diff --git a/requirements-swh.txt b/requirements-swh.txt
--- a/requirements-swh.txt
+++ b/requirements-swh.txt
@@ -3,3 +3,4 @@
 swh.vault >= 0.0.23
 swh.indexer >= 0.0.120
 swh.scheduler >= 0.0.31
+swh.search >= 0.0.3
diff --git a/swh/web/api/views/origin.py b/swh/web/api/views/origin.py
--- a/swh/web/api/views/origin.py
+++ b/swh/web/api/views/origin.py
@@ -194,7 +194,7 @@
             and only the Link header should be used for paginating through
             results.
 
-        :param string url_pattern: a string pattern or a regular expression
+        :param string url_pattern: a string pattern
         :query int limit: the maximum number of found origins to return
             (bounded to 1000)
         :query boolean regexp: if true, consider provided pattern as a regular
@@ -219,21 +219,19 @@
             :swh_web_api:`origin/search/python/?limit=2`
     """
     result = {}
-    offset = int(request.query_params.get('offset', '0'))
     limit = min(int(request.query_params.get('limit', '70')), 1000)
-    regexp = request.query_params.get('regexp', 'false')
+    page_token = request.query_params.get('page_token')
     with_visit = request.query_params.get('with_visit', 'false')
 
-    results = api_lookup(service.search_origin, url_pattern, offset, limit,
-                         bool(strtobool(regexp)), bool(strtobool(with_visit)),
-                         enrich_fn=_enrich_origin)
+    (results, page_token) = api_lookup(
+        service.search_origin, url_pattern, limit,
+        page_token, bool(strtobool(with_visit)),
+        enrich_fn=_enrich_origin)
 
-    nb_results = len(results)
-    if nb_results == limit:
+    if page_token is not None:
         query_params = {}
-        query_params['offset'] = offset + limit
         query_params['limit'] = limit
-        query_params['regexp'] = regexp
+        query_params['page_token'] = page_token
 
         result['headers'] = {
             'link-next': reverse('api-1-origin-search',
@@ -242,7 +240,7 @@
         }
 
     result.update({
-        'results': results
+        'results': list(results)
     })
 
     return result
diff --git a/swh/web/common/service.py b/swh/web/common/service.py
--- a/swh/web/common/service.py
+++ b/swh/web/common/service.py
@@ -3,9 +3,7 @@
 # License: GNU Affero General Public License version 3, or any later version
 # See top-level LICENSE file for more information
 
-import itertools
 import os
-import re
 
 from collections import defaultdict
 from typing import Any, Dict
@@ -23,6 +21,7 @@
 from swh.web.common.origin_visits import get_origin_visit
 from swh.web import config
 
+search = config.search()
 storage = config.storage()
 vault = config.vault()
 idx_storage = config.indexer_storage()
@@ -247,8 +246,7 @@
     return map(converters.from_origin, origins)
 
 
-def search_origin(url_pattern, offset=0, limit=50, regexp=False,
-                  with_visit=False):
+def search_origin(url_pattern, limit=50, page_token=None, with_visit=False):
     """Search for origins whose urls contain a provided string pattern
     or match a provided regular expression.
 
@@ -261,21 +259,11 @@
         list of origin information as dict.
 
     """
-    if not regexp:
-        # If the query is not a regexp, rewrite it as a regexp.
-        regexp = True
-        search_words = [re.escape(word) for word in url_pattern.split()]
-        if len(search_words) >= 7:
-            url_pattern = '.*'.join(search_words)
-        else:
-            pattern_parts = []
-            for permut in itertools.permutations(search_words):
-                pattern_parts.append('.*'.join(permut))
-            url_pattern = '|'.join(pattern_parts)
-
-    origins = storage.origin_search(url_pattern, offset, limit, regexp,
-                                    with_visit)
-    return map(converters.from_origin, origins)
+    results = search.origin_search(url_pattern=url_pattern, count=limit,
+                                   page_token=page_token,
+                                   with_visit=with_visit)
+    origins = map(converters.from_origin, results['results'])
+    return (origins, results['next_page_token'])
 
 
 def search_origin_metadata(fulltext, limit=50):
diff --git a/swh/web/config.py b/swh/web/config.py
--- a/swh/web/config.py
+++ b/swh/web/config.py
@@ -10,6 +10,7 @@
 from swh.core import config
 from swh.indexer.storage import get_indexer_storage
 from swh.scheduler import get_scheduler
+from swh.search import get_search
 from swh.storage import get_storage
 from swh.vault import get_vault
 from swh.web import settings
@@ -30,6 +31,13 @@
             'timeout': 1,
         }
     }),
+    'search': ('dict', {
+        'cls': 'remote',
+        'args': {
+            'url': 'http://127.0.0.1:5010/',
+            'timeout': 10,
+        },
+    }),
     'log_dir': ('string', '/tmp/swh/log'),
     'debug': ('bool', False),
     'serve_assets': ('bool', False),
@@ -127,6 +135,7 @@
         cfg = config.load_named_config(config_file, DEFAULT_CONFIG)
         swhweb_config.update(cfg)
         config.prepare_folders(swhweb_config, 'log_dir')
+        swhweb_config['search'] = get_search(**swhweb_config['search'])
         swhweb_config['storage'] = get_storage(**swhweb_config['storage'])
         swhweb_config['vault'] = get_vault(**swhweb_config['vault'])
         swhweb_config['indexer_storage'] = \
@@ -136,6 +145,13 @@
     return swhweb_config
 
