diff --git a/batch4.go b/batch4.go
index 709f2bf..d2c2795 100644
--- a/batch4.go
+++ b/batch4.go
@@ -8,70 +8,16 @@ import (
 	"sync/atomic"
 )
 
-/*
-Full guide.
-
-	bc.Queue().In() // get in the queue
-
-	// Get data ready for the commit.
-	// All the blocking operations should be done here.
-	data := 1
-
-	if leave() {
-		bc.Queue().Out() // get out of the queue
-		bc.Notify()      // notify waiting goroutines we won't come
-		return
-	}
-
-	idx := bc.Enter(true) // true to block and wait, false for non-blocking return if batch is in the process of commiting
-	                     // just like with the Mutex there are no guaranties how will enter the first
-	if idx < 0 {        // in non-blocking mode we didn't enter the batch, it's always >= 0 in blocking mode
-		return         // no need to do anything here
-	}
-
-	defer bc.Exit() // if we entered we must exit
-	_ = 0          // calling it with defer ensures state consistency in case of panic
-
-	// We are inside locked Mutex between Enter and Exit.
-	// So the shared state can be modified safely.
-	// That also means that all the long and heavy operations
-	// should be done before Enter.
-
-	if idx == 0 { // we are first in the batch, reset the state
-		sum = 0
-	}
-
-	if full() {       // if we didn't change the state we can just leave.
-		bc.Trigger() // Optionally we can trigger the batch.
-		            // Common use case is to flush the data if we won't fit.
-		return     // Then we may return and retry in the next batch.
-	}
-
-	sum += data // adding our work to the batch
-
-	if spoilt() {
-		_, err = bc.Cancel(ctx, causeErr) // cancel the batch, commit isn't done, all get causeErr
-		                                 // if causeErr is nil it's set to Canceled
-	}
-
-	if failed() {
-		panic("we can safely panic here") // panic will be propogated to the caller,
-		                                 // other goroutines in the batch will get PanicError
-	}
-
-	if urgent() {
-		bc.Trigger() // do not wait for the others
-	}
-
-	res, err := bc.Commit(ctx) // get ready to commit.
-	                          // The last goroutine entered the batch calls the actual commit.
-	                         // All the others wait and get the same res and error.
-*/
-
 type (
+	// Queue is a queue of waiting workers.
 	Queue int32
 
+	// Coordinator coordinates workers to update shared state,
+	// commit it and deliver result to all participated workers.
 	Coordinator[Res any] struct {
+		// CommitFunc is called to commit shared shate.
+		//
+		// Required.
 		CommitFunc func(ctx context.Context) (Res, error)
 
 		lock
@@ -95,31 +41,50 @@ type (
 		trigger bool
 	}
 
+	// PanicError is returned to all the workers in the batch if one panicked.
+	// The panicked worker recieves panic, not an error.
 	PanicError struct {
 		Panic interface{}
 	}
 )
 
+// Canceled is a default error returned to workers if Cancel was called with nil err.
 var Canceled = errors.New("batch canceled")
 
+// New creates new Coordinator.
 func New[Res any](f func(ctx context.Context) (Res, error)) *Coordinator[Res] {
 	return &Coordinator[Res]{
 		CommitFunc: f,
 	}
 }
 
+// Init initiates zero Coordinator.
+// It can also be used as Reset but not in parallel with its usage.
 func (c *Coordinator[Res]) Init(f func(ctx context.Context) (Res, error)) {
 	c.CommitFunc = f
 }
 
+// Gets the queue where workers get it to be waited for.
+//
+// Worker can leave the Queue, but it must call Notify to wake up waiting workers.
 func (c *Coordinator[Res]) Queue() *Queue {
 	return &c.lock.queue
 }
 
+// Notify wakes up waiting workers.
+// Must be called if the worker left the queue.
 func (c *Coordinator[Res]) Notify() {
 	c.cond.Broadcast()
 }
 
