Bookmarks

If I had to get local SEO clients or I'll be homeless

This is what I'd do (bookmark for later):

1. Target service businesses already spending money - plumbers, HVAC, med spas, lawyers.

Bonus if they're running Google Ads = they understand paying for leads.

Find them by searching "[niche] + [city]" and look for the ad results.

2. Check Angie's List, Thumbtack, HomeAdvisor for active lead buyers.

DM/email them this offer:
"Hey, I noticed you're running ads. I just checked your organic rankings and found some issues costing you leads. Mind if I send over a quick audit?"

20/100 will say "Sure."

3. Use Local Dominator to show them exactly where they rank vs competitors.

Show them their competitor ranking #1 getting 400+ calls per year.

Do the math: 400 calls x 33% close rate x $500 average job = $66,000 in lost revenue.

Ask: "How would an extra $66k change your business?"

4. Send a custom Loom audit showing:

• Their terrible rankings Competitor domination
• Specific fixes needed What success looks like

They'll be shocked no one told them this before.

5. Sign a $1,500/month deal to handle:

• Google Business Profile optimization
• Local citations and directory listings
• Review generation and management
• Weekly GMB posts and updates
• Monthly ranking reports

Get 7 of these clients = $10,500/month

The work? Set up systems in week one, then 80% runs on autopilot.

You're not selling SEO rankings.
You're selling a system that turns their Google presence into a lead machine.

Most agencies charge $50k for what you can deliver for $1,500/month.

They raise venture capital to solve problems they think exist.

You're solving problems you can prove exist with ranking data.

$10k MRR, 85% margins, business worth $600k you own 100%.

Scale to $25k MRR by adding more cities or services.

Still easier than getting a traditional marketing job.

The opportunity is massive - $90 billion SEO market growing 20% yearly.

Most local businesses have terrible SEO and pay for expensive leads instead.

You become the bridge between their lead costs and organic growth.

By the way, we're giving away our 7-figure 50-page proposal template, comment "LOCALRANK" + bookmark this post and we'll DM it to you for free (must be following).

Just caught my client selling MY n8n workflow for $15K.

The same workflow I gave him for $2K.

So fuck it. Here's the exact JSON that's making him rich:

The "Client Thief 3000" automation:
→ Scrapes LinkedIn for CEOs mentioning "inefficient"
→ AI analyzes their company's tech stack
→ Generates a personalized video pitch
→ Books meetings while they sleep
→ Closes deals with 34% conversion

He's closed $73K in 3 weeks. I built it in 2 hours.

The betrayal stings, but your success will heal me.

Follow, RT + Reply "THIEF" & I'll DM the JSON.

(Use it to steal his clients. I'll provide popcorn.)

i work at meta. hr systems. mostly comp processing. quiet job. stable. sometimes i daydream about retiring in portugal.

today a package hit my queue. base + bonus + equity. looks normal at first glance. then i open the details.

$1,000,000,000
over four years.
plus signing.
plus year 1 minimum: $100m.

i stare at it like it’s a typo. check the name. triple check the level. researcher.

coolcoolcool. so now i have to enter this into workday.

i paste the first number, field throws an error.“value must be below $99,999,999.”
lol. okay. i try splitting it. base in one bucket, equity in another. nope. i try scientific notation. it rounds it down.

the system can’t HANDLE a billion dollars.

i call someone on payroll infra. tell them i’ve got a 10-digit comp packet. they think i’m joking. i forward the req. they go quiet.

“we’re gonna need to escalate this,” someone mutters.
“to who?” i ask.
“god, maybe.”

next thing i know zuck’s chief of staff is in the thread. now i’m in a thread with zuck. because of a number.

then i find out the guy declined the offer.
just said no. no negotiation. no counter. just… no.

i’ve been maxing out my 401k for 11 years & this man said no to a billion dollars like he was skipping dessert.

i close the ticket. delete the draft.
go for a walk.

& then i rethink everything.

🎞️ Such an useful upgrade of Google's NotebookLM

NotebookLM just got upgraded with AI Video Summaries and Studio Upgrades.

