@DeRonin_: why i'd build an AI-native service over SaaS for the next 12-24 months: everyone is shipping a wrapper now... that's th…

X AI KOLs Timeline News

Summary

The author advocates for building AI-native services over traditional SaaS by selling completed work instead of tools, citing companies like Harvey, Sierra, and Fin as successful examples with high ARR growth.

why i'd build an AI-native service over SaaS for the next 12-24 months: everyone is shipping a wrapper now... that's the whole problem you can't build a moat out of something a competitor rebuilds over a weekend or even better, it will get replaced by new frontier model at all selling the finished work is the opposite: > the customer never wanted the tool, they wanted their books closed, their contract checked, their claim filed (saas has been a help tool, but you sell complex solution) > the budget already exists, you're replacing a line item they pay every month instead of inventing a category > every model release makes your delivery cheaper, instead of making your product less necessary > you get paid from job one, no year of guessing at product market fit > the moat is the rulebook, every mistake you caught and wrote down, and nobody can download that > your overhead stops growing with every client once intake is a form and quality is a list here's a few case studies of companies already doing it: 1. http://harvey.ai - $100M to $400M ARR - 13 months - the research and drafting an associate used to do 2. http://sierra.ai - $100M to $200M ARR - 7 months - the support queue, billed at about $1.50 per resolved ticket 3. http://fin.ai - $1M to about $100M ARR - 3 years - first line support at $0.99 per solved conversation none of them sold software to do the job, they sold the job what it gives to you: - you're paid to learn what to build, instead of raising money to guess - one person carries the client load that used to need a department, i replaced four of mine last year - margins land closer to software than to an agency, because your cost of one more unit is nearly zero - you own an asset that isn't a seat license, and it gets more valuable with every job you run everyone is building tools and almost NOBODY is selling the work same 24 months, completely different odds and you don't have to bet the house on it (it will cost to you literally LLMs expenses) keep whatever pays you now, pick one deliverable your clients already ask for, do it by hand for five of them and write down every mistake you catch i want to put the bet of my business here next years
Original Article
View Cached Full Text

Cached at: 09/28/26, 09:56 AM

why i’d build an AI-native service over SaaS for the next 12-24 months:

everyone is shipping a wrapper now… that’s the whole problem

you can’t build a moat out of something a competitor rebuilds over a weekend

or even better, it will get replaced by new frontier model at all

selling the finished work is the opposite:

the customer never wanted the tool, they wanted their books closed, their contract checked, their claim filed (saas has been a help tool, but you sell complex solution) the budget already exists, you’re replacing a line item they pay every month instead of inventing a category every model release makes your delivery cheaper, instead of making your product less necessary you get paid from job one, no year of guessing at product market fit the moat is the rulebook, every mistake you caught and wrote down, and nobody can download that your overhead stops growing with every client once intake is a form and quality is a list

here’s a few case studies of companies already doing it:

  1. http://harvey.ai
  • $100M to $400M ARR
  • 13 months
  • the research and drafting an associate used to do
  1. http://sierra.ai
  • $100M to $200M ARR
  • 7 months
  • the support queue, billed at about $1.50 per resolved ticket
  1. http://fin.ai
  • $1M to about $100M ARR
  • 3 years
  • first line support at $0.99 per solved conversation

none of them sold software to do the job, they sold the job

what it gives to you:

  • you’re paid to learn what to build, instead of raising money to guess
  • one person carries the client load that used to need a department, i replaced four of mine last year
  • margins land closer to software than to an agency, because your cost of one more unit is nearly zero
  • you own an asset that isn’t a seat license, and it gets more valuable with every job you run

everyone is building tools and almost NOBODY is selling the work

same 24 months, completely different odds

and you don’t have to bet the house on it (it will cost to you literally LLMs expenses)

keep whatever pays you now, pick one deliverable your clients already ask for, do it by hand for five of them and write down every mistake you catch

i want to put the bet of my business here next years


Harvey – Professional Class AI

Source: https://www.harvey.ai/

Build a Frontier Legal Organization

The world’s top law firms and in-house legal teams trust Harvey with their highest-stakes work.

Request a Demo

Harvey UI showing Acme Acquisitions space with shared documents, agents, and more

2,400+ Legal Organizations Build on :Harvey:

One Platform for Legal Work

Firms and legal departments do their best work with Harvey. Run everything your team and practice needs on one collaborative platform.

New

Introducing :Harvey: II

The most powerful Harvey yet. Smarter agents can now deliver more complex work in exactly the way you like it.

Product Innovation for the Best Legal Minds

Real Impact for Real Clients

Helping Teams Stay Focused and See Measurable Results

Average hours saved per month

25+

Professionals using Harvey

200,000+

Law firms and in-house legal teams using Harvey

2,400+

Countries Harvey is used in

70+

AmLaw 100 firms using Harvey

75+

Similar Articles

Micro-SaaS Is Dead. Service With A Software Replaces It

Hacker News Top

The author argues that the traditional micro-SaaS model is dying because AI makes it easy for anyone to build simple tools and customers are consolidating spending. He proposes 'Service with a Software' as the new indie builder strategy: building custom, overfit software as an instrument to enhance a service, not as a product to sell.