+// Enter enters the batch.
+// When the call returns we are in the critical section.
+// Shared resources can be used safely.
+// It's similar to Mutex.Lock.
+// Pair method Exit must be called if Enter was successful (returned value >= 0).
+// It returns index of entered worker.
+// 0 means we are the first and we should reset the shared state.
+// If blocking == false and batch is not available negative value returned.
 func (c *Coordinator[Res]) Enter(blocking bool) int {
 	c.mu.Lock()
 
@@ -147,6 +112,10 @@ func (c *Coordinator[Res]) Enter(blocking bool) int {
 	return c.cnt - 1
 }
 
+// Exit exits the critical section.
+// It should be called with defer just after we successfuly Enter the batch.
+// It's similar to Mutex.Unlock.
+// Retururns number of workers still in the batch (blocked in Commit/Cancel call).
 func (c *Coordinator[Res]) Exit() int {
 	defer func() {
 		c.mu.Unlock()
@@ -178,14 +147,23 @@ func (c *Coordinator[Res]) Exit() int {
 	return idx
 }
 
+// Trigger triggers the batch to Commit.
+// We can call Commit or return with the error.
+// If we added our data to the batch or if we didn't respectively.
 func (c *Coordinator[Res]) Trigger() {
 	c.trigger = true
 }
 
+// Commit waits for the waiting workes to add its data to the batch,
+// calls Coordinator.Commit only once for the batch,
+// and returns the same shared result to all workers.
 func (c *Coordinator[Res]) Commit(ctx context.Context) (Res, error) {
 	return commit[Res](ctx, &c.lock, &c.coach, nil, c.CommitFunc)
 }
 
+// Cancel aborts current batch and returns the same error to all workers already added their data to the batch.
+// Coordinator.Commit is not called.
+// Waiting workers but not Entered the critical section are not affected.
 func (c *Coordinator[Res]) Cancel(ctx context.Context, err error) (Res, error) {
 	if err == nil {
 		err = Canceled
@@ -235,18 +213,22 @@ again:
 	return cc.res, cc.err
 }
 
+// In gets into the queue.
 func (q *Queue) In() int {
 	return int(atomic.AddInt32((*int32)(q), 1))
 }
 
+// In gets out of the queue.
 func (q *Queue) Out() int {
 	return int(atomic.AddInt32((*int32)(q), -1))
 }
 
+// Len is the number of workers in the queue.
 func (q *Queue) Len() int {
 	return int(atomic.LoadInt32((*int32)(q)))
 }
 
+// AsPanicError unwrapes PanicError.
 func AsPanicError(err error) (PanicError, bool) {
 	var pe PanicError
 
diff --git a/coordinator_example_test.go b/coordinator_example_test.go
index 82d8681..ff682ed 100644
--- a/coordinator_example_test.go
+++ b/coordinator_example_test.go
@@ -12,13 +12,15 @@ import (
 
 var jobs = flag.Int("jobs", 5, "parallel jobs")
 
-type Service struct {
-	sum int // state we collect to commit together
+type (
+	Service struct {
+		sum int // state we collect to commit together
 
-	bc *batch.Coordinator[int] // [int] is the result value type, set to struct{} if don't need it
-}
+		bc *batch.Coordinator[int] // [int] is the result value type, set to struct{} if don't need it
+	}
 
-type contextKey struct{}
+	contextKey struct{}
+)
 
 func NewService() *Service {
 	s := &Service{}
@@ -94,6 +96,5 @@ func ExampleCoordinator() {
 	}
 
