, 'w') as file:
+        json.dump(object, file, indent=indent)
diff --git a/src/jsonchain/tree.py b/src/jsonchain/tree.py
new file mode 100644
index 0000000..30c2dea
--- /dev/null
+++ b/src/jsonchain/tree.py
@@ -0,0 +1,199 @@
+from copy import copy
+from typing import Hashable, Union, Optional, Any
+import operator
+import deepmerge
+
+def compare_tree_values(
+    tree_a: dict | list,
+    tree_b: dict | list,
+    levels_a: list[Hashable | None],
+    levels_b: list[Hashable | None],
+    leaf_a: Hashable,
+    leaf_b: Hashable,
+    compare_func: Union[str, callable, None],
+    comparison_key: Optional[Hashable]= None,
+    *args,
+    **kwargs,
+) -> dict:
+    """
+    Returns a dictionary tree keyed according to 
+    'tree_a': the first tree to compare
+    'tree_b': the second tree to compare
+    'levels_a': The levels to iterate through in order to access the leaf keys in
+        'leaves_a'. If a level is listed is None, then all keys at that level will
+        be iterated over.
+    'levels_b': The levels to iterate through in order to access the leaf keys in
+        'leaves_b'. If a level is listed is None, then all keys at that level will
+        be iterated over.
+    'leaf_a': The leaf in the tree_a to compare to the leaf in tree_b
+    'leaf_b': The leaf in the tree_b to compare to the leaf in tree_a
+    'compare_func': Either one of 
+        {'div', 'sub', 'add', 'mult', 'ge', 'le', 'lt', 'gt', 'eq', 'ne'}, None, or a 
+        user-supplied callable whose call signature takes the values of the individual
+        elements of 'leaf_a' as the first param, the individual elements of 'leaf_b'
+        as the second param. Optionally, args and kwargs can be passed and they
+        will be passed on to the callable.
+        If compare_func is None, then no function is called and the values for
+        comparison are both entered into the returned dictionary but without
+        a special comparison operation performed. If compare_func is None,
+        'comparison_key' is ignored.
+    'comparison_key': If provided, will serve as the key for the comparison value.
+        If None, then the name of the comparison operator will used instead.
+    """
+    ops = {
+        "div": operator.truediv,
+        "sub": operator.sub,
+        "add": operator.add,
+        "mul": operator.mul,
+        "ge": operator.ge,
+        "le": operator.le,
+        "lt": operator.lt,
+        "gt": operator.gt,
+        "eq": operator.eq,
+        "ne": operator.ne,
+    }
+    env_acc = {}
+    # If we are at the last branch...
+    subtree_a = retrieve_leaves(tree_a, levels_a, leaf_a)
+    subtree_b = retrieve_leaves(tree_b, levels_b, leaf_b)
+
+    branch_a = trim_branches(subtree_a, levels_a)
+    branch_b = trim_branches(subtree_b, levels_b)
+
+    for trunk in branch_a.keys():
+        value_a = branch_a[trunk]
+        value_b = branch_b[trunk]
+        env_acc.setdefault(trunk, {})
+        env_acc[trunk].setdefault(leaf_a, value_a)
+        env_acc[trunk].setdefault(leaf_b, value_b)
+        comparison_operator = ops.get(compare_func, compare_func)
+        if comparison_operator is not None:
+            compared_value = comparison_operator(value_a, value_b)
+            if comparison_key is None:
+                comparison_key = comparison_operator.__name__
+            env_acc[trunk].setdefault(comparison_key, compared_value)
+    return env_acc
+
+
+def trim_branches(
+    tree: dict | list,
+    levels: list[Hashable | None],
+):
+    """
+    Returns a copy of the 'tree' but with the branches in 
+    'levels' trimmed off.