Google says users can think of Video Overviews as a visual alternative to Audio Overviews.

📹 Video Overviews allow users to turn dense multimedia, such as raw notes, PDFs, and images, into digestible visual presentations.

The video overviews feature enables NotebookLM to produce narrated video summaries, complete with visuals and voiceovers, drawing directly from user-uploaded sources like PDFs, web pages, or notes.

This isn’t just a simple text-to-video conversion; it’s an AI-orchestrated presentation that highlights key insights, timelines, and connections, making it ideal for quick briefings or educational content.

It also lets each notebook hold multiple custom outputs. NotebookLM already offered Audio Overviews, tiny podcasts that read course packs or legal briefs so users could learn hands‑free. Handy in traffic, but charts, formulas, and process diagrams still begged for visuals.

Video Overviews plugs that hole. An AI host builds short narrated slides, pulling images, diagrams, quotes, and raw numbers straight from the upload, then plays the set like a brisk explainer.

Prompts let users set topics, goals, and audience depth.

The updated Studio panel adds 4 tiles and stores many outputs of each type, making multilingual audio, role‑specific video, or chapter‑based mind maps possible in one notebook. It even runs an Audio Overview while you explore other views, trimming tool‑switching friction.

The paper shows inverse reinforcement learning (IRL) lets LLMs learn their own reward signal, so they align better and think deeper.

Most teams still push imitation fine‑tuning and preference PPO, and that path stalls because no shared reward lives underneath.

IRL flips the script.

It watches expert traces or pairwise votes, then learns a neural function that scores any output.

With that score the model can search, plan, or self‑train without new human labels each step.

Math problems highlight the upside.

A simple right‑wrong reward lets agents grow long chains of thought and backtrack until answers lock in.

Chat models gain too, because pair choices are far cheaper than full demonstrations.

The authors map classic RL tools onto this setting and compare routes like PPO, DPO, Monte Carlo policy gradients, and Best‑of‑N filtering.

They warn of reward hacking, show ensemble or generative checks to spot it, and stress fresh data over sheer volume.

The bottom line is clear.

Learn the reward, not just the policy, and the model stays on target longer.

----

Paper – arxiv. org/abs/2507.13158

Paper Title: "Inverse Reinforcement Learning Meets LLM Post-Training: Basics, Advances, and Opportunities"

Szenen eines jüdischen (ausgewanderten) Lebens in Berlin:

► Erstens: Wir bestellten bei Flink. Der Auslieferer sieht die Mesusa (jüdisches Schutzzeichen) an unserer Tür. „Er ruft: Haut ab, ihr Juden!“ Er wirft mir die Kartons an den Kopf und geht. 

► Zweitens: „Unsere kleinen Kinder werden in der Schule regelmäßig von arabischen und iranischen Schülern belästigt und bedroht. Sie beschimpfen meinen Sohn: „Du gehörst hier nicht hin, du bist ein Jude!“

► Drittens: Fast alle Uber-Fahrer fragen mich, wo ich herkomme. Einmal sagte ich: „Israel“. Da hielt der Wagen an und ich sollte aussteigen. 

► Viertens: Wenn wir als Familie unterwegs sind, sprechen wir kein Hebräisch mehr, denn wir wurden deshalb schon angegangen. Die Kinder haben gelernt, ihre Identität in der Öffentlichkeit zu verleugnen. 

► Fünftens: Meine Frau meidet die öffentlichen Verkehrsmittel, aus genau denselben Gründen. 

► Sechstens: Mein Büro wurde mehrfach aufgebrochen, die Wände beschmiert mit „Free Palestine“ und „Juden raus“. 

► Siebtens: Wenn ich einen Antrag im Bezirksamtstelle, wird alles verzögert, sobald klar wird, wer ich bin. 

► Achtens: „In Neukölln sagte mir ein Araber: „Verschwindet lieber jetzt, in 20 Jahren gehört hier alles uns.“

(Link in den Kommentaren.)