 	wg.Wait()
-
 	// Output:
 }
diff --git a/doc.go b/doc.go
new file mode 100644
index 0000000..7ae0deb
--- /dev/null
+++ b/doc.go
@@ -0,0 +1,100 @@
+/*
+batch made easy.
+
+batch is a safe and efficient way to combine results of multiple parallel tasks
+into one heavy "commit" (save) operation.
+batch ensures each task is either receive successful result if its results
+have been committed (possibly as a parth of a batch) or an error in other case.
+
+Logic for Coordinator
+
+	                    -------------------------
+	-> worker0 ------> | common cumulative state |            /--> worker0
+	-> worker1 ------> | collected from multiple | -> result  ---> worker1
+	-> worker2 ------> | workers and committed   |            \--> worker2
+	-> worker3 ------> | all at once             |             \-> worker3
+	                    -------------------------
+
+# Logic for Multi
+
+Workers are distributed among multiple coaches and then each coach works the same as in Coordinator case.
+
+	                    -------------------------
+	              | -> | state is combined in a  |            /--> worker0
+	-> worker0 -> | -> | few independent coaches | -> result  ---> worker1
+	-> worker1 -> |     -------------------------
+	-> worker2 -> |
+	-> worker3 -> |     -------------------------
+	              | -> | each is committed as an |            /--> worker2
+	              | -> | independent part        | -> result  ---> worker3
+	                    -------------------------
+
+Full Guide
+
+	bc.Queue().In() // get in the queue
+
+	// Get data ready for the commit.
+	// All the blocking operations should be done here.
+	data := 1
+
+	if leave() {
+		bc.Queue().Out() // get out of the queue
+		bc.Notify()      // notify waiting goroutines we won't come
+		return
+	}
+
+	idx := bc.Enter(true) // true to block and wait, false for non-blocking return if batch is in the process of commiting
+	                     // just like with the Mutex there are no guaranties how will enter the first
+	if idx < 0 {        // in non-blocking mode we didn't enter the batch, it's always >= 0 in blocking mode
+		return         // no need to do anything here
+	}
+
+	defer bc.Exit() // if we entered we must exit
+	_ = 0          // calling it with defer ensures state consistency in case of panic
+
+	// We are inside locked Mutex between Enter and Exit.
+	// So the shared state can be modified safely.
+	// That also means that all the long and heavy operations
+	// should be done before Enter.
+
+	if idx == 0 { // we are first in the batch, reset the state
+		sum = 0
+	}
+
+	if full() {       // if we didn't change the state we can just leave.
+		bc.Trigger() // Optionally we can trigger the batch.
+		            // Common use case is to flush the data if we won't fit.
+		return     // Then we may return and retry in the next batch.
+	}
+
+	sum += data // adding our work to the batch
+
+	if spoilt() {
+		_, err = bc.Cancel(ctx, causeErr) // cancel the batch, commit isn't done, all get causeErr
+		                                 // if causeErr is nil it's set to Canceled
+	}
+
+	if failed() {
+		panic("we can safely panic here") // panic will be propogated to the caller,
+		                                 // other goroutines in the batch will get PanicError
+	}
+
+	if urgent() {
+		bc.Trigger() // do not wait for the others
+	}
+
+	res, err := bc.Commit(ctx) // get ready to commit.
+	                          // The last goroutine entered the batch calls the actual commit.
+	                         // All the others wait and get the same res and error.
+
+Batch is a safe wrapper around Coordinator.
+It will call missed methods if that makes sense or it will panic otherwise.
+
+Multi is a coordinator with N available parallel batches.
+Suppose you have 3 db replicas so you can distribute load across them.
+Multi.Enter will enter the first free coach (replica in our example) and return its index.
+The rest is similar. Custom logic for choosing coach can be used by setting Multi.Balancer.
+
+MultiBatch is a safe wrapper around Multi just like Batch wraps Coordinator.
+*/
+package batch
diff --git a/multi.go b/multi.go
index 2a99ea6..dd3386a 100644
--- a/multi.go
+++ b/multi.go
@@ -5,9 +5,16 @@ import (
 )
 
 type (
+	// Multi is a coordinator to choose and access multiple coaches.
 	Multi[Res any] struct {
+		// CommitFunc is called to commit result of coach number coach.
 		CommitFunc func(ctx context.Context, coach int) (Res, error)
 
+		// Balancer chooses replica among available or it can choose to wait for more.
+		// bitset is a set of available coaches.
+		// Coach n is available <=> bitset[n/64] & (1<<(n%64)) != 0.
+		// If n is >= 0 and the coach is available it proceeds with it.
+		// If it's not Enter blocks or returns -1, -1 according to Enter argument.
 		Balancer  func(bitset []uint64) int
 		available []uint64
 
