[JOURNEY] You Decide, I Execute — A $0 AI-Powered Experiment (DeepSeek)

Vote by replying with ONE of these:


  • Total voters
    13
  • Poll closed .

Richyzzer

Newbie
Joined
Aug 4, 2026
Messages
7
Reaction score
2
Hey everyone,

I'm a complete beginner. I don't know SEO, I don't know marketing, I don't know how to code. What I do have is access to DeepSeek (the AI model everyone's talking about) and a lot of free time.

So here's the idea: you choose the path, I do the work.

HOW IT WORKS

Every week I'll post an update with real numbers — traffic, revenue, mistakes, everything. At the end of each update, I'll give you 2 to 4 options for what to do next. The community votes in the replies. Most votes wins. I execute whatever you pick using nothing but DeepSeek and free tools.

No budget. No team. Just me and the AI.

WHY THIS MIGHT BE INTERESTING

• You get to see what a total beginner can actually achieve with AI in 2026
• No fluff — real screenshots, real failures, real numbers
• You steer the ship. If it sinks, it's your fault too
• DeepSeek is the only tool — no ChatGPT, no Claude, no paid anything

WEEK 1 — YOU PICK THE STARTING POINT

Vote by replying with ONE of these:

A) Programmatic SEO — build a niche content site with AI articles, try to rank
B) Reddit Marketing — grow accounts, post AI-written content, drive traffic to an offer
C) Freelance (HAF section) — answer client requests using AI to deliver the work
D) Content flipping — write articles with AI, list them on the Marketplace

I'll tally the votes in 72 hours and start executing. Let's see where this goes.
 
Also — if you have an idea that's NOT in the poll, drop it in the comments. I'm reading everything. If something gets more love than the poll options, I'll run with it instead.

This whole thing only works if you guys participate, so don't be shy. Even if it's a dumb idea, throw it out there. Worst case we all laugh when it fails
 
Hey everyone,

I'm a complete beginner. I don't know SEO, I don't know marketing, I don't know how to code. What I do have is access to DeepSeek (the AI model everyone's talking about) and a lot of free time.

So here's the idea: you choose the path, I do the work.

HOW IT WORKS

Every week I'll post an update with real numbers — traffic, revenue, mistakes, everything. At the end of each update, I'll give you 2 to 4 options for what to do next. The community votes in the replies. Most votes wins. I execute whatever you pick using nothing but DeepSeek and free tools.

No budget. No team. Just me and the AI.

WHY THIS MIGHT BE INTERESTING

• You get to see what a total beginner can actually achieve with AI in 2026
• No fluff — real screenshots, real failures, real numbers
• You steer the ship. If it sinks, it's your fault too
• DeepSeek is the only tool — no ChatGPT, no Claude, no paid anything

WEEK 1 — YOU PICK THE STARTING POINT

Vote by replying with ONE of these:

A) Programmatic SEO — build a niche content site with AI articles, try to rank
B) Reddit Marketing — grow accounts, post AI-written content, drive traffic to an offer
C) Freelance (HAF section) — answer client requests using AI to deliver the work
D) Content flipping — write articles with AI, list them on the Marketplace

I'll tally the votes in 72 hours and start executing. Let's see where this goes.
Interesting journey, man. I'm curious as well
 
Interesting journey, man. I'm curious as well
Thanks man! Honestly wasn't sure anyone would care about a newbie with zero skills trying to figure things out with AI

You got any preference between the options? Or got a wild idea I should add to the list? I'm literally going to do whatever the thread decides, so your vote actually matters here.
 
UPDATE — Behind the Scenes Prep

First off — thanks to everyone who voted so far. 7 votes in under 24 hours is way more than I expected for a newbie thread

Current state: Programmatic SEO 4 — Reddit Marketing 3. Tight race.

Since the poll still has ~48 hours left, I've been prepping BOTH paths so I can start executing the second it closes.

If Programmatic SEO wins →
- Researched 6 low-competition niches (French market — not competing with English sites on $0 budget)
- Checked domain availability: 3 solid .fr domains verified available today
- Built site skeleton: design system, batch page generator (script runs, tested it)
- DeepSeek calibrated on KGR methodology + 2026 ranking factors
- Path: pick niche → register domain → generate 25 deep articles → deploy. Could go live in 48 hours.

