EvoLink

Always Pay Less for Top AI Image, Video & Chat Models

Visit Website
March 20, 2026 Quit my 10-yr Big Tech PM job to build an AI API aggregator.

Hey IH,

I recently made the biggest jump of my career. After 10 years working as a Senior Product Manager at a Chinese Big Tech company, I realized the AI wave is just too big to watch from the sidelines. So, I quit my comfortable job to go all-in.

I built Evolink.ai — a unified API gateway for LLMs. Basically, you use one API key to access Claude, Gemini, GPT, and others, acting as a drop-in replacement for the OpenAI SDK.

The Business Model & The Struggle My core strategy is "low margin, high volume." Because of smart routing, I can offer developers significantly cheaper API costs.

The good news? It’s working. I’ve already acquired a decent initial user base through community hustle and word-of-mouth. The bad news? A low-margin business only survives on massive scale, and my current top-of-funnel just isn't wide enough yet. There is a huge room for growth, but I'm hitting a wall on how to systematically acquire traffic.

As a PM, building the product was the easy part. Distribution is where I need your help. I have three specific questions for the veterans here:

1. SEO for a Developer Tool (API) How do you actually do SEO for a product like this? B2C "AI tool directories" are mostly useless for me since my target audience is developers. Standard blog spam feels wrong. Should I focus on technical tutorials? Programmatic SEO (e.g., comparing API costs dynamically)? What actually moves the needle for dev tools?

2. Designing a Referral / Affiliate Program Do "refer-a-friend" programs work for API infrastructure? If so, how should I design the incentives? Should I give out free API credits (e.g., "Give $5, Get $5") or actual cash commissions? Developers usually hate spamming their friends, so how do I make it feel natural and not scammy?

3. Steady Traffic Scaling & Budget Allocation I have a modest marketing budget. How should I distribute it to steadily open up the traffic floodgates? Should I sponsor dev-focused newsletters? Pay tech YouTubers for integrations? Or ignore paid channels and dump it all into content/SEO?

If you've built dev tools, SaaS, or high-volume platforms before, I’d love to hear your thoughts. Feel free to roast my approach!

12 Comments

  1. 1

    Man, your story is almost identical to mine. Also a Chinese PM background, recently built the exact same thing - an AI API aggregation layer with OpenAI-compatible SDK unified access to Claude/Gemini/GPT/etc. Even the "low margin, high volume" business model matches.

    On your three questions - here's what's actually working for us so far:

    1. SEO for dev tools: Programmatic pages for "API cost comparison" do work - developers do search for "Claude vs GPT pricing". But the biggest driver has been technical tutorials - like "how to build a customer support bot with AI API in 5 minutes". We rank on use cases, not tool keywords. One more thing - open source contributions. PRs to popular AI libraries on GitHub with backlinks in readmes drive surprising traffic.

    2. Referral program: Free API credits work better than cash. Developers don't want to shill for $5, but when it's "give your friend free credits and you get free credits too", they naturally share within their teams. The most successful channel for us has been Slack communities - developers ask "what API layer are you using?" and the word spreads organically.

    3. Paid acquisition / budget: Don't do paid ads initially. Sponsor dev newsletters (e.g. Ben's Bites, The Rundown) - their conversion quality is much higher than Google Ads for us. Our $100 spent on in-product "Build in Public" content outperformed $500 in ad spend.

    Btw - we built almost the exact same thing, called ModelBridge. If you're down to exchange strategies, feel free to DM me on X (@yangyuzhen888). Not here to compete - I think we're facing the same challenges and the space should help each other. Good luck brother! Maybe we should co-write a "two Chinese PMs bet on AI API" post sometime. 😄

  2. 2

    Quitting a 10-year PM role to go all-in is a proper bet. On the referral question - give credits, not cash. Developers hate feeling like they're pushing a product on friends, but "here's $5 off your API bill" feels like sharing something useful. That's the line worth finding. What does your current churn look like on the initial user base you've built?

    1. 1

      Really appreciate you breaking it down so clearly. I’ll definitely keep this in mind when I design the referral mechanics.

      On churn: since the current user base is still relatively small and early, the churn is on the higher side right now. A lot of people are just trying Evolink for a specific experiment or side project, so I see quite a few sign‑ups that spike usage for a short time and then drop off.

      I’m trying to treat this phase more as “learning from behavior” than locking in perfect retention, but I know I need to tighten this up as I grow the top of the funnel.

  3. 2

    I recommend writing technical blogs in this space, how you built it or user trends! a great blog that resonates with your community is amazing.

    1. 3

      Thanks for the suggestion — really appreciate you taking the time to share this. Writing more technical blogs about how I’m building Evolink and the user trends around it is a great idea, and I’ll definitely keep it in mind as I plan my next steps.

  4. 2

    Inspiring story! The AI API space is fascinating.

    What's been your biggest technical challenge

    with the aggregator so far?

    1. 1

      Great question — and thanks for the kind words, really appreciate it.

      The biggest technical challenge so far has actually been pricing and billing across all the different providers and resellers. Everyone has their own pricing model, quotas and quirks, so normalizing that into something consistent and predictable for users while still keeping it accurate is surprisingly complex.

      On the infra side, the aggregator has been running fairly stably so far, but I’m constantly refining how we handle edge cases and provider changes under the hood.

  5. 2

    Nice congrats! I too quit my PM job last june. Now fulltime indie. Pretty cool idea you have. Referrals definitely work, pretty big b2b companies are doing it. Also has the chance to make your product go "viral" if people start sharing it more and more.

    1. 1

      Thanks a lot, really appreciate the encouragement! And huge congrats to you as well for making the jump from PM to full-time indie — feels exciting (and a bit scary) but totally worth it.

      Great to hear your take on referrals too. Knowing that bigger B2B companies are making it work and that it can help things go “viral” gives me more confidence to experiment with it for Evolink.

    2. 1

      This comment was deleted 5 months ago

  6. 1

    This makes sense as a product, especially in the current AI stack.The challenge now is trust and visibility. For developers, nothing beats seeing the product in action fast, clear, no fluff. Showing real integrations and outcomes can unlock a lot more traction than just explaining it. a strong demo can become your best growth asset across multiple channels. I help build those kinds of demos that drive engagement and conversions, so if you want to scale Evolink, I’d be glad to help.

  7. 1

    At $80K/mo on low margins, what happens when the model providers cut prices by 50% again like they did last year? Does your margin compress or does volume spike enough to offset it?

About

I’m fed up with constantly having to tweak my parameters whenever I switch between models from different providers, and I’m not prepared to pay extra for AI services anymore, so I created Evolink.