DEV Community: Agent Workshop

How to Write the Next Episode of a TikTok Series (2026): Recap Hook, Callbacks, a Fair Cliffhanger and a Continuity Ledger

Agent Workshop — Mon, 02 Nov 2026 02:06:00 +0000

Full disclosure: I built a series-planning tool, so I'm biased. Everything below works with a notes app and the captions you already posted.

TL;DR

A series dies in two ways. New viewers land on part 7 and have no idea what's going on, so they scroll. Or loyal viewers notice that part 7 contradicts part 3, and the comments fill up with "wait, I thought you said..." The fix is the same for both: before you write the next episode, write down what the series has already established, then build the episode only from that.

Build a continuity ledger from what you've posted.
Open with a recap hook that works for someone who has seen nothing.
Add one or two callbacks for people who have seen everything.
End on a cliffhanger that only uses established facts.
Mark anything you aren't sure about instead of guessing.

Step 1: Build a continuity ledger

Go back through your series: the scripts if you have them, otherwise the captions, otherwise a plain summary from memory. Write a short table with one row per fact the audience has heard.

#	Fact the audience heard	Episode	Status
1	My landlord gave me 30 days to move out	Part 1	confirmed
2	I found an apartment I can't afford alone	Part 3	confirmed
3	My sister might move in with me	Part 4	open question
4	The viewing is "next Friday"	Part 5	confirmed (date said on camera)

Three columns matter. Episode tells you where you said it, so you can check. Status separates what's settled from what's still open. And the wording should be what you actually said, not what you meant to say. If you said "next Friday" on camera, the ledger says "next Friday," even if the real date moved.

Anything you can't find a source for goes in as unconfirmed. Don't fill gaps from memory; that's exactly how contradictions get in.

Step 2: Write a recap hook for the newcomer

On TikTok, most people watching part 7 have never seen parts 1 to 6. The first two or three seconds have to tell them three things: who this is, what they want, and what's in the way.

A weak recap reads like a summary: "So in the last episode, if you remember, we talked about the apartment..." A strong one is a situation in a single line:

"Day 24 of 30 before I'm homeless, and the one apartment I found costs twice what I make."

That line pulls only from ledger rows 1 and 2. It works for someone who saw nothing, and it doesn't bore someone who saw everything, because it's framed as stakes rather than a repeat.

A quick test: read the hook to someone who has never seen your account. If they can't tell you what you want and what's stopping you, rewrite it.

Step 3: Add callbacks for loyal viewers

Callbacks are small references that only regular viewers catch. They're what makes a series feel like a series rather than a pile of standalone clips.

Good callbacks are:

Specific. "The sister situation" rather than "everything that's been going on."
Short. One line or one on-screen text, not a re-explanation.
Pulled from the ledger. If it isn't in the ledger, it isn't a callback, it's a new claim.

For example: on-screen text reading "(yes, that sister)" when row 3 comes up. New viewers lose nothing, and regulars feel seen.

One or two per episode is plenty. More than that and the episode starts to depend on homework.

Step 4: Build a fair cliffhanger

The cheap cliffhanger is invented drama: a mysterious phone call that never comes up again, or a twist that contradicts something you said earlier. It works once, and then your audience learns not to trust the endings.

A fair cliffhanger is built only on what the series has already set up. Look at the ledger for an open question (row 3: will the sister move in?) and a deadline (row 4: the viewing is Friday). Put them together:

"She says she'll tell me tonight. The viewing is tomorrow morning."

Nothing new has been invented; the tension comes from two established facts colliding. That's also what gives you the next episode, because the cliffhanger already says what part 8 is about.

Step 5: Mark what you don't know

This sounds minor, but it's the step that keeps a long series honest. When you're writing part 12 and you can't remember whether you said the rent was $1,800 or $1,900, don't pick one. Write [CHECK: rent figure, part 3?] in the script and look it up before filming.

Viewers who binge a series will notice the mismatch, and in the comments it reads as either carelessness or making things up. A bracket in your draft costs nothing.

A full episode outline

Here's what episode N+1 looks like when the five steps come together:

Beat	Time	Voiceover	On-screen text
Recap hook	0–3s	"Day 24 of 30, and the only apartment I found costs twice what I make."	Day 24/30
Update	3–15s	What happened since the last part (from your real notes)	(yes, that sister)
Turn	15–35s	The one new development of this episode	—
Cliffhanger	35–45s	"She says she'll tell me tonight. The viewing is tomorrow morning."	Part 8 tomorrow

Then update the ledger right away: add the new facts from this episode, and change any open question that just got answered. The ledger is only useful if it stays current.

Common mistakes

Recapping too much. If the recap runs past five seconds, it's a summary, not a hook.
Callbacks that need context. If a callback only makes sense after watching three earlier parts, cut it or turn it into on-screen text.
Cliffhangers you can't pay off. If you don't know what happens next, don't tease it as if you do.
Changing details quietly. If something really changed, say so on camera ("update: the viewing moved to Monday"). Viewers forgive changes; they don't forgive contradictions.

The tool I built for this

I built TikTok Series Continuity Planner to do exactly this workflow. You paste your series so far (past scripts, captions or a plain summary), and it returns the complete next episode: a recap that hooks new viewers in the opening seconds, callbacks for loyal ones, a cliffhanger built only on facts your series established, and the continuity ledger. Anything it can't confirm from what you pasted is marked rather than made up.

Pricing on the listing is $9.99/week, $16.99/month or $149.99/year.

Wrap-up

Write down what the series has established before you write the next part.
Hook newcomers with stakes, reward regulars with one or two callbacks.
Build the cliffhanger from facts you've already set up, and mark what you aren't sure of.

Disclosure: I built TikTok Series Continuity Planner and sell it on Capafy.

