Learning-in-public · Claude Code

I build AI for complex
product companies — in public.

Short episodes showing exactly how the systems get made — the prompts, the wins, the glitches. Watch how it's built, steal the code, or hire the operator behind it.

Watch the latest episode Work with us
50 episodes 5 proven cases 100% free code
01 The episodes

Everything I've built, one episode at a time

Each episode is one real thing I built with Claude Code — shown end to end, with the exact prompts and the free code to run it yourself.

My AI supervisor burned 9.9 million Fable tokens in a week re-judging tabs that hadn’t changed
EP.51 · SKIP THELATEST

My AI supervisor burned 9.9 million Fable tokens in a week re-judging tabs that hadn’t changed

A scheduled Claude Code run checked my other AI sessions every 15 minutes and re-read the same idle tabs each time: 374 runs and 9.9M Fable output tokens in one week, the weekly limit gone two weeks running. The fix is a content fingerprint per tab and an exit-code gate that only starts the model when something changed.

Read the episode →
My AI listed 12 million files on my NAS, deleted nothing, and found a 4.2 TB folder no listing shows
EP.50 · DELETE KEY

My AI listed 12 million files on my NAS, deleted nothing, and found a 4.2 TB folder no listing shows

My NAS said the disk was nearly full and could not say why. I had Claude Code list every file on it, 12.2 million rows into one SQLite table, with only ls, find and du allowed to run on the NAS. It built a page that sums what I tick. The biggest gap turned out to be Synology Drive version history, 4.2 TB of it, invisible to a normal file listing. It deleted nothing; the delete key stays with me.

Read the episode →
I built a smoke alarm for my AI’s merges, and the first thing it caught was itself
EP.49 · SMOKE ALARM

I built a smoke alarm for my AI’s merges, and the first thing it caught was itself

AI agents merge their own tickets into my dashboard. I wrote a deterministic tripwire that judges every merge after it lands, ran it silently over the last 50 merges, fixed three false fires with a test each, and wired it in as a shadow step that logs and never blocks. Then it flagged its own merge.

Read the episode →
My local voice AI invented 7 answers out of 24. One stricter instruction took it to zero
EP.48 · INVENTED NONE

My local voice AI invented 7 answers out of 24. One stricter instruction took it to zero

I asked my local voice assistant 24 questions about my own day and it made things up in 7 answers. A test that counts lies plus four strict lines in the system prompt took that to zero, and the reply now starts in a third of a second.

Read the episode →
My merge button failed 44 times out of 65. Now it tells me why before I press it
EP.47 · MERGE PREFLIGHT

My merge button failed 44 times out of 65. Now it tells me why before I press it

My AI agents work on branches and one button merges their work to main. It died half way in 44 of 65 presses. A read-only preflight, a gate earlier in the chain and a Fix line on every stop turned the failures into one-line answers, and a tally keeps the score.

Read the episode →
I built a paid transcription service in one evening, and the privacy promise lives in the code
EP.46 · DELETED WHEN

I built a paid transcription service in one evening, and the privacy promise lives in the code

Upload audio, get a transcript, nothing lingers. With Claude Code I built a whole transcription product in 20 commits: audio deleted the moment the job finishes, the transcript on first download, everything else gone within 24 hours. The rules came before the features.

Read the episode →
One secret opened two doors: the convenient default in my gateway config
EP.45 · TWO DOORS

One secret opened two doors: the convenient default in my gateway config

My Pocket Operator gateway had one tidy line: if no billing secret is set, reuse the connect secret. It meant the key for the account-linking door also opened the billing door, and nothing would ever warn me. Now unset means billing is off, and equal secrets make the gateway refuse to start.

Read the episode →
A free local AI improves my web page one rule at a time, and numbers decide what it keeps
EP.44 · KEPT OR

A free local AI improves my web page one rule at a time, and numbers decide what it keeps

I gave a free local model a 179-rule rulebook and one web page. It proposes one small patch at a time, a measuring tool decides keep or revert, git remembers every keep, and the tool has no publish button.

