LIVETeams Business Start your build

Platform brief

You don't get software. You get a staff member.

We build a company its platform, then staff it with an AI agent that actually operates it — and a second agent that keeps writing the code. Humans approve anything consequential.

Read time 6 minutes Status live in production Reference client Aliyah Foundation
01 — The idea

Software is sold as a tool. We sell the operator.

Every small organisation that buys software inherits a second job: running it. Our product removes that job.

What a client buys

A platform built for their operation — bookings, tours, payments, CRM — plus an agent that runs it day to day. They ask for things in plain language; the agent does the work and asks before anything consequential.

What we operate

A fleet of autonomous coding agents that fix bugs, ship features and verify their own work on a schedule — reporting in plain English, and escalating to a human when the call isn't theirs to make.

02 — Commercial model

Three ways in, one operating promise

Every lane ships a working platform with an operator behind it — the difference is how much is bespoke.

Lane 1 · Self-serve

Widget build

from $59/mo

Pick a template, choose widgets, provisioned automatically. Live the same day.

Lane 3 · Bespoke

Custom platform

$5k–25k

Built around their exact operation, then operated under the same agent + approval model. Ops plan from $250/mo.

Add-ons: AI agent seats $49/mo each · SMS packs · extra modules · white-label.

03 — How a client comes on

Signup to running platform

The path a new business takes, and where a human is required.

1

They choose and describe

Marketing site → pick a lane → create an account → land in their own portal with a live build tracker.

2

Their workspace is provisioned

An isolated tenant with its own data boundary is created automatically — their records never sit alongside another client's.

3

The coding agent builds it

It creates the repository, writes the site, deploys it to a live URL, and reports what it did and what it did not verify.

4

They meet their agent

From then on they talk to it: ask for changes, get a confirmation card, approve, done — with every applied change written to an audit trail they can see.

Proven, not theoretical. A restaurant went from intake form to a live, deployed website built entirely by the agent — including the agent catching and fixing its own deployment mistake mid-build.

04 — The two agents

One operates the business. One writes the code.

Keeping these separate is deliberate — they carry different risk.

 Client's operations agentOur coding routines
Who uses itThe client, in their own consoleNobody — runs on a schedule
What it touchesTheir own data onlySource code and deployments
Can it act alone?No — proposes, human confirmsYes for fixes; escalates decisions
OversightConfirmation card + audit logAdversarial review before shipping

The coding fleet runs continuously across every product, fixing reported issues, verifying in a real browser, and writing a plain-English report each cycle. When one of them faces a judgement call — should travellers sign in to book, or book as guests? — it doesn't guess and doesn't stall. It asks.

05 — The loop that compounds

The product extends itself from real use

A client asks for something the agent can't do. That gap becomes work, and then becomes capability.

A client asks their agent for something new.
Has the abilityProposes the change → the client confirms → applied and audited.
Doesn't have itFiles a capability request — it never improvises.
In scope for their planThe coding routine builds it.
New widget or moduleAppears on our dashboard for approval → we quote it → the client pays.
The capability is granted permanently — every future change inside it is included, with no new approval.

Why this matters commercially: paying for a capability widens that client's scope for good. They pay once to open a door; every refinement inside it afterwards is included — which is what makes the relationship feel like staff rather than a support queue.

06 — Governance

Autonomy with a short leash

Automatic

Bug fixes, content and copy, data the client already owns, and anything inside a module they've paid for. The agent acts and reports.

Needs a human

New widgets or modules, payments, permissions, anything crossing client boundaries — and any genuine product decision. These reach a person by SMS and by dashboard, and the answer is applied on the next cycle.

IsolatedEach client's data in its own boundary
AuditedEvery applied change attributable
HonestAgents report what they did not verify

Testing runs under a dedicated identity with no password — access is granted for minutes at a time, logged on every use, and confined to a sandbox. It cannot read a client's real customer records.

07 — Where it stands

Live today, and what's next

CapabilityState
Marketing site, intake and lead pipelineLive verified end to end
Client portal — signup, workspace, build trackerLive
Autonomous coding fleet across every productLive running on schedule
Agent-built client site, repo to live URLProven reference build shipped
Decision escalation to the right human by SMSLive
Client operations agent with confirm-and-auditIn build first client
Quote → pay → capability grantedIn build
Template & widget self-service pickerNext

LIVE Teams — platforms, built and operated. Figures reflect current published pricing in US dollars.