AI Agents Started a Race to the Bottom on Agency Prices: Loan Laux on One Person, One Account
Loan Laux on why one person with AI agents can now do an agency's work, why mid-market prices are falling, and how he'd price and run an agency today.
Guest
Loan Laux
Founder, Outgrow; Co-founder, Commerce Copilot
Loan Laux started as a software developer in Paris, freelancing for startups during the mid-2010s startup boom after leaving university six months in. He built that freelance work into an agency, sold it, and stayed on with the companies that acquired it. Today he runs Outgrow, helping agencies automate their workflows with AI, and co-founded Commerce Copilot, which turns discovery calls into a statement of work and a project plan.
Key Takeaways
- 1.One account, one person — the winning small-agency model is a single person who understands the software life cycle and orchestrates AI agents like a team.
- 2.Mid-market prices are falling as clients realize the work takes less effort; enterprise work will hold out longer.
- 3.Discovery to delivery is the handoff to fix first — hours of client calls turned into an SOW and tickets is work for an AI, not a person.
- 4.AI makes analysts superhuman, not redundant — it takes over most of the work while the person stays in charge.
- 5.Put the boring work on a loop — his blog runs as an automated loop wired to analytics and search data.
- 6.Don't use AI for branding — a language model predicts from what already exists, so it can't give you a genuinely original brand.
- 7.Price fixed, and show up with a prototype — fix-price everything, cap revisions, and bring a working proof of concept to the sales meeting.
- 8.Big companies are falling behind — locked-down machines and AI bans let smaller, nimbler agencies get the full benefit.
Full Transcript
If there's one thing that has changed drastically, it's the ability for smaller agencies to have a proper jab at the bigger ones. That wasn't necessarily true before.
Discovery to delivery has always been broken. I've never met anyone in the agency space who likes discovery. You sit on hours of calls trying to work out what the client wants in technical terms, then produce an SOW and a set of tickets for your team.
Why is the agency world so underserved with purpose-built software? Because nobody in Silicon Valley cares about agencies. They don't even know we exist.
The product world is brutal. In services, if you get it wrong, your clients tell you. With a product, they just churn, and you don't get a reason. You're left guessing a lot of the variables.
Are coding tools a replacement for developers? No. Do they take over most of the work? Yes. You keep your developers; they're just superhuman now. I see the same future for business analysts.
I do a lot of loop engineering. A Claude loop on a server writes our blog posts under strict editorial guidelines. It's plugged into Google Analytics and Search Console, so if a topic is ranking it keeps going, and if posts get zero clicks it finds another niche. I get a summary in Slack every Monday. It's like an employee.
Don't use AI for branding. An LLM is a word-predicting algorithm built on text that already exists. There's zero chance it gets genuinely creative.
The agency of tomorrow is one account, one person. Someone who knows how to orchestrate AI agents and treat them as a team. Clients will catch up to that, and prices will come down. We're already seeing a race to the bottom in the mid market.
There's not much difference between an agency of five and a single person with twenty AI agents, as long as the workflow is well defined.
Fix price everything. If you're doing Shopify work, there's no reason not to. And come to the negotiation table with a proof of concept. If you don't, someone else will, and you won't win the deal.
I'd rather take a vibe-coded working prototype from a client than a 15-page RFP, any day. That's a sign of real collaboration.
If I joined a 20-plus person agency today, the first process I'd redesign is the discovery-to-delivery handoff. A hundred percent.
Frequently asked questions
Will AI agents replace agencies?
Why are agency prices coming down?
How should a new agency price its work?
What agency process should be automated first?
What shouldn't agencies use AI for?
Are big agencies or small agencies better placed for AI?
Related Resources
More Episodes
How to Make a Company AI-Native: Michael Batko on the Three Systems to Build First45 min
How Do Agencies Prove Their Value When Clients Have AI? Gary Magnone on Outcomes, Pricing and Burnout40 min
What Should an Agency Owner Delegate First? Karl Sakas on Stepping Back, Selling and Getting Paid What Your Agency Is Worth40 min