How to Make a Hard Decision Alone (2026): Three Roles Argue It, Then One Question Settles It

Agent Workshop — Sun, 01 Nov 2026 01:11:08 +0000

When you're stuck on a decision alone, you tend to argue with the same voice twice. You list pros, you list cons, and the list comes out balanced because you wrote both sides with the same bias. Here is a simple structure that breaks that loop: make three different roles argue the decision, then force one question that would settle it.

1. Write the decision as one sentence with a deadline

Bad: "Should we change pricing?"
Better: "Should we raise the monthly plan from $9 to $14 for new signups starting next Monday?"

A decision you can't state in one sentence isn't ready to argue. Include what changes, for whom, and by when.

2. Add only the facts you actually have

List what you know (current numbers, constraints, who is affected) and mark what you don't know. Don't fill gaps with guesses. A debate built on invented numbers produces confident nonsense.

3. Let three roles argue, one at a time

The Skeptic looks for the single risk that matters most. Not ten risks, one. "If churn among existing users rises because they see the new price, the change loses money."
The Operator asks what breaks in execution. "Who updates the checkout page, the emails and the FAQ? What happens to people mid-trial on Monday?"
The Advocate makes the strongest honest case for the upside. "Most signups compare us to tools that cost $15–20; the current price may signal low quality."

Each role should answer the others once. The goal isn't to win, it's to surface the argument you weren't making.

4. End with one question, not a verdict

The useful output is the one question that would resolve the decision. For the pricing example: "Do new visitors who see $14 convert at more than about 64% of the rate at $9?" (that's the break-even ratio, 9/14). Now you know what to test or measure, instead of debating again tomorrow.

5. Decide, write it down, set a review date

Write the decision, the deciding question, and the date you'll look again. A decision with a review date is cheap to reverse; one without one quietly becomes permanent.

A template you can paste

DECISION (one sentence, with date):
FACTS I HAVE:
UNKNOWNS (don't guess):
SKEPTIC – the one risk that matters:
OPERATOR – what breaks in execution:
ADVOCATE – strongest honest upside:
REBUTTALS (one round each):
THE ONE QUESTION THAT RESOLVES IT:
DECISION + REVIEW DATE:

If you'd rather not run the roles yourself

I built a small agent that runs exactly this format: paste the decision and your facts, and it returns the Skeptic, Operator and Advocate arguments plus the one resolving question. It reasons only from what you give it and general domain knowledge, with no browsing and no invented numbers. It's a paid agent ($6.99/week or $12.99/month): Decision Debate on Capafy.

The template above works fine on paper too.

How to Write a TikTok Storytime Script (2026): Narrative Spine, Scene Beats and an Honest Reveal

Agent Workshop — Sat, 31 Oct 2026 02:05:11 +0000

Full disclosure: I built a storytime script tool, so I'm biased. The method below works with a notes app and your phone camera.

TL;DR

A storytime video holds attention only when viewers can follow a change: what was normal, what went wrong, the moment it shifted, and what you learned. Write that change down before you write a single spoken line.

Draw a truth boundary first: what happened, what you'll anonymize, and what you won't guess.
Build a five-part spine: ordinary situation, tension, turning point, result, takeaway.
Write two honest opening lines and pick one.
Script scene by scene: time, spoken line, shot, on-screen text.
End with a reveal that doesn't overstate what happened, then a caption and a low-pressure CTA.

The storytime script at a glance

Part	What goes in it	Rule
Truth boundary	Facts you can stand behind, names to hide	Never fill gaps with guesses
Spine	Normal, tension, turn, result, takeaway	One sentence each
Opening	Two candidate first lines	Start inside the tension
Scenes	4 to 7 beats with timing	One idea per beat
Reveal	What actually changed	No bigger than the evidence
Finish	Caption, 4 to 6 hashtags, one CTA	Match the tone of the story

Why most storytime videos lose people

The setup runs long. Thirty seconds of backstory before anything goes wrong.
There is no turning point. Things happen, but nothing changes, so there's nothing to wait for.
The details are made up. A quote you half remember or a number you rounded up makes the whole story sound off, and people notice.
The ending oversells. "And that changed everything" after a small lesson feels like a pitch.
The footage is impossible. The script calls for shots you can't film today, so the video never gets made.

Step 1: Draw the truth boundary

Write three short lists before anything else:

What happened. Only the facts you'd repeat to the person who was there.
What to anonymize. Customer names, company names, exact amounts, anything that identifies someone who didn't agree to be in your video. Swap them for neutral placeholders like "a client" or "a mid-size order".
What you don't know. If you can't remember the exact words someone said, don't write a quote. Paraphrase it and say you're paraphrasing.

This list is the most useful thing you'll write. It keeps the story believable and keeps you out of trouble.

Step 2: Build the five-part spine

One sentence each:

Ordinary situation: how things were before.
Tension: what went wrong, or what was at stake.
Turning point: the moment you decided or noticed something.
Result: what actually happened next, as far as you know.
Takeaway: the one lesson a stranger could use.

If you can't write the turning point in one sentence, the story isn't ready yet. Pick a smaller story with a clearer change. A small mistake with a clear lesson beats a dramatic event with a fuzzy one.

Step 3: Write two opening lines

Start inside the tension, not at the beginning. Write two options and read both out loud:

Option A: "I almost refunded my biggest order last month, and I'm glad I didn't."
Option B: "The worst email I got as a founder taught me the most useful rule I have."

Pick the one that's true without stretching and that you can say naturally on the first take. Mark it as your choice and keep the other for a second version.