Read the episode →
I stopped guessing which AI model to use: a score database picks the cheapest one that can do the job
EP.43 · NO DATA

I stopped guessing which AI model to use: a score database picks the cheapest one that can do the job

Every prompt in my AIOS went to the big cloud model by habit. I tested 12 free local models on a real job, wrote every score to a database, and built a router that reads it: it names the cheapest model that measured good enough, and when it has no data it says so instead of guessing.

Read the episode →
My AI read 5 live role postings and mapped my own skills against every one
EP.42 · GAP MATRIX

My AI read 5 live role postings and mapped my own skills against every one

I pointed my AI at five live role postings and it built a competency matrix against my own profile — Have, Partial, Missing, ranked by fit — before I'd read a single line. A skills radar, not a scroll.

Read the episode →
I gave my AI a Slack seat - and it still can't send a word without me
EP.41 · SLACK SEAT

I gave my AI a Slack seat - and it still can't send a word without me

The Telegram bot has been the AI's mouth for months, but Telegram is a channel between me and it - Slack is where a decision gets made in a thread on a Tuesday and never gets written down anywhere else. So it got the same read access I have: channels, threads, DMs, full-text search. scripts/slack_client.py wraps the Slack Web API in eight subcommands; the first six are read-only and unguarded, because reading a workspace is a private act. Posting into one is not - it lands in front of other people, under my name, with no undo. So send refuses by default: it composes the message, prints it, and waits for four characters of confirmation. Getting there cost two token traps and a scope typo - a bot token can't see DMs at all (it needs a user token), usergroups:read is not groups:read, and a token pasted where it shouldn't have been took the whole encrypted secrets file down with it.

Read the episode →
I let a fresh AI judge three rewrites of itself
EP.40 · REWRITE SHOWDOWN
Posted 2026-10-01

I let a fresh AI judge three rewrites of itself

My AIOS redesign forked into three parallel prototypes with no clear winner. I brought in Fable 5 — a brand-new Claude model with no history — to scout all three plus production and pick the one that deserved to survive. It did, and then it told me not to build a fourth.

Read the episode →
My AI found a lead I'd let go cold — and wrote the proposal by morning
EP.39 · COLD LEAD
Posted 2026-10-01

My AI found a lead I'd let go cold — and wrote the proposal by morning

My overnight AI audit flagged a consulting proposal I'd let go cold for five weeks. By morning it had workshopped the pitch three ways and generated the full styled proposal — and a human just has to hit send.

Read the episode →
I voice-command any Claude Code window on any screen
EP.38 · VOICE ROUTER
Posted 2026-09-30

I voice-command any Claude Code window on any screen

I said "on screen 2, tab 3: implement the plan" out loud — and the right Claude Code session on the right monitor picked it up. A free spacedesk screen fabric, a Win32 window router, and a regex intercept that never spends a token.

Read the episode →
The deploy job is now a free checkbox
EP.37 · CLOUDFLARE PAGES
Posted 2026-09-30

The deploy job is now a free checkbox

I push one commit to GitHub and thirty seconds later Cloudflare Pages has rebuilt and redeployed the live site — no server, no FTP, no deploy button. The pipeline that used to be a whole DevOps role is a box you tick once.

Read the episode →
99 skills, scattered across 20 folders — one search box found them all
EP.36 · SKILLS REGISTRY
Posted 2026-09-29

99 skills, scattered across 20 folders — one search box found them all

Skills, commands and agents pile up in your global config, inside every project, and in plugin caches — no single view, and duplicates drift out of sync in silence. I had my dashboard build a registry that walks every location, catalogs all 99, and surfaces the strays in one search box.