@@ -17,6 +24,7 @@ type (
 	}
 )
 
+// NewMulti create new Multi coordinator with n parallel coaches.
 func NewMulti[Res any](n int, f func(ctx context.Context, coach int) (Res, error)) *Multi[Res] {
 	return &Multi[Res]{
 		CommitFunc: f,
@@ -25,19 +33,31 @@ func NewMulti[Res any](n int, f func(ctx context.Context, coach int) (Res, error
 	}
 }
 
+// Init initiates zero Multi.
+// It can also be used as Reset but not in parallel with its usage.
 func (c *Multi[Res]) Init(n int, f func(ctx context.Context, coach int) (Res, error)) {
 	c.CommitFunc = f
 	c.cs = make([]coach[Res], n)
 }
 
+// Queue returns common queue for entering coaches.
+// Getting into it means already started batches will wait for it not committing yet.
+//
+// Worker can leave the Queue, but it must call Notify to wake up waiting workers.
 func (c *Multi[Res]) Queue() *Queue {
 	return &c.lock.queue
 }
 
+// Notify wakes up waiting workers. Can be used if you were waited,
+// but now situation has changed.
 func (c *Multi[Res]) Notify() {
 	c.cond.Broadcast()
 }
 
+// Enter enters available batch.
+// It will return -1, -1 if no coaches available and blocking == false.
+//
+// coach choice can be configured by setting custom Multi.Balancer.
 func (c *Multi[Res]) Enter(blocking bool) (coach, idx int) {
 	c.mu.Lock()
 
@@ -104,6 +124,8 @@ func (c *Multi[Res]) enterBalancer() (coach, idx int) {
 	return coach, c.cs[coach].cnt - 1
 }
 
+// Exit exits the batch. Should be called with defer.
+// It works similar to Mutex.Unlock in the sense it unlocks shared resources.
 func (c *Multi[Res]) Exit(coach int) int {
 	defer func() {
 		c.mu.Unlock()
@@ -137,16 +159,29 @@ func (c *Multi[Res]) Exit(coach int) int {
 	return idx
 }
 
+// Trigger triggers the batch to commit.
+// Must be called inside Enter-Exit section.
+//
+// Can be used to flush current batch and retry.
+// Commit only happens when current worker Exits from the section.
+// So you need to Exit and then get into the Queue and Enter again to retry.
 func (c *Multi[Res]) Trigger(coach int) {
 	c.cs[coach].trigger = true
 }
 
+// Commit indicates the current worker is done with the shared data and ready to Commit it.
+// Commit blocks until batch is committed. The same Res is returned to all the workers in the batch.
+// (Res, error) is what the Multi.Commit returns.
 func (c *Multi[Res]) Commit(ctx context.Context, coach int) (Res, error) {
 	return commit(ctx, &c.lock, &c.cs[coach], nil, func(ctx context.Context) (Res, error) {
 		return c.CommitFunc(ctx, coach)
 	})
 }
 
