DEV Community: Agent Workshop

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.

How to Turn Messy Meeting Notes Into Action Items (2026): A TL;DR, Decisions, Owners and Open Questions

Agent Workshop — Thu, 15 Oct 2026 23:51:50 +0000

Full disclosure: my team built a tool that does this, so I'm biased. The format below works in any doc or chat, with or without it.

TL;DR

A good meeting recap has four parts and takes a few minutes to write. Send it the same day, and mark anything you don't know as TBD instead of guessing.

TL;DR: two or three sentences.
Decisions: only what was actually agreed.
Action items: a table with task, owner, due date.
Open questions: what is still unanswered, and who has to answer.

The recap at a glance

Part	What goes in it	Rule
TL;DR	What the meeting was about, what changed	2 to 3 sentences
Decisions	One line each, with the reason if you have it	Only what was agreed
Action items	Task, Owner, Due, Notes	One clear next step per row
Open questions	What was raised but not resolved	Name who needs to answer

Why follow-ups slip

Notes are a transcript, not a plan. Raw notes record what was said. Nobody can act on "talked about the pricing page."
Owners are implied. "We should update the doc" has no owner, so nobody does.
Dates are vague. "By Friday" means different things on a Tuesday and a Thursday.
Decisions and ideas get mixed. Readers can't tell what's settled and what's still being discussed.

Step 1: Write the TL;DR last, in two or three sentences

Say what the meeting was about and what changed. If nothing changed, say that. It's the part most people will read, so keep it plain.

Step 2: List decisions only if they were agreed

A decision is something the group settled, not something one person suggested. One line each, and add the reason if it's in your notes. If you aren't sure it was agreed, put it under open questions instead.

Step 3: Make every action item a verb with an owner

Use a small table:

Task	Owner	Due	Notes
Send revised quote to the client	[Name]	[Date]	Use the new scope
Check whether the vendor can deliver by month end	TBD	TBD	Ask in the team channel

Start each task with a verb. If the owner or date wasn't stated, write TBD. A visible TBD gets fixed. A guessed owner creates a wrong assumption.

Step 4: Convert relative dates

"By Friday" and "next week" lose meaning in a few days. Write today's date at the top of your notes and convert them to real dates when you write the recap.

Step 5: List what's still open

Anything raised but not resolved goes here, with the name of the person who needs to answer. This also keeps decisions clean.

A made-up example

This example is invented for illustration.

TL;DR: We reviewed the launch checklist. The date moved by a week because the review isn't finished.

Decisions

Launch moves to [new date].

Action items
| Task | Owner | Due | Notes |
|---|---|---|---|
| Finish the final review | [Name] | [Date] | Needs the updated copy |
| Tell the partner about the new date | TBD | TBD | Decide who sends it |

Open questions

Do we still need legal sign-off? ([Name] to check)

What I'd do today

After your next meeting, write the date at the top of your notes.
Fill in the four parts in under ten minutes.
Turn every TBD into a name or a question in the team chat.
Send the recap the same day, in the channel where the team already works.

FAQ

How long should a recap be?
Short enough that people read it. One screen is a good target.

Should I include everything that was said?
No. The recap is for decisions and next steps. Keep the raw notes or transcript separately if you need them.

What if two people disagree on who owns something?
Put it under open questions and name who resolves it.

Can AI summarize meetings?
It can, but check names, dates and decisions against your own notes before you share it, especially if the notes were incomplete.

Doing it faster with Meeting Notes to Action Items

If you'd like a first draft, my team built a chat agent called Meeting Notes → Action Items & Decisions. It runs on Capafy, an AI agent marketplace.

What you send: your notes or a transcript, plus optional context such as attendee names, today's date, and the format you want (Slack, email, doc).

What you get back, according to the listing:

a short TL;DR,
decisions that were actually agreed,
an action-items table with task, owner and due date,
open questions that are still unanswered.

The listing says it takes owners and dates only from what you paste and marks unknowns as TBD instead of guessing, and that a date you provide turns "by Friday" into a real date. It works from the text you send, so a thin set of notes gives a thin recap, and you should still read it before sharing.

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

Wrap-up

Four parts: TL;DR, decisions, action items, open questions.
Every action has a verb, an owner and a date, or a visible TBD.
Convert relative dates and send the recap the same day.

Disclosure: my team built Meeting Notes → Action Items & Decisions and sells it on Capafy.

How to Write a Cold Email That Gets Replies (2026): Subject, Opener, One Ask, and a Follow-Up Cadence

Agent Workshop — Thu, 15 Oct 2026 12:13:31 +0000

Full disclosure: my team built a tool that writes cold email, so I'm biased. I'll show you the method first, and it works fine with just a text editor.

