30-Minute Temporal Talk – Slide Storyline Slide 1 – Hook
Title: We think we see everything.
Show:
FINISHED FILES ARE THE RESULT OF YEARS OF SCIENTIFIC STUDY COMBINED WITH THE EXPERIENCE OF YEARS.
Ask them to count the F’s.
Most say 3. Answer: 6.
Say: If we miss half the letters in one sentence, imagine the failures we miss across retries, crashes, timeouts, redeploys.
Distributed systems fail in ways humans are terrible at noticing.
Slide 2 – Production reality
Title: Systems are messy.
Even in good environments:
requests timeout
responses get lost
retries duplicate work
pods restart
deployments happen mid-flow
humans are late
dependencies are slow
Business expectation?
✅ always correct ✅ never lose progress ✅ easy to debug
Slide 3 – A bug we’ve all written
Title: Looks fine. Isn’t.
Withdraw → Deposit.
Try/catch. Log error.
Withdraw succeeds. Deposit fails.
Money is gone.
Say: Logging an error is not a recovery strategy.
Slide 4 – Why this keeps happening
Because normal services were built for:
👉 request → response
But business processes are:
👉 long 👉 multi-step 👉 cross-service 👉 failure prone 👉 sometimes human driven
We force a square peg into a round hole.
Slide 5 – Time breaks everything
Title: Now wait 2 days for approval.
What do we do?
keep thread alive?
write to DB and poll?
cron jobs?
manual scripts?
And what happens when we deploy tomorrow?
Who remembers we were waiting?
Slide 6 – Async callback illusion
We fire an event.
If downstream is slow or unavailable:
different systems believe different truths
reconciliation jobs appear
DLQs appear
replay tools appear
We accidentally build infrastructure instead of product.
Slide 7 – Even when nobody is “down”
This is where you answer product.
Failures without outages:
caller crashes after success
network drops reply
retry runs twice
deployment mid-transaction
race conditions
timeouts
Up ≠ correct.
Slide 8 – Enter Temporal
Title: What if your code never forgot?
Temporal provides:
✅ durable execution ✅ automatic retries ✅ persistent state ✅ timers without threads ✅ survives crashes & deploys ✅ complete history
You write business logic. Temporal handles survival.
Slide 9 – The mental shift
Title: Describe the process to Temporal.
Your service doesn’t fight failures anymore.
It says:
Here are the steps. Make sure they happen.
Slide 10 – Architecture Client | Service ──► Temporal | Task Queue | Workers (anytime)
Workers can die, scale, redeploy.
Progress is safe.
Slide 11 – The magic trick
Title: Workflows resume. They don’t restart.
If a worker crashes in step 7 of 20:
It continues at step 7.
Because history is recorded.
Like a save game.
Slide 12 – Why checkpoints are huge
Temporal stores:
inputs
outputs
decisions
timers
signals
When it runs again, it deterministically lands exactly where it left off.
No guessing. No rebuilding state.
Slide 13 – Revisit bank transfer
Withdraw → Deposit.
If deposit fails:
retry automatically
or compensate
never lose where we were
No custom saga framework. No manual repair.
Slide 14 – Revisit 2-day approval requestApproval() waitForSignal() continue()
Service can restart 50 times.
Still waiting.
Slide 15 – Revisit downstream dependency
Instead of:
A → call → B and pray
You do:
A → workflow B → worker
If B is down → task waits.
Loose coupling. Safe progress.
Slide 16 – Here’s what people miss
Temporal is not only for failure.
It makes normal development better.
Slide 17 – Engineering gets simpler
You delete:
❌ retry code ❌ idempotency hacks ❌ state tables ❌ polling ❌ cron recovery ❌ DLQ replayers
Focus returns to business logic.
Slide 18 – Visibility becomes insane
Open a workflow and see:
every step
every retry
inputs & outputs
current state
Ask:
Where is order 123 stuck?
Answer in seconds.
Slide 19 – Product people love this
Now you can measure:
approval times
bottlenecks
failure rates
SLA by step
Without building extra tracking.
Slide 20 – New product capabilities
Things that were painful become easy:
✨ human in the loop ✨ long timers ✨ pause/resume ✨ multi-stage onboarding ✨ subscriptions ✨ migrations mid-process
Reliability enables innovation.
Slide 21 – Real world examples
Where companies use it:
payments & refunds
food delivery lifecycle
travel booking
account provisioning
KYC flows
infrastructure creation
returns & exchanges
If it spans time → Temporal.
Slide 22 – Safe deployments
You can push new workers while processes are running.
They pick up safely.
Huge operational win.
Slide 23 – When not to use it
Simple CRUD. Single quick calls.
If there’s no state over time → skip it.
(credibility booster)
Slide 24 – What happens after teams adopt it
People start asking:
“Should this be a workflow?”
And reliability maturity jumps.
Slide 25 – Summary
Temporal = durable execution for business logic.
It remembers so your services don’t have to.
Slide 26 – Closing punch
“If the process matters to the business, it should survive failure.”