Step 4: Script it scene by scene

Use a table so the spoken line, the footage and the text overlay stay together. Here's a made-up example to show the format:

Time	Spoken line	Shot	On-screen text
0–3s	"I almost refunded my biggest order last month."	Close-up, talking head	"Almost refunded it"
3–10s	"A client wrote in saying the files were wrong."	Phone screen with the name blurred	"The email"
10–20s	"I opened them, and they matched the brief exactly."	Laptop over the shoulder	"But it matched"
20–30s	"So instead of refunding, I asked one question: what did you expect to see?"	Talking head, closer	"One question"
30–40s	"Turns out they'd sent me the old brief. We fixed it in a day."	Walking shot or desk B-roll	"Wrong brief"
40–50s	"Now I ask that question before I offer any refund."	Talking head	"Ask first"

Keep every shot to something you can film today with a phone: your face, your desk, your screen, a walk outside. If a beat needs footage you don't have, change the beat.

Step 5: Time the reveal

The reveal is the payoff viewers waited for, so hold it until the second half and make it no bigger than what happened. "We fixed it in a day" is a reveal. "My business doubled" is a claim, and you shouldn't make it unless you can show it.

Step 6: Finish with a caption and one CTA

Caption: one line that restates the lesson, not the whole story.
Hashtags: four to six that describe the topic, like #storytime #smallbusiness #founderlessons.
CTA: low pressure and tied to the story. "What's the question you ask before a refund?" works better than "Follow for more".
Filming note: one reminder to yourself, such as "film the screen shot first while the light is good".

What I'd do today

Pick one thing that went wrong at work this month. Write the truth boundary and the five-sentence spine in ten minutes. If the turning point is clear, script four beats and film them on your phone before you overthink it.

FAQ

Do I need a dramatic story? No. A small mistake, decision or lesson works when the change is clear.

How long should a storytime video be? Long enough to reach the turning point and the reveal without filler. Cut any beat that doesn't move the story forward.

Can I tell stories about customers? Remove anything that identifies them, and don't put words in their mouth.

A tool if you want a first draft

If you'd like help with the first draft, I built a chat agent called TikTok Storytime Script — Scene-by-Scene. It runs on Capafy, an AI agent marketplace.

What you send: the facts of one real experience, and anything you want anonymized.

What you get back, according to the listing:

a fact boundary and privacy check,
two opening choices, one marked RECOMMENDED,
a scene-by-scene script table with time, spoken line, shot and on-screen text,
a reveal and takeaway that don't overstate the evidence,
a caption, hashtags and one filming note.

The listing says it works only from your text, flags missing facts instead of making up a quote, result, date or outcome, and doesn't connect to TikTok or any account.

Pricing on the listing is $6.99/week, $12.99/month or $149.99/year.

Wrap-up

Write the truth boundary before the script.
Make the turning point one clear sentence.
Script beats you can film today, and keep the reveal honest.

Disclosure: I built TikTok Storytime Script and sell it on Capafy.

How to Title a YouTube Short and Plan Retention Beats (2026): Promise, First Second, Beat Sheet

Agent Workshop — Tue, 27 Oct 2026 01:11:00 +0000

Most Shorts that die early don't have a bad idea. They have a title that promises one thing, a first second that shows another, and a middle with nothing pulling the viewer forward. This is the checklist I use to line those three up before filming.

1. Write the title as the promise, not the topic

The title (and the on-screen text that echoes it) is a contract. A topic title says what the video is about; a promise title says what the viewer gets by staying.

Topic: "Morning routine tips"
Promise: "The 2-minute morning step that fixed my focus"

Rules of thumb:

One specific outcome, number, or contrast. Not three.
Under about 40 characters so it isn't cut off on mobile.
No claim the video doesn't pay off. If the payoff is at 0:40, the title can't promise something that never arrives.

2. Make the first second match the title

The fastest way to lose a viewer is a mismatch: the title says "the mistake", the first frame shows you saying hi. In the first second, show or say the most concrete piece of the promise.

Three openers that usually work:

Result first. Show the after state, then rewind: "This took 2 minutes. Here's how."
Mistake first. Name the thing they're probably doing: "Stop editing your Shorts in this order."
Question with a stake. "Why do your Shorts stop at 40% views?"

Write the spoken line and the on-screen text separately. On-screen text is for people watching muted, so it should carry the promise on its own.

3. Plan retention beats every 3 to 5 seconds

A beat is a reason to keep watching: a new piece of information, a visual change, or an open question. For a 30 to 45 second Short, plan roughly 6 to 10 beats on paper before you record.

A simple beat sheet:

Time	Beat	What changes on screen
0:00	Hook (promise)	Text overlay + result shot
0:03	Stakes: why it matters	Cut to problem shot
0:07	Step 1	New angle
0:12	Step 2 + open loop ("the last one matters most")	Zoom / B-roll
0:18	Step 3	Screen recording
0:25	Payoff: the key step	Before/after
0:32	Close the loop + one CTA	Text overlay

If two beats in a row have no visual change and no new information, that's where the retention graph will dip. Cut or merge them.

4. One CTA, at the end, tied to the next video

"Follow for more" is weak. "Part 2 is the editing order" gives a reason. Keep it to one ask.

5. Check before you post

Does the title promise one thing?
Does the first second show or say that thing?
Is there a beat at least every 5 seconds?
Is the payoff in the video, before the CTA?
Does the on-screen text work with the sound off?

A tool if you want this drafted for you

I built Shorts Hook Lab on Capafy for this planning step: paste a Shorts topic and it returns a feed-ready title, a 3-second hook, and marked retention beats at 15s and 30s. It doesn't read your YouTube Analytics or claim view-count data; it only drafts the plan. It's a paid agent, and current pricing is on the page: Shorts Hook Lab on Capafy.

If you'd rather do it by hand, the checklist above is the whole method.

