diff --git a/discovery_client.go b/discovery_client.go
new file mode 100644
index 0000000..3a25253
--- /dev/null
+++ b/discovery_client.go
@@ -0,0 +1,134 @@
+package rendezvous
+
+import (
+	"context"
+	"github.com/libp2p/go-libp2p-core/discovery"
+	"github.com/libp2p/go-libp2p-core/host"
+	"github.com/libp2p/go-libp2p-core/peer"
+	"math"
+	"math/rand"
+	"sync"
+	"time"
+)
+
+type rendezvousDiscoveryClient struct {
+	rp        RendezvousPoint
+	peerCache sync.Map //is a map[string]discoveredPeerCache
+	rng       *rand.Rand
+	rngMux    sync.Mutex
+}
+
+type discoveredPeerCache struct {
+	cachedRegs []Registration
+	mux        sync.Mutex
+}
+
+func NewRendezvousDiscoveryClient(host host.Host, rendezvousPeer peer.ID) discovery.Discovery {
+	rp := NewRendezvousPoint(host, rendezvousPeer)
+	return &rendezvousDiscoveryClient{rp, sync.Map{}, rand.New(rand.NewSource(rand.Int63())), sync.Mutex{}}
+}
+
+func (c *rendezvousDiscoveryClient) Advertise(ctx context.Context, ns string, opts ...discovery.Option) (time.Duration, error) {
+	// Get options
+	var options discovery.Options
+	err := options.Apply(opts...)
+	if err != nil {
+		return 0, err
+	}
+
+	ttl := options.Ttl
+	var ttlSeconds int
+
+	// Default is minimum duration
+	if ttl == 0 {
+		ttlSeconds = 120
+	} else {
+		ttlSeconds = int(math.Round(ttl.Seconds()))
+	}
+
+	if err := c.rp.Register(ctx, ns, ttlSeconds); err != nil {
+		return 0, err
+	}
+
+	actualTTL := time.Duration(ttlSeconds) * time.Second
+	return actualTTL, nil
+}
+
+func (c *rendezvousDiscoveryClient) FindPeers(ctx context.Context, ns string, opts ...discovery.Option) (<-chan peer.AddrInfo, error) {
+	// Get options
+	var options discovery.Options
+	err := options.Apply(opts...)
+	if err != nil {
+		return nil, err
+	}
+
+	const maxLimit = 1000
+	limit := options.Limit
+	if limit == 0 || limit > maxLimit {
+		limit = maxLimit
+	}
+
+	// Get cached peers
+	var cache *discoveredPeerCache
+
+	genericCache, _ := c.peerCache.LoadOrStore(ns, &discoveredPeerCache{})
+	cache = genericCache.(*discoveredPeerCache)
+
+	cache.mux.Lock()
+	cachedRegs := cache.cachedRegs
+
+	// Remove all expired entries from cache
+	currentTime := int(time.Now().Unix())
+	newCacheSize := len(cachedRegs)
+
+	for i := 0; i < newCacheSize; i++ {
+		reg := cachedRegs[i]
+		if reg.Ttl < currentTime {
+			newCacheSize--
+			if i != newCacheSize {
+				cachedRegs[i] = cachedRegs[newCacheSize]
+				i--
+			}
+		}
+	}
+	cache.cachedRegs = cachedRegs[:newCacheSize]
+	cache.mux.Unlock()
+
+	// Discover new records if we don't have enough
+	var discoveryErr error
+	if newCacheSize < limit {
+		if discoveryRecords, _, err := c.rp.Discover(ctx, ns, limit, nil); err == nil {
+			for i := range discoveryRecords {
+				discoveryRecords[i].Ttl += currentTime
+			}
+			cache.mux.Lock()
+			cache.cachedRegs = discoveryRecords
+			cache.mux.Unlock()
+		} else {
+			// TODO: Should we return error even if we have valid cached results?