Read the episode →
One markdown file tracks every open thread in my life — closeable from my phone
EP.35 · OPEN THREADS
Posted 2026-09-22

One markdown file tracks every open thread in my life — closeable from my phone

The proper fix for tracking loose ends is a dashboard heatmap (todo #34) — but that needs a desktop session. So I built the usable-today version first: inbox/open-threads.md, one line per open thread, closed by texting 'close 3' from any device. Writing it down became a free audit — 11 loose threads surfaced the moment it existed, two I'd have sworn were done. The scrappy text file that beat the fancy dashboard. Channel A, on camera.

Read the episode →
My AI audits every project every night — and just learned to spot a fake one
EP.34 · FAKE PROJECT
Posted 2026-09-22

My AI audits every project every night — and just learned to spot a fake one

A master-architect agent sweeps every project folder in the workspace each night to decide what's alive and what's stale. Two folders with perfectly written plan files passed as real work while their content folders held nothing but a .gitkeep — so the discovery filter learned to look inside.

Read the episode →
My finished-looking dashboard was leaking raw markdown
EP.33 · MARKDOWN BLEED
Posted 2026-09-22

My finished-looking dashboard was leaking raw markdown

My AIOS rewrite grew a slick new Todo board — then every row was leaking raw **markdown**: literal asterisks and backticks in a UI that's supposed to look done. The fix: a RichText renderer for the drawer, a plainTitle helper that strips markdown from every row title.

Read the episode →
My AI bot went quiet mid-job — so I closed the loop
EP.32 · BOT GOES QUIET
Posted 2026-09-22

My AI bot went quiet mid-job — so I closed the loop

An overnight render died halfway and my Telegram AI said nothing — a per-message claude -p process exits after replying, so anything it backgrounds dies untracked. I built run-detached-notify.ps1 (one guaranteed Done/FAILED message on finish), then taught the same bot to read my 👍/👎 and mail a weekly report card (weekly-feedback-digest.ps1). The AI's feedback loop, closed: it reports back and it listens back. Channel A, on camera.

Read the episode →
One command ships my whole AI workspace
EP.31 · RELEASE WORKSPACE
Posted 2026-09-02

One command ships my whole AI workspace

One push to GitHub, and /release fans out lint, test, commit, push and deploy across roughly 20 repos — the DevOps job that used to be a manual per-repo chore is a single command now.

Read the episode →
The gate that blocks my own pushes
EP.30 · CI GATE
Posted 2026-09-02

The gate that blocks my own pushes

All the tests passed, I pushed, and two minutes later GitHub mailed a Run failed notice with seven red tests on a public repo. The tests weren't wrong - this laptop is too rich. It has every library and API key I've ever installed, while the CI runner installs exactly three things: pytest, detect-secrets, bandit. So two test files quietly leaned on what only exists here and died on a clean machine. Which makes the obvious fix useless: running them again locally reproduces the same blind spot, because green on my machine is not evidence of anything. The gate is a pre-push hook that runs a throwaway environment holding exactly what the runner holds - same three checks, same order - and if any fail the push never leaves the laptop. Two details make or break it: the hook is pinned to Unix line endings (one stray carriage return and it stops running silently, which is worse than no gate at all) and it's committed as executable, or git just skips it.

Read the episode →
My AI dashboard now shows every service running on my machine — with live port-collision warnings
EP.29 · SERVICES CARD
Posted 2026-09-02

My AI dashboard now shows every service running on my machine — with live port-collision warnings

One card reads every running process, maps it to a service and a port, and flags collisions. It caught two services fighting over port 8080 the instant it loaded.

Read the episode →
The proof page — what changed, not what I built
EP.28 · PROOF OF WORK
Posted 2026-08-14

The proof page — what changed, not what I built

Most portfolios list tools. I built a proof page that shows the delta. Five real wins, each one the same shape: the thinking constraint that was blocking me, the KPI before and after, and the AI system that moved it — 2-4 h to 0 h, 20 min to 3 min, scraping to one command, silent context deaths to zero. Every number links to the episode that proves it. The story of building showcase/cases.html.