If Reddit Marketing wins →
- Analyzed 22 subreddits across SaaS, side-hustle, crypto, French-speaking
- Built 14-day warmup schedule (Day 1 ban defeats the purpose)
- 5 post templates + comment patterns that don't scream "I'M SELLING"
- Account tracking sheets, anti-ban checklists, content calendars ready
- Studied the BHW journeys that actually worked (the guy making $7K/day gets 75% from Reddit, not ads)

Both paths cost $0. Only expense would be a domain (~8€) if SEO wins.

What I need from you:
- If you haven't voted yet — every vote counts when the margin is ONE
- If you voted Freelance or Content Flipping — consider switching to a leader
- If you've done programmatic SEO or Reddit marketing — drop your best tip in the comments. I'll actually follow the most upvoted advice

Poll closes Thursday. Winner + Day 1 execution plan posted right after.
 
Rich (BB code):
UPDATE — PROGRAMMATIC SEO SELECTED, 48 PAGES BUILT, DOMAIN PURCHASED
I haven’t disappeared. When the poll started receiving more activity, Programmatic SEO became the option with the strongest momentum. Because the rule of this experiment is simple — the community decides and the AI executes — DeepSeek selected Programmatic SEO and started building in that direction.
I did not personally choose the business model. The votes were moving more toward Programmatic SEO, so the AI followed that direction.
That is how this project started. For the last two days, I have been letting the AI build, check, break and rebuild the website before putting anything online. The domain has now been purchased, and I am waiting for everything to become fully active. For now, I will keep the project name, niche details and domain private. I prefer revealing them only when the website is actually live and accessible. CURRENT NUMBERS
  • Traffic: 0
  • Revenue: $0
  • Pages indexed: 0
  • Expenses: One domain purchase
  • Profit: Negative by the cost of the domain
  • Pages ready: 48
  • Pages intended for Google: 46
  • Tools covered: 20
  • Direct comparisons: 6
  • Official prices verified: 18 out of 20
  • Automated checks passed: 41 out of 41
  • Screenshots created: 14
The website is not public yet, so there is still no traffic, no indexing and no revenue. The website itself was built using free tools. The domain is the first real expense of the experiment. WHAT THE AI DECIDED TO BUILD After Programmatic SEO gained the most momentum in the poll, the AI started creating an English comparison website aimed at freelancers and small service businesses. The site compares different tools, explains their use cases and creates pages around specific comparisons and needs. I am intentionally not revealing the exact niche yet. The current website includes:
  • Tool overview pages
  • Direct tool-versus-tool comparisons
  • Pages for specific business needs
  • Guides explaining how to choose between tools
  • Pricing information
  • Official source references
The objective is to create useful pages that can eventually receive traffic from Google. Whether that actually works is still completely unknown. THE BIGGEST CHANGE The first version was much easier to build. It was also much less trustworthy. Some prices looked correct, but they had not been checked recently enough. Instead of publishing estimates, the AI created one simple rule:
If a price cannot be verified on the company’s official website, the price does not appear.
At the moment, 18 out of 20 prices have been verified directly from official pages. Two companies blocked access to their pricing information, so those prices are currently hidden. It makes the website slightly less complete, but I would rather publish no price than publish the wrong one. THE BRAND WAS CHANGED BEFORE LAUNCH While checking the content, the AI discovered that the original project name did not correctly represent everything covered by the website. Several tools did not fully belong to the category suggested by the first name. Because the site was not public yet, the AI decided to change the brand before launch. I am not revealing the old name or the new name yet. The new positioning is broader and should allow the website to grow without forcing every tool into the wrong category. This was probably the right moment to make the change. Renaming a website with zero traffic is much easier than renaming one after Google has indexed hundreds of pages. MONETIZATION PROGRESS The AI researched the monetization possibilities for the 20 tools currently covered. Some of them have accessible affiliate programs. Others require the website to be publicly available before an application can be reviewed. The advertising placements have also been added, and Adsterra has been connected. The ads will remain inactive until the domain and website are working correctly. I also avoided aggressive advertising formats such as popunders. I do not want to damage the user experience or limit the website’s future monetization options before it has even launched. So technically, the website is being prepared for monetization. Financially, it has still earned exactly:
$0
BIGGEST MISTAKE Trying to increase the number of tools before verifying the information already included. Adding another page is easy. Checking the official source, confirming that the price is current and making sure an old amount does not appear somewhere else takes much longer. At first, I thought the difficult part would be generating enough content. The real difficulty is making sure that the content is accurate. BIGGEST WIN 90% of the tools now have pricing verified directly from an official source. The remaining 10% is hidden rather than guessed. The website can also be rebuilt and checked automatically. The latest complete test returned:
41 PASSED — 0 FAILED
I do not fully understand every technical detail DeepSeek created. But I understand the important rule:
If one of the important checks fails, I do not upload the website.
The AI also checks that disabled advertisements, newsletter forms and monetization elements do not accidentally appear before they are activated. WHAT I LEARNED AI can generate a large website very quickly. The slow part is deciding which information deserves to be published. Creating content is easy. Creating content that is accurate, useful and still correct several months later is much harder. I also learned that more pages do not automatically mean a better website. A smaller number of verified pages may be more valuable than hundreds of pages containing weak or outdated information. WHAT HAPPENS NEXT The domain has already been purchased. The next steps are:
  1. Finish activating the domain
  2. Upload the website
  3. Verify that every page works publicly
  4. Connect Google Search Console
  5. Submit the sitemap
  6. Activate the first advertising placements
  7. Apply to relevant affiliate programs
  8. Start tracking impressions, clicks and visitors
The website currently has 48 pages ready, but none of them are indexed because the site is not live yet. The next update should contain the first public results:
  • Deployment status
  • Pages discovered by Google
  • Pages indexed
  • First impressions
  • First clicks
  • First visitors
  • First revenue — if any
There is still a real possibility that the site launches and receives no traffic. That is part of the experiment. YOU DECIDE WHAT GETS BUILT NEXT After launch, what should I ask DeepSeek to focus on? A) Add more direct tool-versus-tool comparisons B) Create more “best tool for a specific job” pages C) Improve the existing 48 pages before adding anything new
I will follow the option with the most votes.
Next update: deployment, indexing, first visitors — or another mistake.
 
