v2026.4 / In market / Shipping weekly
codas-labs / homebase.tsx

Superpowers for people who ship.

We build software on one idea: the work that used to need a whole department should need one motivated person and the right tool. Our products are how that idea reaches the people actually doing the work.

CODASMEDIA Also writing at Codas Media
The Multiplier, a weekly letter on building this way
From the founder

Why we build this way.

Two minutes on what fifteen years of running software companies taught me about where the real cost sits, and what changed when it stopped being true.

Recorded 2026 · Raleigh-Durham, NC
Mockup. Real footage drops in here
Portfolio

Our brands.

Different industries, same premise. Take the work that used to need a team and put it in one person's hands.

Construction Live
The challenge

Every change order lives in somebody's inbox.

One builder runs the schedule, the selections, the change orders and every client conversation, without a back office behind them.

Visit builderpad.com
B2B Data Live
The challenge

Real research takes an analyst you can't afford.

One seller gets the work a whole research team used to produce, at the speed of a database query.

Visit leadtale.com
Creator Live
The challenge

A community only works if somebody reads all of it.

One creator runs a space that would normally need moderators, a summariser, and somebody chasing the follow-ups.

Visit tribetale.com
Research Live
The challenge

The answers arrive. Nobody has time to read them.

One operator gets the themes, the outliers and the follow-up question they should have asked, without hiring a researcher to find them.

Visit surveytale.com
Customer support Live
The challenge

More customers has always meant more people.

One founder runs a support desk that replies in seconds, at three in the morning, in the customer's own language, and hands over the one conversation that genuinely needs a person.

Visit tore.ai
AI memory Live
The challenge

Every AI assistant forgets you the moment you leave.

One shared memory that every agent reads from and writes back to. The missing layer that other companies are raising millions to build on its own.

Visit meshtale.com
What it used to cost

The category leaders in our markets raised $1.92 billion between them.

They are good companies. They were also built in an era when that was genuinely what it took: hundreds of people, a decade, and a lot of somebody else's money. That is no longer the price.

Category Category leader Ours
Built the old way Established leaders, years of hiring, other people’s money.
Construction
project management
$450M raised · founded 2002 · ~4,700 people $0 time to build · team size — you tell me
Surveys
& research
$694M raised · founded 1999 · ~1,300 people $0 time to build · team size — you tell me
B2B data
& lead intelligence
$204M raised · founded 2017 · ~1,000 people $0 time to build · team size — you tell me
Community
for creators
$30M raised · founded 2020 · ~356 people $0 time to build · team size — you tell me
Customer support
& help desks
$545M raised* · founded 2011 · ~1,800 people $0 time to build · team size — you tell me
Across all five $1.92Band roughly 9,200 people $0and a team you could fit round one table
And the new wave Young, well-backed, and each one funded to build a single piece.
AI memory
for agents
$19M raised across two startups ($10M and $9M), each of which does memory and nothing else $0 time to build · team size — you tell me

To be straight about it: plenty of good software gets built without funding. Two of the companies we looked at raised nothing and are doing very well. Taking no money is not the interesting part. The interesting part is the ground a handful of people can now cover, and how little of it needs anyone's permission.

* The customer-support figure includes borrowed money alongside investor capital. The other four are investor capital only.

Thesis / 001

The bottleneck was never the idea.

Fifteen years of running software companies taught this the expensive way. The ideas were never the scarce thing. What cost the money was everything between the idea and the working version of it: the specifications, the handoffs, the coordination, the waiting.

That gap has narrowed enormously, and most software still hasn't been rebuilt around the fact. Codas Labs exists to do that rebuilding, and then to hand the same leverage to the person using the product, not just keep it for ourselves.

The constraint was never imagination. It was everything between the idea and the working thing.

In practice that means our customers feel it as speed: a bug reported on Monday that is fixed on Monday, a feature request that does not join a twelve-month queue, and pricing that reflects what the software actually costs to run.

How we build

We built the system for ourselves first.

An AI-native operating system runs the studio off one shared knowledge base: research, build, review, support. We are not theorising about what that changes. We ship on it every week, and everything below is what it taught us.

01

Value before ask

Deliver something useful before requesting attention, clicks or money. Every piece of work should leave the person on the other end better off.

02

Humble and hungry

Never take the position for granted. Grateful for what has been built, still with something to prove. That energy shows up in the product.

03

Substance over style

The message matters more than the production value. Visual polish never substitutes for a real insight, and we would rather show the work than describe it.

04

Systems, not features

Solve whole workflows rather than shipping another checkbox. Build assets that keep working after the week you built them in.

Behind the scenes

The Multiplier

Every week Mark writes up what actually happened building this way: what worked, what broke, and what it cost. No theory that has not shipped.

Read this week's issue
The Multiplier · Issue 024

The week a test suite ate the machine

One command took 663% of the CPU and killed five running sessions. What that taught me about giving the build its own hardware.

6 min read · Sent Thursday
Founder

Fifteen years of learning where the cost actually sits.

Mark Thompson has been building and running software products since 2011, through the hiring, the outsourcing, the payroll and the coordination tax that came with all three. Codas Labs is what he built once the constraint that justified them stopped binding.

Read the full story
2011
Started building
2013
First SaaS product
2023
First exit
Weekly
Shipping cadence
Work with us

Building something that shouldn't need a department?

Tell us what you are trying to ship. If we are the wrong people, we will say so quickly.

Get in touch