Read the episode →
One Cloudflare command = HTTPS in seconds
EP.27 · CLOUDFLARE TUNNEL
Posted 2026-08-14

One Cloudflare command = HTTPS in seconds

My phone voice AI needed a trusted HTTPS URL. I was using Tailscale — account, device enrollment, DNS. Then I found cloudflared quick tunnels: zero account, one command, a live URL with a real cert in seconds. Wrapped it in run-tunnel.ps1 and ripped Tailscale out.

Read the episode →
The skill that writes my movie scripts
EP.26 · SEEDANCE SCRIPT
Posted 2026-08-05

The skill that writes my movie scripts

One sentence in, a full Seedance 2.0 shot list out. I built a Claude Code skill that writes the script for any AI-video idea — copy-paste prompts, CapCut steps, a credit budget. Here's the anatomy.

Read the episode →
My AI dashboard reads every conversation I've ever had — and resumes them from the browser
EP.25 · RESUME SESSIONS
Posted 2026-08-05

My AI dashboard reads every conversation I've ever had — and resumes them from the browser

I built a dashboard button that opens any past Claude Code session in the browser, shows the full parsed conversation — tool calls and all — then resumes it in the terminal with one click.

Read the episode →
I used Claude to diagnose my boat
EP.24 · DIAGNOSE MY BOAT
Posted 2026-08-05

I used Claude to diagnose my boat

My outboard died twice at sea with no fault code and restarted instantly. I fed the real manuals to Claude Code — it built a cited knowledge base of the whole boat, ranked six suspects (kill-switch stop-circuit #1), and handed me a free dockside test plan before any mechanic. Off-brand domain, on-brand method: AI debugs a real-world problem from the manuals up. Channel A, on camera.

Read the episode →
My AI shops for the cheapest brain for the job
EP.23 · LLM ROUTING
Posted 2026-08-01

My AI shops for the cheapest brain for the job

Running AI at home burns real money. So I had Claude build a two-layer router: one flag picks the cheapest model that's still good enough, then the cheapest host to serve it. Same prompts, same results, a fraction of the bill.

Read the episode →
20 agents audit my machine while I sleep
EP.22 · OVERNIGHT AI
Posted 2026-08-01

20 agents audit my machine while I sleep

One cron job at 01:00 fans out 20 AI agents that audit every repo on my machine for health and security, then write a single ranked morning report — done before I pour the coffee.

Read the episode →
The next 15 dares, scripted in one night
EP.21 · CLAUDE CHALLENGE
Posted 2026-08-01

The next 15 dares, scripted in one night

Fifteen daily Claude Code dares scripted in one night build — 9 built-in commands + 6 anonymized power-user shortcuts, one caption-first 9:16 spine.

Read the episode →
I fed 946 of my prompts to six AI agents
EP.20 · AUDIT MY AI
Posted 2026-08-01

I fed 946 of my prompts to six AI agents

Six parallel read-only Claude agents read 946 of my own prompts across 351 sessions, found my exact pain points, and handed me a five-pillar redesign. Here's the method, so you can audit your own AI usage.

Read the episode →
Why my Telegram AI beats the official Claude app
EP.19 · TELEGRAM APP
Posted 2026-08-01

Why my Telegram AI beats the official Claude app

Everyone opens the polished Claude app and desktop. I mostly use a Telegram bot and a bare terminal instead — same AI brain, but they sit next to my real files, inbox and scripts, so they can actually DO things. Page tag EP.11; built by the night build.

Read the episode →
I let Claude build my n8n workflow — live
EP.17 · CLAUDE BUILDS
Posted 2026-07-25

I let Claude build my n8n workflow — live

A client wanted to see how the automation actually works. So I described one real workflow out loud and Claude wired the whole n8n pipeline node by node, then fired it: draft episode to live URL in one click.