 
+def search():
+    """Return the current application's search.
+
+    """
+    return get_config()['search']
+
+
 def storage():
     """Return the current application's storage.
 
diff --git a/swh/web/settings/tests.py b/swh/web/settings/tests.py
--- a/swh/web/settings/tests.py
+++ b/swh/web/settings/tests.py
@@ -101,7 +101,8 @@
     })
     from swh.web.tests.data import get_tests_data, override_storages # noqa
     test_data = get_tests_data()
-    override_storages(test_data['storage'], test_data['idx_storage'])
+    override_storages(test_data['storage'], test_data['idx_storage'],
+                      test_data['search'])
 else:
     ALLOWED_HOSTS += ['testserver']
 
diff --git a/swh/web/tests/api/views/test_origin.py b/swh/web/tests/api/views/test_origin.py
--- a/swh/web/tests/api/views/test_origin.py
+++ b/swh/web/tests/api/views/test_origin.py
@@ -497,22 +497,6 @@
         == {'https://github.com/memononen/libtess2'}
 
 
-def test_api_origin_search_regexp(api_client):
-    expected_origins = {
-        'https://github.com/memononen/libtess2',
-        'repo_with_submodules'
-    }
-
-    url = reverse('api-1-origin-search',
-                  url_args={'url_pattern': '(repo|libtess)'},
-                  query_params={'limit': 10,
-                                'regexp': True})
-    rv = api_client.get(url)
-    assert rv.status_code == 200, rv.data
-    assert rv['Content-Type'] == 'application/json'
-    assert {origin['url'] for origin in rv.data} == expected_origins
-
-
 @pytest.mark.parametrize('limit', [1, 2, 3, 10])
 def test_api_origin_search_scroll(api_client, archive_data, limit):
     expected_origins = {
@@ -529,8 +513,8 @@
     assert {origin['url'] for origin in results} == expected_origins
 
 
-def test_api_origin_search_limit(api_client, archive_data):
-    archive_data.origin_add([
+def test_api_origin_search_limit(api_client, archive_data, tests_data):
+    tests_data['search'].origin_update([
         {'url': 'http://foobar/{}'.format(i)}
         for i in range(2000)
     ])
diff --git a/swh/web/tests/conftest.py b/swh/web/tests/conftest.py
--- a/swh/web/tests/conftest.py
+++ b/swh/web/tests/conftest.py
@@ -95,7 +95,8 @@
     data = get_tests_data(reset=True)
     # Update swh-web configuration to use the in-memory storages
     # instantiated in the tests.data module
-    override_storages(data['storage'], data['idx_storage'])
+    override_storages(data['storage'], data['idx_storage'],
+                      data['search'])
     return data
 
 
diff --git a/swh/web/tests/data.py b/swh/web/tests/data.py
--- a/swh/web/tests/data.py
+++ b/swh/web/tests/data.py
@@ -20,6 +20,7 @@
 from swh.model.hashutil import hash_to_hex, hash_to_bytes, DEFAULT_ALGORITHMS
 from swh.model.identifiers import directory_identifier
 from swh.loader.git.from_disk import GitLoaderFromArchive
+from swh.search import get_search
 from swh.storage.algos.dir_iterators import dir_iterator
 from swh.web import config
 from swh.web.browse.utils import (
@@ -250,6 +251,11 @@
     # Create indexer storage instance that will be shared by indexers
     idx_storage = get_indexer_storage('memory', {})
 
+    # Create search instance
+    search = get_search('memory', {})
+    search.initialize()
+    search.origin_update({'url': origin['url']}for origin in _TEST_ORIGINS)
+
     # Add the empty directory to the test archive
     empty_dir_id = directory_identifier({'entries': []})
     empty_dir_id_bin = hash_to_bytes(empty_dir_id)
@@ -259,6 +265,7 @@
     return {
         'storage': storage,
         'idx_storage': idx_storage,
+        'search': search,
         'origins': _TEST_ORIGINS,
         'contents': contents,
         'directories': list(directories),
@@ -317,17 +324,21 @@
     return _current_tests_data
 
 
-def override_storages(storage, idx_storage):
+def override_storages(storage, idx_storage, search):
     """
     Helper function to replace the storages from which archive data
     are fetched.
     """
     swh_config = config.get_config()
-    swh_config.update({'storage': storage})
-    service.storage = storage
+    swh_config.update({
+        'storage': storage,
+        'indexer_storage': idx_storage,
+        'search': search,
+    })
 
-    swh_config.update({'indexer_storage': idx_storage})
+    service.storage = storage
     service.idx_storage = idx_storage
+    service.search = search
 
 
 # Implement some special endpoints used to provide input tests data