+			discoveryErr = err
+		}
+	}
+
+	// Randomize and fill channel with available records
+	cache.mux.Lock()
+	cachedRegs = cache.cachedRegs
+	sendQuantity := len(cachedRegs)
+	if limit < sendQuantity {
+		sendQuantity = limit
+	}
+
+	chPeer := make(chan peer.AddrInfo, sendQuantity)
+
+	c.rngMux.Lock()
+	perm := c.rng.Perm(len(cachedRegs))[0:sendQuantity]
+	c.rngMux.Unlock()
+
+	for _, i := range perm {
+		chPeer <- cachedRegs[i].Peer
+	}
+
+	cache.mux.Unlock()
+	close(chPeer)
+	return chPeer, discoveryErr
+}
diff --git a/discovery_client_test.go b/discovery_client_test.go
new file mode 100644
index 0000000..854737e
--- /dev/null
+++ b/discovery_client_test.go
@@ -0,0 +1,163 @@
+package rendezvous
+
+import (
+	"context"
+	"github.com/libp2p/go-libp2p-core/discovery"
+	"github.com/libp2p/go-libp2p-core/host"
+	"github.com/libp2p/go-libp2p-core/peer"
+	"math/rand"
+	"sync"
+	"testing"
+	"time"
+)
+
+func getRendezvousDiscovery(hosts []host.Host) []discovery.Discovery {
+	clients := make([]discovery.Discovery, len(hosts)-1)
+	rendezvousPeer := hosts[0].ID()
+	for i, h := range hosts[1:] {
+		rp := NewRendezvousPoint(h, rendezvousPeer)
+		rng := rand.New(rand.NewSource(int64(i)))
+		clients[i] = &rendezvousDiscoveryClient{rp: rp, peerCache: sync.Map{}, rng: rng}
+	}
+	return clients
+}
+
+func peerChannelToArray(pch <-chan peer.AddrInfo) []peer.AddrInfo {
+	pi := make([]peer.AddrInfo, len(pch))
+	peerIndex := 0
+	for p := range pch {
+		pi[peerIndex] = p
+		peerIndex++
+	}
+	return pi
+}
+
+func checkAvailablePeers(t *testing.T, ctx context.Context, client discovery.Discovery, namespace string, expectedNumPeers int) {
+	pch, err := client.FindPeers(ctx, namespace)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	pi := peerChannelToArray(pch)
+
+	if len(pi) != expectedNumPeers {
+		t.Fatalf("Expected %d peers", expectedNumPeers)
+	}
+}
+
+func TestDiscoveryClientAdvertiseAndFindPeers(t *testing.T) {
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	// Define parameters
+	const namespace = "foo1"
+	const numClients = 4
+	const ttl = DefaultTTL * time.Second
+
+	// Instantiate server and clients
+	hosts := getRendezvousHosts(t, ctx, numClients+1)
+
+	svc, err := makeRendezvousService(ctx, hosts[0], ":memory:")
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer svc.DB.Close()
+
+	clients := getRendezvousDiscovery(hosts)
+
+	// Advertise and check one peer
+	_, err = clients[0].Advertise(ctx, namespace, discovery.TTL(ttl))
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	checkAvailablePeers(t, ctx, clients[0], namespace, 1)
+
+	// Advertise and check the rest of the peers incrementally
+	for i, client := range clients[1:] {
+		if _, err = client.Advertise(ctx, namespace, discovery.TTL(ttl)); err != nil {
+			t.Fatal(err)
+		}
+
+		checkAvailablePeers(t, ctx, client, namespace, i+2)
+	}
+
+	// Check that the first peer can get all the new records
+	checkAvailablePeers(t, ctx, clients[0], namespace, numClients)
+}
+
+func TestDiscoveryClientExpiredCachedRecords(t *testing.T) {
+	BaseDiscoveryClientCacheExpirationTest(t, true)
+}
+
+func TestDiscoveryClientExpiredManyCachedRecords(t *testing.T) {
+	BaseDiscoveryClientCacheExpirationTest(t, false)
+}
+
+func BaseDiscoveryClientCacheExpirationTest(t *testing.T, onlyRequestFromCache bool) {
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	// Define parameters
+	const numShortLivedRegs = 5
+	const everyIthRegIsLongTTL = 2
+	const numBaseRegs = numShortLivedRegs * everyIthRegIsLongTTL
+	const namespace = "foo1"
+	const longTTL = DefaultTTL * time.Second
+	const shortTTL = 2 * time.Second
+
+	// Instantiate server and clients
+	hosts := getRendezvousHosts(t, ctx, numBaseRegs+3)
+
+	svc, err := makeRendezvousService(ctx, hosts[0], ":memory:")
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer svc.DB.Close()
+	clients := getRendezvousDiscovery(hosts)
+
+	// Advertise most clients
+	for i, client := range clients[2:] {
+		ttl := shortTTL
+		if i%everyIthRegIsLongTTL == 0 {
+			ttl = longTTL
+		}
+
+		if _, err = client.Advertise(ctx, namespace, discovery.TTL(ttl)); err != nil {
+			t.Fatal(err)
+		}
+	}
+
+	// Find peers from an unrelated client (results should be cached)
+	pch, err := clients[0].FindPeers(ctx, namespace)
+	if err != nil {
+		t.Fatal(err)
+	}
+	pi := peerChannelToArray(pch)
+	if len(pi) != numBaseRegs {
+		t.Fatalf("expected %d registrations", numBaseRegs)
+	}
+
+	// Advertise from a new unrelated peer
+	if _, err := clients[1].Advertise(ctx, namespace, discovery.TTL(longTTL)); err != nil {
+		t.Fatal(err)
+	}
+
+	// Wait for cache expiration
+	time.Sleep(shortTTL + time.Second)
+
+	// Check if number of retrieved records matches caching expectations after expiration
+	expectedNumClients := numShortLivedRegs
+	if !onlyRequestFromCache {
+		expectedNumClients++
+	}
+	pch, err = clients[0].FindPeers(ctx, namespace, discovery.Limit(expectedNumClients))
+	if err != nil {
+		t.Fatal(err)
+	}
+	pi = peerChannelToArray(pch)
+
+	if len(pi) != expectedNumClients {
+		t.Fatalf("received an incorrect number of records: %d", len(pi))
+	}
+}