Read the episode →
The doctor-vs-pharmacist positioning and the Constraint to KPI to Build method for deciding what's worth building with AI
EP.16 · THE METHOD
Posted 2026-06-25

When AI can build anything, knowing what to build is the only edge

88% of companies use AI. Only 6% are good at it. 30% of AI projects get abandoned. That gap is the entire opportunity. Here is the consulting pitch, the doctor-vs-pharmacist positioning, the Constraint→KPI→Build method, and the 60-second Matrix-operator video script.

Read the episode →
A push-to-talk AI app installed on a phone home screen, talking back out loud and reaching every project on the desktop
EP.15 · VOICE AI
Posted 2026-07-05

A voice AI on my phone that knows my whole life

Siri can't reach my files and the Claude app has no desktop voice. So I built a push-to-talk PWA that installs to my phone's home screen, routes each question across two AI tiers, talks back out loud, and already knows every project I'm running.

Read the episode →
Claude Code sessions replayed and scored for novelty, feeding a pre-qualified episode backlog
EP.14 · CHAT HISTORY
Posted 2026-07-11

I mine my Claude Code chat history for episodes

The sharpest ideas happen mid-session and vanish when you close the tab. I taught an AI to read my saved sessions back, score each chunk for novelty and surprise, and hand me a pre-qualified backlog — each idea carrying its own source transcript.

Read the episode →
Jacob dropped into a Matrix-style scene as the operator who downloads AI skills on command
EP.13 · MAKE MOVIES
Posted 2026-06-25

Now a computer geek can make movies

I'm not a filmmaker. But I built a skill that gave Claude the script, Seedance 2.0 the shots, and CapCut the edit — 18 seconds, two generations, zero camera. The 6-frame gallery shows the exact prompt behind each shot.

Read the episode →
Today's Claude Challenge — a grid of 21 daily Claude Code dares across two tiers
EP.12 · DAILY DARE
Posted 2026-06-25

Today's Claude Challenge — dare to accept?

A faceless daily series: one small Claude Code move a day. 21 dares in two tiers — built-in commands and skills I built myself — each one you can try in five minutes.

See all 21 →
A Claude Code terminal acting as a media factory: typed idea flowing through three glowing stages — generate, edit, post — all in one chat
EP.11 · MEDIA FACTORY
Posted 2026-07-26

My AI makes, edits, and posts my videos — I just type the idea

The n8n media agent plus paid poster, rebuilt inside Claude Code. Type a topic: it expands the prompt, fires kie.ai → Sora 2, burns in captions, trims dead air, and queues the Reel in self-hosted Postiz. $0/mo on the posting.

Read the episode →
A terminal status line showing the model, effort, and two live fuel bars for context and session tokens
EP.10 · FUEL GAUGE
Posted 2026-06-21

My terminal now shows how much brain my AI has left

Long sessions die silently as context fills. I had the AI build its own cockpit status line — model, effort, and two live fuel bars — so I know exactly when to /compact.

Read the episode →
A website drawn as a glowing building with one lit doorway labelled llms.txt, an AI agent walking straight in
EP.09 · FRONT DOOR
Posted 2026-06-18

I built my website a front door for AI

Agents scrape a site built for human eyes badly. So I gave mine an llms.txt, a JSON index, a tiny CLI and an MCP server — now any Claude reads the whole site from one command.

Read the episode →
A paused tutorial dissolving into a grid of frames feeding a terminal — learn by not watching
EP.08 · NOT WATCHING
Posted 2026-06-11

Learn things by not watching

A 20-minute tutorial I never played: one slash command cut it into 80 frames, read 565 transcript segments — then three research agents proved its money claims were ~4× inflated.

Read the episode →
The night shift — a fresh AI session building an episode page on its own at one in the morning
EP.07 · NIGHT SHIFT
Posted 2026-06-11

I slept. A fresh AI built the next episode without me.

Half past eight I gave it the idea; it asked six questions. I answered two from the couch and fell asleep. By one in the morning it had read the brief and built this whole page alone — then left a note listing every guess it made.