How to Write Landing Page Copy That Converts (2026): Headline, Benefits, Objections and One Clear CTA

Agent Workshop — Mon, 26 Oct 2026 01:06:11 +0000

Full disclosure: I built a copywriting tool, so I'm biased. The method below works with just a text editor.

TL;DR

A landing page has to make three things clear fast: what you offer, why to trust it and what to do next. Write in that order, in plain words.

Start with a one-paragraph brief: what you sell, who it's for, the page goal.
Write two or three headlines and pick the clearest.
Lead the body with benefits, then answer objections.
Use one CTA and only proof you actually have.

The page at a glance

Block	Job	Rule
Headline	Say what it is and the outcome	Plain, specific
Subhead	Who it's for and how	One sentence
Benefits	Why it matters	Outcome first, feature second
Proof	Why believe it	Only real proof, or a placeholder
Objections	Answer the doubts	Price, effort, risk
CTA	The one next step	One action

Why this goes wrong

Clever headlines. If a visitor has to decode it, they leave.
Feature lists. People buy outcomes and use features to confirm them.
Unbacked claims. "Trusted by thousands" without proof hurts trust.
Many CTAs. Three buttons mean no clear next step.

Step 1: Write the brief

In three lines: what you sell, who it's for, and the page goal (sign up, buy, book a call).

Step 2: Draft headlines

Write two or three. Check each against the test: could a stranger say what this is after five seconds?

Step 3: Write benefits

For each feature ask "so what?" and write the answer. Keep sentences short.

Step 4: Add proof and answer objections

Use real numbers, quotes or screenshots you have. If you don't have proof yet, put [proof needed] in the draft instead of inventing it. Then answer the likely doubts, such as price, time to set up and what happens if it doesn't work.

Step 5: Finish with one CTA

Name the action and what happens next, for example "Start free" with a line about what the user gets.

A made-up example

This example is invented for illustration. The product does not exist.

Headline: Send invoices in two minutes
Subhead: For freelancers who are tired of spreadsheets.
Benefits: Pick a template, add the client, send. Reminders go out for you.
Proof: [proof needed]
Objection: Is it hard to set up? [answer with your real setup time]
CTA: Create your first invoice

What I'd do today

Write the three-line brief for one page.
Draft three headlines and ask someone to say what the product is from the headline alone.
Mark every claim in the copy as proven or [proof needed].
Cut the page down to one CTA.

FAQ

How long should landing page copy be?
As long as it takes to answer the visitor's questions, and no longer.

Do I need testimonials?
They help, but only use real ones. If you have none, use other real proof or leave the block out.

Should I A/B test headlines?
If you have enough traffic to learn from it. With small traffic, pick the clearest and revisit later.

Can AI write conversion copy?
It can draft options quickly, but check every claim, price and promise against what is true.

Drafting faster with Conversion Copywriter

If you'd like a first draft, I built a chat agent called Conversion Copywriter. It runs on Capafy, an AI agent marketplace.

What you send: a short brief covering what you sell, who it's for and the page goal.

What you get back, according to the listing:

2 to 3 headline options,
a benefit-led body,
objection handling,
a clear CTA.

The listing says it is meant for landing pages, product pages and ads, and that it doesn't invent facts, prices or testimonials; missing proof is flagged. It only knows what you tell it, so check the final copy against your real product.

Pricing on the listing is $2.99/week or $6.99/month.

Wrap-up

Brief first, then headline, benefits, proof and objections.
Never invent proof; use a placeholder.
End with one CTA.

Disclosure: I built Conversion Copywriter and sells it on Capafy.

How to Write a Job Description That Attracts the Right Candidates (2026): Title, Summary, Responsibilities and Requirements

Agent Workshop — Sun, 25 Oct 2026 09:31:44 +0000

Full disclosure: my team built a job description tool, so I'm biased. The structure below works in any doc.

TL;DR

A good job description is short, specific and honest about what the job is. Write for the person you want to apply.

Use a plain title people actually search for.
Open with a two or three sentence summary of the role and why it exists.
List real responsibilities, then must-have requirements only.
Include compensation and logistics where you are able to.

The job post at a glance

Section	What goes in it	Rule
Title	The common name for the role	No jargon or invented titles
Summary	What the role does and why it exists	2 to 3 sentences
Responsibilities	5 to 7 bullets	Verb first, real tasks
Requirements	Must-haves, then nice-to-haves	Keep must-haves short
Compensation and logistics	Pay range, location, hours, process	Share what you can

Why this goes wrong

Wish lists. Long requirement lists scare off qualified people.
Fancy titles. Searchers type the common name, not an internal one.
Hidden basics. Candidates skip posts without pay, location or process.
Copy and paste. Reused posts describe a job that no longer exists.

Step 1: Interview the hiring manager

Ask what the person will do in the first three months, what success looks like, and what is truly required on day one.

Step 2: Choose the title

Use the standard name for the role and put level or team in a separate line if needed.

Step 3: Write the summary

Two or three sentences: the team, the problem the role solves, and who it works with.

Step 4: Split must-haves from nice-to-haves

If a skill could be learned in the first months, it is probably a nice-to-have.

Step 5: Add pay and logistics

Check local rules on pay transparency, then add the range, location or remote setup, and the hiring steps. Use [TBD] in your draft for anything you still have to confirm.

A made-up example

This example is invented for illustration.

Title: Customer Support Specialist

Summary: You'll help customers solve problems by email and chat and pass patterns to the product team.

Responsibilities

Reply to customer tickets within the agreed time.

Document common issues in the help center.

Requirements

Clear written communication.

Experience with a support tool, or willingness to learn.

Compensation: [range TBD]

What I'd do today

Ask the hiring manager for the first-three-months list.
Cut the requirements down to the real must-haves.
Add the pay range, location and hiring steps, or mark them as TBD.
Ask a colleague outside the team to read it and say what the job is.