TL;DR

Most cold email gets ignored because it reads like a template. A reply-worthy email has five parts in a fixed order, stays short, and asks for one small thing.

Subject: short and specific, no clickbait.
Opener: a real reason you're writing to this person.
Value: one concrete outcome tied to their world.
Proof: only proof you actually have.
Ask: one low-friction question.

The email at a glance

Part	What it does	Rough length
Subject	Gets the open without tricking anyone	2 to 6 words
Opener	Shows this wasn't mail-merged	1 sentence
Value	One specific outcome for them	1 to 2 sentences
Proof	Makes the claim believable	1 sentence, or leave it out
Ask	Makes replying easy	1 sentence

Aim for roughly 50 to 125 words total, so it reads easily on a phone.

Why most cold email fails

The opener is filler. "Hope you're well" tells the reader nothing and signals a template.
It pitches three things. If the reader has to decide which one matters, they decide none of them do.
The ask is heavy. "Can we do a 30-minute demo next week?" is a big request from a stranger.
The proof is invented. Vague claims like "companies like yours see big gains" are easy to spot and easy to distrust.

Step 1: Write the subject last, and keep it plain

Write two options. Say what the email is about in the words the recipient would use. Avoid all caps, fake "Re:" prefixes and urgency tricks. A subject like Question about onboarding at [Company] beats a clever one that gets ignored.

Step 2: Open with a real reason

Use something true and specific: their role, a post they wrote, a launch, a hiring page, a tool they use. If you can't find one, don't fake it. Use a placeholder and go find it:

[ADD: one specific reason I'm emailing this person]

Step 3: One value, one sentence

Say what changes for them, not what your product is. "Cut the time your team spends on X" is about them. "We're an AI-powered platform" is about you. Keep it concrete, and don't promise numbers you can't back up.

Step 4: Proof only if it's real

If you have a real customer, result or number, use it once. If you don't, say nothing rather than stretch. Mark the gap so you remember to fill it later: [ADD: real proof point].

Step 5: Ask one small question

Good asks are easy to answer: "Is this something you own, or is someone else the right person?" or "Worth sending a two-line example?" Put it last, and make it a single question.

A made-up example

This example is invented for illustration. The company and facts are placeholders.

Subject: Question about onboarding at [Company]

Hi [Name],

I saw [Company] is hiring two support agents, so I'm guessing ticket volume is growing.

We help small support teams draft first replies faster, so new hires ramp up without a long training period.

[ADD: one real customer or result, if you have it]

Is support something you own, or is there someone else I should ask?

[Your name]

A follow-up cadence that isn't pushy

A sequence of 4 to 6 emails over a couple of weeks is common. The key is to change the angle each time instead of repeating "just following up":

First touch: the email above.
A few days later: a different benefit, or a short proof point.
A few days after that: a different angle, such as a question about their current process.
A short bump that makes it easy to say no.
A polite break-up note that closes the loop.

Space them out, and stop when someone asks you to. Check the rules for the country you're sending to, including unsubscribe requirements, before you run any outbound campaign.

What I'd do today

Pick one segment and one person. Write the opener from something you actually found.
Draft the email using the five parts, and cut it to under 125 words.
Replace every guessed fact with an [ADD: ...] marker, then fill those in or delete the line.
Send ten, read the replies, and change one thing at a time.

FAQ

How long should a cold email be?
Short enough to read on a phone, roughly 50 to 125 words. Cut anything that isn't the opener, value, proof or ask.

How many follow-ups should I send?
Often 3 to 5 after the first email, with a different angle each time. Stop if the person asks you to.

Should I mention a competitor?
Only if it's relevant and accurate. Don't make claims about a competitor you can't back up.

Can AI write cold email?
It can draft one, but you need to supply the real facts. Check every claim about the recipient before you send.

Doing the drafting faster with Cold Email Writer

If you'd like a first draft, my team built a chat agent called Cold Email Writer: Get Replies, Not Templates. It runs on Capafy, an AI agent marketplace.

What you send: who you're emailing, what you offer, and the goal.

What you get back, according to the listing:

B2B cold-email copy with two subject options,
a relevant opener, one specific value, and a low-friction ask,
three modes: a single first-touch email, a full 4 to 6 email cadence with rotating angles, or a rewrite of an email you already have,
notes on send spacing, what to personalize, and any [ADD: ...] gaps.

The listing says it uses only the proof you give it and never invents recipient facts or fake stats. It writes copy only: it doesn't send email or find leads, and you're responsible for what you send.

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

Wrap-up

Five parts, in order: subject, opener, value, proof, ask.
One ask, one benefit, under about 125 words.
Mark unknowns with a placeholder instead of guessing.
Change the angle in each follow-up.