Read the episode →
The globe overlay — 1,782 LinkedIn contacts and live packet-trace routes streaming over the teal Earth
EP.06 · YOUR WORLD
Posted 2026-06-09

I put my whole network on the planet

The AI-built globe got personal: 1,782 LinkedIn contacts stream onto it as a living layer — thin pins, a terminal at 50 names/sec, and three real packet-trace routes. One toggle fades it all back to the plain globe.

Read the episode →
The live Bite Size AI global activity network globe — glowing teal Earth with node and request counters
EP.05 · THE GLOBE
Posted 2026-06-08

I gave an AI one sentence — it built a planet

A rotating, futuristic Earth for the brand, designed in one chat on claude.ai/design. Every prompt, the palette it guessed from a screenshot, and a live globe you can drag to spin.

Read the episode →
The winning aurora v2 app-store screen style — dark teal poster with a phone mockup
EP.04 · OWN STYLE
Posted 2026-06-08

I ditched the borrowed pencil style — my AI designs app-store screens now

Five art directions fought a style-off over the same screen. The winner became the house style — and sixteen screens later, the look is mine.

Read the episode →
A hand-sketched hub-and-spoke map of Jacob's whole AI operating system, drawn by an image model
EP.03 · THE GLITCH
Posted 2026-06-08

I asked AI to draw my whole system — it leaked secret codes

One prompt mapped my entire AI OS as a sketch — and the model quietly leaked raw colour codes into the labels. The glitch, and the 5-cent fix.

Read the episode →
A hand-drawn-style diagram generated by Claude Code from a single prompt
EP.02 · VISUALS

I taught Claude Code to draw like a human

One skill turns a plain prompt into a hand-sketched diagram — no design tool, ~5 cents a render.

Read the episode →
The winning Calm Lagoon palette on the test sketch — one AI shared across work, partner and kids
EP.01 · THE PALETTE
Posted 2026-06-08

What colour is trust? I asked an AI — it overruled my pick

Six drawn candidate boards in two rounds, a winner called Calm Lagoon — and the moment the AI rejected my favourite shade for accessibility.

Read the episode →
02 Proof of results

Constraints removed, KPIs moved

Each case: one thinking constraint that was blocking progress, the metric that proves it's gone, and the AI system that fixed it.

Case 01Publishing

Episode builds were a manual 2–4 h job blocking every publish

0 h Jacob-time — AI builds unsupervised overnight
Case 02Research

Evaluating a tutorial took 20 minutes of passive watching

20 min → 3 min · found 4× inflated income claims
Case 03Infrastructure

AI agents couldn’t read the site — every automation was blocked

1 command — any AI reads the full site in context
Case 04Observability

Context filling was invisible — sessions died with no warning

0 silent context deaths · live fuel bars in the terminal
Case 05In-house

The team’s AI problems had no owner — colleagues came to me instead

Inbound requests: 0 → team’s default first call for anything AI-shaped
Client workEnterprise

Ten AI systems inside one product company

The client and enterprise cases: what was blocking, what moved, and the system that ran it.

Open the case book →
03 Work with us
Jacob Skogström — AI agent operator, Malmö

I'm Jacob — an AI agent operator for complex product companies

I turn messy, real-world product operations into AI systems that actually run. Everything on this site, I built myself with Claude Code. I'm open to a Product Owner / AI role in the Malmö–Lund region and to select consulting work.

04 Get the weekly build

One new AI build in your inbox, weekly

No noise — just the episode, the prompt, and the code, the moment it ships.

No inbox? On the videos, comment BUILD and the bot DMs you the link.

05 Explore

Where to go next

Steal the code

Every episode's free, open download in one place.

Live

Work with us

Hire the operator behind the builds — role or consulting.

Live

Skool community

Learn the AI-agent-operator playbook alongside others.

Coming soon

Personal AI silos

The product: a private AI assistant per person, now Pocket Operator.

Invite-only beta