FAQ

How long should a job post be?
Short enough to scan, usually one page.

Should I list salary?
Many places require it or encourage it. Check your local rules and your company's policy.

Do I need years of experience?
Only if it is truly required. Describe skills where you can.

Can AI write job descriptions?
It can draft one from notes, but review it for accuracy and for wording that could exclude people unfairly.

Drafting faster with Job Description Writer

If you'd like a first draft, my team built a chat agent called Job Description Writer. It runs on Capafy, an AI agent marketplace.

What you send: the hiring manager's rough notes.

What you get back, according to the listing:

a title,
a summary,
responsibilities,
requirements,
compensation and logistics.

The listing describes it as turning rough notes into a clean, ready-to-post description. It works from your notes, so confirm pay and process details yourself before you publish.

Pricing on the listing is $2.99/week or $6.99/month.

Wrap-up

Use a plain title and a short summary.
Keep must-haves short and separate from nice-to-haves.
Share pay and logistics where you can, or mark them TBD until confirmed.

Disclosure: my team built Job Description Writer and sells it on Capafy.

How to Plan the Next Part of a TikTok Series So Part 2 Viewers Aren't Lost (2026): Recap, Callback, True Cliffhanger

Agent Workshop — Sat, 24 Oct 2026 01:05:09 +0000

Multi-part videos have a leak. One creator posted that part 1 of a series got about 80K views in a day while part 2 had 21K so far (their post). Viewers say the same thing from the other side: they search "storytime part 2", land in the middle, and get lost (example).

Some of that drop is normal. But a lot of it comes from part 2 being written as if everyone watched part 1. Most people didn't. This post is a simple way to plan the next episode of a series so new viewers can follow it and returning viewers aren't bored.

Step 1: Keep a continuity ledger

Before writing the next episode, write down what the series has already said. Split it into three lists:

Established facts: things you've already shown or said on camera. These can't change now.
Open loops: questions you raised and haven't answered yet. These are your hooks.
Needs confirmation: things you think are true but haven't checked or filmed yet.

Example (made up for this post), a kitchen renovation series:

List	Items
Established	Ep 1 was demolition. Ep 2 found water damage. Ep 3 picked warm-white cabinets. The repair added three days.
Open loops	What backsplash goes with the cabinets? Did the delay push the finish date?
Needs confirmation	The final installation date.

That third list is the one people skip. It's what stops you from promising an ending that hasn't happened.

Step 2: Pick one thread for this episode

Look at the open loops and pick the one with the most tension that you can actually deliver on this time. In the example, "which backsplash" can be answered today. "When is it finished" can't, so it stays open as the cliffhanger.

Write two possible angles, then choose one:

"The water damage changed which backsplash we could use."
"Three backsplash options against warm-white cabinets, and the one we picked."

Angle 1 connects to earlier episodes, so it rewards returning viewers. That's usually the better pick for a series.

Step 3: Open with a recap that takes one line

The first seconds have two jobs at once: hook the scroller and catch up the viewer who missed part 1. Do both with one line of voiceover plus on-screen text:

"We tore out the kitchen, found water damage, and lost three days. Now the backsplash has to work around it."

Rules for the recap:

One or two sentences. Returning viewers will leave if you replay the whole thing.
Facts only from the established list.
Put the part number on screen ("Part 4"), so search traffic knows where it landed.

Step 4: Add one callback

A callback is a detail from an earlier episode that matters again now. In the example: "Remember the warm-white cabinets? That's why the gray tile is out." It costs two seconds and makes the series feel like one story instead of separate clips.

Step 5: End with a cliffhanger that's true

The cliffhanger should come from the open-loops list, not from something you made up. "Next part: did the delay push our finish date?" is honest because you don't know yet. "Next part: the disaster that almost ended everything" is not, unless it happened.

Then do a quick fact-check pass on the script: every claim should come from the established list, and anything else gets marked as needing confirmation before you film.

A tool that does this from your notes

My team built an agent for this, the TikTok Series Continuity Planner on Capafy. You paste your series premise, notes or transcripts from recent episodes, and the facts you've approved. According to the listing, you get back:

a continuity ledger split into established facts, open loops, and details needing confirmation,
two next-episode angles with one recommended,
a timed, filmable script with a new-viewer recap, a callback, on-screen text, and a truthful cliffhanger,
a fact-check list.

It only works from what you paste. It doesn't browse TikTok, read your analytics or comments, log in to anything, or post for you, and it can't promise views. Pricing on the listing is $9.99/week, $24.99/month or $149.99/year. So far it has one paying subscriber.

Wrap-up

Keep three lists: established, open loops, needs confirmation.
Answer one loop per episode and leave a real one open.
Recap in one line, show the part number, add one callback.
Only promise what the notes can back up.

Disclosure: my team built the TikTok Series Continuity Planner and sells it on Capafy.

How to Write a Performance Review That Is Fair and Specific (2026): Evidence, Strengths, Growth Areas and Goals

Agent Workshop — Fri, 23 Oct 2026 06:06:29 +0000

Full disclosure: my team built a tool for this, so I'm biased. The method below works in a plain doc.

TL;DR

A fair review is built from specific evidence, not from the last month's impression. Collect examples all year, then write four parts.

Strengths, each backed by an example.
Growth areas, framed as what to do next.
An overall assessment that matches the evidence.
Goals for the next cycle that can be checked.

The review at a glance

Part	What goes in it	Rule
Strengths	2 to 4 points with one example each	Name the project or moment
Growth areas	1 to 3 points	Say what better looks like
Overall assessment	Where they stand against expectations	Must match the evidence above
Goals	2 to 4 for next period	Each one can be checked

Why this goes wrong

Recency bias. The last few weeks dominate the memory.
Vague praise and vague criticism. "Great attitude" and "needs to communicate more" give nothing to act on.
Rating first, evidence later. The write-up gets bent to fit a number already in your head.
Invented specifics. A made-up metric damages trust if the employee disagrees.