+// Cancel indicates the current worked is done with shared data but it can't be committed.
+// All the workers from the same batch receive zero Res and the same error.
+//
+// It can be used if batch shared state have been spoilt as a result of error or something.
 func (c *Multi[Res]) Cancel(ctx context.Context, coach int, err error) (Res, error) {
 	if err == nil {
 		err = Canceled
diff --git a/safe_example_test.go b/safe_example_test.go
new file mode 100644
index 0000000..f3f711d
--- /dev/null
+++ b/safe_example_test.go
@@ -0,0 +1,94 @@
+package batch_test
+
+import (
+	"context"
+	"errors"
+	"log"
+	"sync"
+
+	"nikand.dev/go/batch"
+)
+
+type SafeService struct {
+	sum int // state we collect to commit together
+
+	bc *batch.Coordinator[int] // [int] is the result value type, set to struct{} if don't need it
+}
+
+func NewSafeService() *SafeService {
+	s := &SafeService{}
+
+	s.bc = batch.New(s.commit)
+
+	return s
+}
+
+func (s *SafeService) commit(ctx context.Context) (int, error) {
+	// suppose some heavy operation here
+	// update a file or write to db
+
+	log.Printf("* * *  commit %2d  * * *", s.sum)
+
+	return s.sum, nil
+}
+
+func (s *SafeService) DoWork(ctx context.Context, data int) (int, error) {
+	b := batch.By(s.bc)
+	defer b.Exit() // it's like Mutex.Unlock, but safely works even if we didn't enter
+	_ = 0          // Must be called with defer to outlive panics
+
+	b.QueueIn() // let others know we are going to join
+
+	_ = data // prepare data
+
+	idx := b.Enter(true) // true for blocking, false if we want to leave instead of waiting
+	if idx < 0 {         // we haven't entered the batch in non blocking mode
+		return 0, errors.New("not this time") // we have to leave in that case
+	}
+
+	if idx == 0 { // we are first in the batch, reset the state
+		s.sum = 0
+		log.Printf("* * * reset batch * * *")
+	}
+
+	log.Printf("worker %2d got in with index %2d", ctx.Value(contextKey{}), idx)
+
+	s.sum += data // add our work to the batch
+
+	// only one of return/Cancel/Commit must be called and only once
+	res, err := b.Commit(ctx)
+	if err != nil { // batch failed, each worker in it will get the same error
+		return 0, err
+	}
+
+	log.Printf("worker %2d got result %v %v", ctx.Value(contextKey{}), res, err)
+
+	// if we are here, all of the workers have their work committed
+
+	return res, nil
+}
+
+func ExampleBatch() {
+	s := NewSafeService()
+
+	// let's spin up some workers
+	var wg sync.WaitGroup
+
+	for j := 0; j < *jobs; j++ {
+		j := j
+		wg.Add(1)
+
+		go func() {
+			defer wg.Done()
+
+			ctx := context.Background() // passed to commit function
+			ctx = context.WithValue(ctx, contextKey{}, j)
+
+			res, err := s.DoWork(ctx, 1)
+			_, _ = res, err
+		}()
+	}
+
+	wg.Wait()
+	// Output:
+}
diff --git a/semaphore.go b/semaphore.go
index 31f2395..fc8f08a 100644
--- a/semaphore.go
+++ b/semaphore.go
@@ -3,6 +3,9 @@ package batch
 import "sync"
 
 type (
+	// Semaphore is a classical semaphore synchronization primitive.
+	// All methods can be safely called on nil Semaphore.
+	// They will do nothing like you have unlimited semaphore.
 	Semaphore struct {
 		mu   sync.Mutex
 		cond sync.Cond
@@ -11,6 +14,7 @@ type (
 	}
 )
 
+// NewSemaphore creates a new semaphore with capacity of n.
 func NewSemaphore(n int) *Semaphore {
 	b := &Semaphore{}
 
@@ -19,6 +23,10 @@ func NewSemaphore(n int) *Semaphore {
 	return b
 }
 
+// Reset resets semaphore capacity.
+// But not the current value, which means it can be used
+// to update limit on the fly, but it can't be used to reset
+// inconsistent semaphore.
 func (b *Semaphore) Reset(n int) {
 	if b == nil {
 		return
@@ -34,6 +42,7 @@ func (b *Semaphore) Reset(n int) {
 	b.lim = n
 }
 
+// Enter critical section.
 func (b *Semaphore) Enter() int {
 	if b == nil {
 		return 0
@@ -51,6 +60,7 @@ func (b *Semaphore) Enter() int {
 	return b.n
 }
 
+// Exit from critical section.
 func (b *Semaphore) Exit() {
 	if b == nil {
 		return
@@ -63,6 +73,7 @@ func (b *Semaphore) Exit() {
 	b.cond.Signal()
 }
 
+// Len is a number of tasks in the critical section.
 func (b *Semaphore) Len() int {
 	if b == nil {
 		return 0
@@ -74,6 +85,7 @@ func (b *Semaphore) Len() int {
 	return b.n
 }
 
+// Cap is a semaphore capacity.
 func (b *Semaphore) Cap() int {
 	if b == nil {
 		return 0