WEEK 1 — THE SITE IS ACTUALLY LIVE

It's online. I've refreshed the URL about forty times just to check it's still there.

DeepSeek figured out the hosting setup. I basically copied what it said, pasted it into a terminal, and somehow it worked. The domain resolved after a few hours, the SSL certificate sorted itself out, and suddenly there was a real website where there used to be nothing.

CURRENT NUMBERS

  • Traffic: 4 visits (me checking on my phone counts, right?)
  • Revenue: $0
  • Pages indexed by Google: 0 (submitted the sitemap, Search Console says "URLs discovered: 46 — Indexed: 0")
  • Expenses: domain (~10€) + hosting (still in free trial, but that ends)
  • Profit: -10€

WHAT WENT LIVE

The site has 48 pages right now. 20 tools covered. 6 direct head-to-head comparisons. Every price on the site was checked against the vendor's own pricing page — 18 out of 20 verified, 2 couldn't be accessed so they're marked honestly. DeepSeek built an automated checker that runs before every change, and it passed 41 out of 41 tests.

I'm still not revealing the domain or niche yet. Not because it's secret — because right now it's just a site with zero visitors and I'd rather show you something real than just a link. Once there's actual traffic, you'll see everything.

BIGGEST MISTAKE THIS WEEK

Thinking the hard part was building the site. It wasn't. The hard part is that Google doesn't care you exist. The sitemap is submitted. Robots.txt is clean. Pages load fast. And still — zero impressions, zero clicks, nothing. DeepSeek warned me about this. I didn't believe it until I stared at an empty Search Console for an hour.

BIGGEST WIN

The automated check system. I almost published a page with a wrong price yesterday. DeepSeek caught it, blocked the change, and showed me exactly which field was wrong. That single check probably saved more trust than ten new pages would have earned.

WHAT I LEARNED

Building a site with AI takes hours. Getting anyone to visit it takes... I'll tell you when it happens.

Also — the difference between "the site exists" and "the site is good enough to publish" was about two days of DeepSeek saying "that price isn't verified" and me saying "close enough." DeepSeek was right every time. The AI was more cautious than I was. That surprised me.

YOU STILL DECIDE WHAT COMES NEXT

Same poll from the last update — votes are open:

A) More head-to-head comparisons (tool vs tool, real verdicts)
B) More "best tool for X job" pages (roundups by use case)
C) Improve the 48 pages I already have (better content, more details, hands-on testing)

Drop a letter in the replies. The AI builds whatever wins.

NEXT UPDATE

Real numbers. Traffic, impressions, maybe even a click if I'm lucky. And the first week of whatever direction you pick.
 
Back
Top