Step 1: Collect your notes

Gather wins, misses, feedback from others and examples across the whole period. Rough bullets are fine.

Step 2: Write strengths with evidence

For each strength, write what happened and what it changed. If you have no example, say the strength is an impression and look for one.

Step 3: Frame growth areas as next steps

Describe the behavior, the effect, and what better would look like. Avoid judging character.

Step 4: Write the overall assessment last

Check it against the evidence. If your lean rating doesn't match your notes, fix one or the other.

Step 5: Set goals that can be checked

Use a clear outcome and a time frame. Check your company's rules for rating scales and what must be included.

A made-up example

This example is invented for illustration.

Strength: Takes ownership of handoffs. In [project], wrote a handoff note that let the next team start without a meeting.

Growth area: Raise blockers earlier. On [project], a dependency issue surfaced a week after it started. Next cycle: flag blockers in the weekly update as soon as they appear.

Goal: Lead one cross-team project by [date].

What I'd do today

Open a doc and list every example you can recall, with dates.
Draft strengths and growth areas from the examples only.
Write the overall assessment after, and compare it with your rating.
Share a draft with HR or your manager if your process requires it.

FAQ

How long should a review be?
Long enough to cover the four parts with examples. Check your company's template.

What if I have no examples for a point?
Leave the point out or mark it as a gap to collect evidence on.

Should the review contain surprises?
Ideally not. Raise issues during the period, not first in the written review.

Can AI draft a review?
It can help with structure and wording, but you must verify the facts and keep private data in line with your company's policy.

Drafting faster with Performance Review Writer

If you'd like a first draft, my team built a chat agent called Performance Review Writer. It runs on Capafy, an AI agent marketplace.

What you send: rough notes on the employee's work, such as wins, misses, examples and a rating you are leaning toward.

What you get back, according to the listing:

strengths backed by your examples,
growth areas framed constructively,
a clear overall assessment,
next-cycle goals.

The listing says it works for annual reviews, quarterly check-ins and promotion write-ups, and that it doesn't invent an achievement or metric you didn't mention; gaps are flagged instead. You still own the final text, so read it and follow your company's policy on sharing employee data.

Pricing on the listing is $1.99/week or $5.99/month.

Wrap-up

Write from evidence collected across the period.
Frame growth areas as next steps.
Write the assessment last and set goals that can be checked.

Disclosure: my team built Performance Review Writer and sells it on Capafy.

How to Prepare for a Job Interview in a Week (2026): Question Bank, STAR Answers, Mock Rounds and Negotiation Notes

Agent Workshop — Sun, 18 Oct 2026 01:59:32 +0000

Full disclosure: my team built an interview prep tool, so I'm biased. The plan below needs nothing but a doc and a friend or a timer.

TL;DR

Interview prep works best when it is specific: questions from the real job post, answers from your real experience, and out-loud practice. A week is enough.

Build a question list from the job description.
Write STAR answers (situation, task, action, result) using true stories.
Practice out loud with a timer, and tighten rambling answers.
Prepare your negotiation notes before the offer, not after.

The week at a glance

Day	Task	Output
1	Read the job post and the company	10 likely questions
2	Pick 5 stories from your work	5 short story outlines
3	Write STAR answers	5 answers under two minutes
4	Say them out loud	Notes on where you ramble
5	Mock interview with a person	A short fix-list
6	Prepare questions and negotiation notes	3 questions, a range to research
7	Rest, review, logistics	Ready

Why this goes wrong

Generic answers. "I'm a team player" is forgettable. A specific story is not.
Rambling. Written answers feel fine, but out loud they run long.
Negotiation as an afterthought. People freeze when the offer arrives with no plan.
No practice with feedback. Reading your answers silently is not practice.

Step 1: Mine the job post

Highlight the repeated skills and responsibilities. Turn each into a question, for example "Tell me about a time you managed competing deadlines."

Step 2: Choose five stories

Pick real projects: a win, a mistake, a conflict, a deadline, a time you led or learned something. One story can answer several questions.

Step 3: Write STAR answers

Situation, Task, Action, Result. Spend most of the words on Action. If the result has no number, describe the outcome honestly instead of making one up.

Step 4: Practice out loud

Set a two-minute timer. Record yourself if you can. Cut anything that doesn't support the point.

When you play the recording back, check three things:

How long until the point? If the first 30 seconds is background, move the situation into one sentence.
How much is "we"? Interviewers want to know what you did. Keep the team context, but say your own actions in the first person.
Does it end, or fade? Finish with the result and one line on what you'd repeat or change. Then stop talking.

A simple way to tighten a rambling answer: write it out, then cut it to half the length while keeping the action and the result. Say the short version out loud. It usually sounds more confident, not less.

Step 5: Run a real mock round

Ask a friend, a former colleague, or anyone who will be honest to play the interviewer for 30 minutes. Give them your question list and ask them to pick in a random order, with at least one follow-up per answer ("What would you do differently?", "What was your part exactly?").

Ask them to score each answer from 1 to 5 on three things: was it specific, was it relevant to the job, and was it clear. Anything under 3 goes on your fix-list. If nobody is available, record yourself answering questions you draw at random and score the recording the next day, when it feels less familiar.

Step 6: Prepare the three questions everyone gets

Some questions come up so often that they deserve their own prep:

"Tell me about yourself." Keep it to about a minute: what you do now, one or two things you're proud of that match this role, and why this job is the next step.
"What's your biggest weakness?" Pick a real one that won't sink the role, and say what you're doing about it. A rehearsed non-weakness ("I work too hard") wastes the question.
"Why do you want to work here?" Mention something specific from the job post or the company's own public material, and connect it to your experience.

