Get tasks doneBusiness size: Solo

London solo consultant runs his client notes, tasks and webinars from one AI-maintained folder and says proposals now take 30 minutes, not 4 hours

Nick Zervoudis · United Kingdom · Claude Code, Codex, Obsidian, Granola, Wispr Flow, Telegram, Gmail, Circle, Luma, Lovable

Go to the skill

What the agent did

Working from one folder of plain text files, Claude Code or Codex updates a note for each company, person and opportunity after every synced call transcript, creates tasks with priority and an estimated duration, starts promised work such as intro emails, and writes a daily brief that also arrives on Telegram. For webinars it makes the banners, drafts the descriptions and publishes the event to Circle and three Luma calendars once he says publish.

Result, as claimed

His own estimates, shared with Create With: proposal prep down from about 4 hours to about 30 minutes, an event banner and descriptions from about 40 minutes to about 5, and delivery effort down about 45% on one engagement. Average monthly revenue in the first half of 2026 ran at 2.4 times the 2025 monthly average, which he credits partly to a young business growing. His community is on course for about 33 virtual events in 2026, against 8 in 2025. The setup costs around £250 a month.

This is the result as claimed in the public source linked below. It has not been checked and is not a typical outcome.

Proof

Read the public source

Try it yourself

Keep one plain-text file per client and have your AI update it after every call, then draft the follow-ups for you to send.

Difficulty: Beginner

Prompt

I run [type of business]. Here is my client file for [client name]: [paste file, or write none yet]. Here are my notes or transcript from today's call: [paste]. Update the file with who the people are, what each cares about and where the work stands, quoting their own words for the problem. List commitments, mine and theirs, with any dates. Draft each follow-up email I promised, in my voice, using these samples: [paste two of my emails]. Mark every draft DRAFT. If a name or detail is missing, ask me instead of guessing. Do not send anything. Return a draft only. Never send, spend, change settings or write to a live system.

Prompt: starter by Solopreneur Code

Tools

  • Claude Code Source-named tool. Official product URL not independently verified during intake, omitted.
  • Codex Source-named tool. Official product URL not independently verified during intake, omitted.
  • Granola Source-named tool. Official product URL not independently verified during intake, omitted.

Steps

  1. Write one text file for one client: people, what you are doing, where it stands.
  2. After the next call, paste the file and your notes or transcript into the prompt.
  3. Check the updated file, task list and drafts line by line.
  4. Send approved emails yourself from your own inbox.
  5. When something comes out wrong, add a standing rule to the file rather than only fixing that output.

Watch out: The source reports the AI invented a surname during one job. Check every name and figure against your own records before sending.

Starter version by Solopreneur Code

Skill: Sales and quotes

Add this skill

Sales and quotesDifficulty: Beginner

Keep one client file current after each call

Keeps one file per client current after each call, lists commitments, turns yours into tasks and drafts follow-ups you send yourself.

Works with
Claude, ChatGPT, Gemini, Any AI chat The story used Claude Code and Codex on a markdown repo with Granola transcripts. In a normal AI chat you paste the file and notes and save the result yourself. Not tested.
Tools in the story
Claude Code, Codex, Granola, Gmail
Country
United Kingdom

How to use this: Copy the skill, paste it into Claude, ChatGPT or Gemini, and ask it to save it as a skill. If it cannot, it will still follow the steps.

Show skill

If the Copy button does not work, select this text and copy it.

Sign up for the newsletter

Get the next story from Solopreneur Code in your inbox.

Form not loading? Subscribe on Substack

Back to all stories

More like this

Browse all stories
Get tasks doneBusiness size: SoloIncludes prompt

A marketing educator has Claude route her inbox into CRM deals, tasks and draft replies, and says agents save her 3 to 4 hours a week

Cristina Peñaloza Morales · Country unknown · Claude, Gmail, HubSpot, Todoist, Notion, Riverside, Gamma

Reads new email and routes each one. Training enquiries become a HubSpot contact and deal with a drafted reply. Community messages get a draft reply. Possible collaborations become a Todoist task. Invoices and admin get a label. Also runs a scheduled Monday AI news digest.

Result, as claimed

She says agents save her between 3 and 4 hours a week.

Get tasks doneBusiness size: SoloIncludes prompt

A newsletter writer's Claude Code agent turned screenshots into 38 scheduled Substack Notes he did not write, with no engagement lift

Dheeraj Sharma @genaiunplugged · Country unknown · Claude Code, SubflowAI Substack connector, macOS scheduler, Substack Notes

Each morning a headless Claude Code run takes the oldest screenshot, classifies it, writes a 45 to 115 word Note in his voice, checks it against copy gates, uploads the image, schedules it for the next free day, reads the schedule back and archives the screenshot. Screenshots with emails, phone numbers or orders are set aside.

Result, as claimed

29 Notes scheduled by 23 Sep and 38 by 12 Oct with none written by him. 12 measured Notes had a median of 3.5 likes against 5 for his other Notes, which he reads as no lift.

Get tasks doneBusiness size: SoloIncludes prompt

A Fukuoka developer had Claude Code build a bookkeeping pipeline that reconciled 200 transactions and flagged its own doubts

Ken Imoto @kenimo49 · Japan · Claude Code, MCP connection to his accounting platform (platform not named)

Built a re-runnable pipeline that pulls the month's transactions over MCP, matches bank rows to receipts by amount and date, proposes a category with a one-line reason for each row and flags rows it isn't sure about.

Result, as claimed

200 transactions reconciled and 11 miscategorised, a 94.5% hit rate. Review shrank to 18 flagged rows plus a sample. 7 of the 11 errors were in the flagged pile and 4 were caught on the sample.