Disclosure: my team built Cold Email Writer and sells it on Capafy.

Stop Running Marketing Checklists: How to Find the One Bottleneck in Your Funnel

Agent Workshop — Thu, 15 Oct 2026 03:57:43 +0000

Traffic is flat, so you read a "20 ways to grow" list. Then you start three of them at once: a new blog post, a few ad tests, a redesigned pricing page. A month later you can't tell which one mattered, or whether any did.

The problem usually isn't effort. It's that a checklist treats every item as equally important, when in most funnels one stage is holding everything else back. Fixing the wrong stage can feel productive and change nothing.

This post walks through a simple way to find that stage, pick one move, build the asset for it, and decide in advance how you'll know it worked.

Why checklists fail

They ignore your numbers. A checklist item like "improve SEO" is advice for a site that has no traffic problem and advice for one that does. It can't tell them apart.
They encourage parallel work. Five changes at once means no clean read on any of them.
They hide the real constraint. If 40 people visit and none sign up, more traffic is the wrong fix. If signups are fine but nobody comes back, so is a better landing page.

Step 1: Write your funnel as plain stages

Keep it short. For most products it looks something like this:

People find you (traffic or awareness)
They sign up or start (conversion)
They get to the first useful moment (activation)
They come back or pay (retention or revenue)

Then write down the rough count at each stage for a recent, fixed period, such as the last 30 days. Use real numbers from your analytics, and if you don't have one, write [ADD: number] and go find it. Don't estimate and carry on as if it were fact.

Step 2: Find the stage that leaks the most

Look at the step-to-step drop, not the totals. A simple example (made up for this post, not real data):

Stage	People	Moved to next stage
Visited the site	1,000	50 signed up
Signed up	50	8 reached the first useful moment
Reached first useful moment	8	3 paid

In this example the biggest relative drop is between signup and the first useful moment. Sending more visitors would just fill the top of a leaky bucket. The stage to work on is activation.

Your own numbers will point somewhere different. The method is the point: pick the stage where the most people are lost relative to what a reasonable product would keep, and be honest if you can't tell yet. If your sample is tiny, the right answer might be "I don't have enough data, so the move is to get more of it."

Step 3: Choose one play, and say why it should work

Resist listing five options. Pick one and write two sentences:

The tactic: what you'll change (for example, shorten onboarding to a single action that delivers the first result).
The reason: the behavior behind it. People drop off when the effort to the first reward feels bigger than the reward they expect. Reducing steps lowers the perceived cost.

If you can't state the reason, you're guessing. That's fine sometimes, but call it a guess and keep the test small.

Step 4: Draft the asset before you start

A plan without an asset stalls. Write the thing you'll ship first: the new onboarding copy, the email subject line, the changed headline, the one-paragraph message to a user who dropped off. A rough version is enough. You'll improve it after you see how people respond.

Step 5: Name one metric and a check date

Pick the single number that should move if the play works, and the date you'll look at it. For the example above, it might be "share of new signups who complete the first action within a day," checked after two weeks.

Write down what result would make you keep the change, and what would make you drop it. Deciding this before you see the data keeps you from explaining away a flat result.

A short checklist

[ ] My funnel is written as plain stages.
[ ] Each stage has a real number for the same period, or an [ADD: …] marker.
[ ] I picked the stage with the biggest leak, not the most interesting one.
[ ] I chose one play and wrote the reason in a sentence.
[ ] The first asset is drafted.
[ ] I named one metric and a date to check it.
[ ] I'm not starting a second change until this one is read.

Doing the diagnosis faster with Marketing Strategist

If you'd like a second pass on the thinking, my team built a chat agent for it called Marketing Strategist — The One Move to Make. It runs on Capafy, an AI agent marketplace.

What you send: what's stuck, such as traffic, signups, activation, ads, SEO, a launch, or conversion, along with whatever numbers and context you have.

What you get back, according to the listing:

a diagnosis of the likely bottleneck in your funnel,
one recommended play, justified by a proven tactic and the psychology behind it,
a drafted asset to execute it,
the metric to watch.

The listing says it reasons from your input and general marketing knowledge, doesn't browse your site, and doesn't invent numbers. So the quality of the answer depends on the detail you give it, and you still decide what to ship.

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

Wrap-up

Write the funnel as stages and put real numbers on them.
Work on the biggest leak, not the most appealing idea.
One play, one reason, one asset, one metric, one check date.
Don't start the next change until you've read this one.

Doing less, in the right place, usually teaches you more than a long list done halfway.

Disclosure: my team built Marketing Strategist and sells it on Capafy.