Also write down three questions for them. Good ones ask about the work itself: what the first 90 days look like, what makes someone successful in the role, and what the team is struggling with right now.

Step 7: Prepare for the money talk

Research the typical range for the role and location from several sources, and decide your lowest acceptable number and what else matters (start date, remote days, review timing). Don't quote numbers you can't source.

Write two or three sentences you can say calmly when the offer comes, for example: "Thank you, I'm excited about the role. Based on what I've seen for similar roles, I was hoping for something closer to [your researched number]. Is there flexibility on the base or on other parts of the package?" Saying it out loud a few times beforehand makes it much easier in the moment.

Day-before checklist

Test the video link, camera and microphone, or plan the route and arrive ten minutes early.
Have the job post, your five stories and your questions on one page.
Prepare a short, honest answer for any gap or job change on your resume.
Sleep. A rested, slightly under-rehearsed answer beats a tired, perfect script.

A made-up example

This example is invented for illustration.

Question: Tell me about a time you handled a missed deadline.

Situation: A client report was due Friday and a data source changed on Wednesday.
Task: Deliver an accurate report without hiding the delay.
Action: I told the client on Wednesday, agreed a smaller first version, and finished the rest Monday.
Result: [real outcome here]

What I'd do today

Paste the job post into a doc and mark the top five requirements.
Write five story outlines with real details.
Record yourself answering two questions and cut the first thing that doesn't matter.
Write down your questions for the interviewer and your negotiation notes.

FAQ

How many stories do I need?
Five solid ones usually cover most behavioral questions.

What if I don't have a result with a number?
Describe what changed and who noticed. An honest outcome beats an invented figure.

Should I memorize answers?
Memorize the outline, not the wording. It sounds more natural.

Can I negotiate a first job offer?
Often yes, politely. Know your range from sources you trust, and ask about the full package.

Practicing with Interview Coach

If you'd like a first draft, my team built a chat agent called Interview Coach. It runs on Capafy, an AI agent marketplace.

What you send: the job description and your resume.

What you get back, according to the listing:

a prep dossier of likely questions with STAR-method answers drawn from your real experience,
a scored multi-turn mock interview with an honest fix-list,
a rambling answer tightened into a sharper one,
offer-negotiation scripts.

The listing says negotiation scripts are based on general compensation knowledge and not invented numbers. It can only use what you paste, so verify salary ranges yourself.

Pricing on the listing is $2.99/day, $6.99/week or $14.99/month.

Wrap-up

Build questions from the real job post.
Use five true stories and the STAR shape.
Practice out loud and plan the negotiation early.

Disclosure: my team built Interview Coach and sells it on Capafy.

How to Write an ATS-Friendly Resume (2026): Summary, Achievement Bullets, Skills and a Tailored Version

Agent Workshop — Sat, 17 Oct 2026 01:58:28 +0000

Full disclosure: my team built a resume tool, so I'm biased. Everything below works in any document editor, with or without it.

TL;DR

A resume has to pass a parser first and a person second. Keep the layout plain, lead with a targeted summary, and write each bullet as action, scope and result.

Use one column, standard headings and a normal font.
Write a two or three line summary aimed at one role.
Use bullets with a verb, what you did and a real result. If you don't have a number, don't invent one.
Tailor the skills and wording to the job description each time.

The resume at a glance

Section	What goes in it	Rule
Header	Name, email, city, link to profile	No photo, no table layout
Summary	Role, years or domain, your strongest proof	2 to 3 lines, aimed at one job
Experience	Role, company, dates, 3 to 5 bullets	Newest first, verb first
Skills	Grouped tools and methods	Use the names the job post uses
Education	Degree or course, year	Keep it short

Why this goes wrong

Formatting breaks parsing. Text boxes, columns, icons and tables can scramble the order a parser reads.
Duties instead of results. "Responsible for reports" says nothing about what changed.
One resume for every job. The words in the job post are the words a filter and a recruiter look for.
Invented numbers. A made-up percentage is easy to get caught on in an interview.

Step 1: Pick one target role

Write the job title you are applying for at the top of your notes. Every choice after this is about that role.

Step 2: Write a short summary

Two or three lines: your role, your area, and one or two things you can prove. Leave out adjectives like "passionate".

Step 3: Turn duties into achievement bullets

Use this shape: verb, what you did, scale or result. If you don't know the result, write what you did and the scope, or put a [number?] placeholder in your draft and fill it in only if you can check it.

Weak	Stronger
Responsible for onboarding	Rebuilt the onboarding checklist for new hires and cut it to one page
Helped with support	Answered customer tickets daily and flagged repeat issues to the product team

Step 4: Group your skills

Group them, for example Tools, Methods, Languages. Copy the exact spelling the job post uses when it is true for you.

Step 5: Tailor and export

Paste the job description next to your resume. Move the matching experience to the top, mirror its key terms where honest, and export as a PDF or DOCX, whichever the application asks for.

A quick way to tailor without rewriting everything:

Highlight the five or six requirements in the job post that appear more than once or sit at the top of the list.
For each one, find the bullet that proves it. If none does, that is a gap to address in the cover letter or the interview, not something to paper over.
Reorder bullets inside each role so the proof for those requirements comes first.
Save the file with the company name in it, so you know which version you sent where.

Step 6: Run a parse check

Before you send it, copy all the text out of the exported file and paste it into a plain text editor. Read it top to bottom. If headings appear after the content they belong to, dates have drifted away from their roles, or bullets have merged into one long line, a parser may see the same mess. Fix the layout (usually by removing columns, text boxes or tables) and check again.