+    """
+    trimmed = tree.copy()
+    for i in range(len(levels)):
+        leaf = levels.pop()
+        trimmed = retrieve_leaves(trimmed, levels, leaf=leaf)
+    return trimmed
+
+
+def retrieve_leaves(
+    tree: dict | list, 
+    levels: list[Hashable | None], 
+    leaf: list[Hashable] | Hashable | None,
+) -> dict:
+    """
+    Envelopes the tree at the leaf node with 'agg_func'.
+    """
+    env_acc = {}
+    key_error_msg = (
+        "Key '{level}' does not exist at this level. Available keys: {keys}. "
+        "Perhaps not all of your tree elements have the same keys. Try enveloping over trees "
+        "that have the same branch structure and leaf names."
+    )
+    # If we are at the last branch...
+    if not levels:
+        if leaf is None:
+            return tree
+        if isinstance(leaf, list):
+            leaf_values = {}
+            for leaf_elem in leaf:
+                try:
+                    tree[leaf_elem]
+                except KeyError:
+                    raise KeyError(key_error_msg.format(level=leaf_elem, keys=list(tree.keys())))
+                leaf_values.update({leaf_elem: tree[leaf_elem]})
+        else:
+            try:
+                tree[leaf]
+            except KeyError:
+                raise KeyError(key_error_msg.format(level=leaf, keys=list(tree.keys())))
+            leaf_values = tree[leaf]
+        return leaf_values
+    else:
+        # Otherwise, pop the next level and dive into the tree on that branch
+        level = levels[0]
+        if level is not None:
+            try:
+                tree[level]
+            except KeyError:
+                raise KeyError(key_error_msg.format(level=level, keys=list(tree.keys())))
+            env_acc.update({level: retrieve_leaves(tree[level], levels[1:], leaf)})
+            return env_acc
+        else:
+            # If None, then walk all branches of this node of the tree
+            if isinstance(tree, list):
+                tree = {idx: leaf for idx, leaf in enumerate(tree)}
+            for k, v in tree.items():
+                env_acc.update({k: retrieve_leaves(v, levels[1:], leaf)})
+            return env_acc
+        
+
+def extract_keys(
+        object: dict[str, Any], 
+        key_name: str,
+        include_startswith: Optional[str] = None,
+        exclude_startswith: Optional[str] = None,
+    ) -> list[dict[str, Any]]:
+    """
+    Returns a list of dicts where each dict has a key of 'key_name'
+    and a value of one of the keys of 'object'.
+
+    e.g.
+    object = {"key1": value, "key2": value, "key3": value}
+    key_name = "label"
+
+    extract_keys(object, key_name) # [{"label": "key1"}, {"label": "key2"}, {"label": "key3"}]
+
+    'include_startswith': If provided, will only include keys that start with this string.
+    'exclude_startswith': If provided, will exclude all keys that start with this string.
+
+    If both 'include_startswith' and 'exclude_startswith' are provided, exclude is executed
+    first.
+    """
+    shortlist = []
+    for key in object.keys():
+        if exclude_startswith is not None and key.startswith(exclude_startswith):
+            continue
+        else:
+            shortlist.append(key)
+
+    acc = []
+    for key in shortlist:
+        if include_startswith is not None and key.startswith(include_startswith):
+            acc.append({key_name: key})
+        elif include_startswith is None:
+            acc.append({key_name: key})
+
+    return acc
+
+    
+
+def merge_trees(trees: list[dict[str, dict]]) -> dict[str, dict]:
+    """
+    Merges all of the tress (dictionaries) in 'result_trees'. 
+
+    This is different than a typical dictionary merge (e.g. a | b)
+    which will merge dictionaries with different keys but will over-
+    write values if two keys are the same.
+
+    Instead, it crawls each branch of the tree and merges the data
+    within each branch, no matter how deep the branches go.
+    """
+    acc = {}
+    for result_tree in trees:
+        acc = deepmerge.always_merger.merge(acc, result_tree)
+    return acc
\ No newline at end of file