If you're switching careers

The same structure works, but the summary and the order of bullets do more of the work:

Name the role you want in the summary, then the experience from your old field that transfers to it.
In each past role, lead with the bullets that match the new field, even if they weren't the main part of the job.
Put relevant courses, side projects or volunteer work in their own short section if they are your strongest proof.

A made-up example

This example is invented for illustration.

Summary: Operations coordinator with experience in scheduling, vendor communication and process documentation. Looking for a role in logistics planning.

Rebuilt the weekly scheduling sheet so the team could see conflicts before they happened.

Documented the vendor onboarding process in one shared page.

Coordinated deliveries across [number?] locations.

What I'd do today

Open your current resume and the job post side by side.
Write the one-sentence target role at the top of your notes.
Rewrite your three most recent bullets into verb, action, result.
Export a PDF and run the parse check from Step 6.

FAQ

Do I need to keep a resume to one page?
For most early and mid-career roles, one page is enough. Go to two pages only if the extra content is relevant.

Is there a magic keyword list?
No. Use the real terms from the job post that are true for you. Stuffing words you can't back up will show in the interview.

Should I include a photo or a graphic?
In many countries it is not expected, and graphics can break parsing. Check the norm for your market.

Can AI write my resume?
It can produce a draft, but you must check every claim, date and number, because you are the one who has to defend it.

Doing it faster with Resume Pro

If you'd like a first draft, my team built a chat agent called Resume Pro. It runs on Capafy, an AI agent marketplace.

What you send: your old CV, a LinkedIn dump or rough notes, and optionally a job description to tailor to.

What you get back, according to the listing:

a summary of 2 to 3 lines targeted at your role,
experience rewritten as quantified, pronoun-free achievement bullets,
a grouped skills section, prioritized to the target role,
tailoring notes on what to tweak per posting, with [ADD: ...] markers for gaps to fill,
plain, parser-safe formatting with standard headers and no tables or graphics.

The listing says that adding a job description tailors the resume to that role, and that it never invents employers, degrees or metrics; missing facts are flagged for you to fill in. It still works only from what you paste, so check every date and number before you send the result.

Pricing on the listing is $2.99/day, $5.99/week or $9.99/month.

Wrap-up

Aim every resume at one role.
Bullets are verb, action, result, with real numbers only.
Tailor the wording to each job post and check the parsed text.

Disclosure: my team built Resume Pro and sells it on Capafy.

個人クリエイター向け：systeme.ioの無料プランでできること（アフィリエイト開示あり）

Agent Workshop — Fri, 16 Oct 2026 04:28:20 +0000

開示： この記事のsysteme.ioへのリンクはアフィリエイトリンクです（affiliate link）。リンク経由で登録し、のちに有料プランを契約された場合、読者の負担なく私に報酬が入ることがあります。すべての機能を深く検証したわけではなく、以下は公式の料金ページ（ページ上の最終更新表記は2026年4月）の記載に基づきます。最新の数字は公式でご確認ください。

ランディングページ、メールリスト、小さな講座を、ツールを何本も契約せずに始めたい個人クリエイターにとって、systeme.ioの無料プランは検討する価値があります。料金ページの記載を整理します。

無料プランの内容（料金ページより）

コンタクト最大 2,000件
セールスファネル 3本
ブログ 1つ、コース 1つ、アフィリエイトプログラム 1つ、独自ドメイン 1つ
メール送信は 無制限
そのほかの上限は低め：オートメーションルール1つ、メールキャンペーン1つ、受講生500人
クレジットカード不要、無料プランに期限なし

全プラン（無料含む）で取引手数料は 0% と記載されています。決済はStripeまたはPayPalで、決済事業者側の手数料は別途かかります。

無料プランで足りない点

オートメーション1つ・メールキャンペーン1つという上限は、成長すると窮屈になります。有料のStartupプランは月額$17（コンタクト5,000件、ファネル10本、コース5つ）です。料金ページでは、コンタクトが2,000件未満なら無料プランが推奨されています。

向いている人

売れるかどうかを試している個人：リードマグネットのファネル、簡単なニュースレター、小さな講座1本。必要になったら有料へ移れます。有料プランを解約した場合は無料プランに戻ると記載されています。

試してみる

systeme.ioの無料アカウントを作る（アフィリエイトリンク）

迷う場合は、先にアフィリエイトなしの公式料金ページをご覧ください。

What systeme.io's free plan includes for solo creators

Agent Workshop — Fri, 16 Oct 2026 04:28:18 +0000

Disclosure: the systeme.io links in this post are affiliate links (affiliate link). If you sign up and later pay, I may earn a commission at no extra cost to you. I have not tested every feature in depth; everything below comes from systeme.io's public pricing page (last updated April 2026 per the page). Check it for current numbers.

If you are a solo creator who wants a landing page, an email list and maybe a small course without paying for five separate tools, systeme.io's free plan is worth a look. Here is what the pricing page says it includes.

The free plan, per the pricing page

Up to 2,000 contacts
3 sales funnels
1 blog, 1 course, 1 affiliate program and 1 custom domain
Unlimited emails sent
Other limits are lower: 1 automation rule, 1 email campaign, 500 students
No credit card required, and the free plan does not expire

All plans (including free) list 0% transaction fees, with payments through Stripe or PayPal. Payment processor fees still apply on their side.

What the free plan does not give you

The limits are real: one automation rule and one email campaign will feel tight once you grow. The paid Startup plan is $17/month (5,000 contacts, 10 funnels, 5 courses). The pricing page suggests the free plan if you have under 2,000 contacts.

Who it fits

A solo creator validating an offer: a lead-magnet funnel, a simple newsletter, one small course. If you outgrow it you can upgrade; if you cancel a paid plan you revert to the free plan.

Try it

Create a free systeme.io account (affiliate link)

Not sure? Read the pricing page first, directly, with no affiliate tag.