Your operation. An AI engineer inside.

An AI engineer builds the software your company runs on: finance and invoicing, reporting, company memory and the tools your team uses every day. Inside your team, in your accounts, with everything yours to keep.

Scope your first systemSee how SMGP runs
Definition

Embedded AI engineering is an AI engineer placed inside your company who builds, runs and hands over the internal software your business runs on, from finance and invoicing to reporting, company memory and the tools your team uses every day.

In the wider industry, an AI engineer is a software engineer who builds products on top of AI models. Embedded AI engineering points that skill at your own company. The AI engineer works inside your team and turns the routines that eat a manager's week into systems: the invoices, the numbers behind the weekly meeting, the notes from every call, the requests that live in someone's inbox.

It is the sister service to embedded GTM engineering. A GTM engineer points AI at revenue coming in. An AI engineer points AI at how the company runs. Our blog post explains the difference between a GTM engineer and an AI engineer in plain English.

What an AI engineer builds.

Four kinds of internal software. We start with the one that costs you the most today.

Finance and invoicing

Invoices that go out on time, payments matched to the right client, a cash forecast and a monthly P&L, without the spreadsheet.

Reporting

The numbers behind the weekly meeting, computed from your systems of record instead of rebuilt by hand.

Company memory

Every call, promise and decision filed within minutes, so the company remembers what its people said.

Internal tools

Forms, dashboards and AI assistants for the jobs your team repeats every week.

Built and running at SMGP.

We run our own company on this work first. Each system below is live inside SMGP as of September 2026.

The numbers

Money in, money out and the pipeline, computed by the system instead of a spreadsheet.

Finance OS

Bank and Stripe transactions synced into Supabase (4,576 transactions across 24 months), a 13-week cash forecast with three scenarios, monthly P&L, and branded invoicing at invoices.smgp.co.

It replaces a bookkeeper's spreadsheet and GoHighLevel invoicing.

The sales CRM

Opportunities board with a weighted pipeline computed in the database, a unified inbox fed by webhooks, activity log and weekly KPI.

It replaces GoHighLevel.

The memory

Calls, promises and daily decisions, filed where the next conversation can find them.

Call-transcript ingest

Every recorded sales call is transcribed, classified and filed into the company knowledge base within 15 minutes, with the account page updated and the team notified.

It replaces note-taking after calls.

The commitments watchdog

Every promise made in Slack or on a call becomes a tracked row; the watchdog posts what is due, overdue and delegated, and files replies automatically.

It replaces a task list nobody opened.

The daily planner and end-of-day capture

A morning planning prompt and an end-of-day capture posted to Slack every working day; the replies are filed into the company knowledge base and turned into tracked commitments.

It replaces the notebook nobody reopened.

The tools

Small pieces of internal software that give a person back part of every day.

The forms builder

Forms.smgp.co: 13+ live forms, Supabase backend, Slack alert on every submission.

It replaces GoHighLevel forms.

Claude Code skills

28 custom skills that run the company's repeatable jobs (drafting, invoicing, kickoff admin, DNS setup, lead checks, call sync) from a single command.

It replaces checklists.

From a routine to a running system.

Understand

Follow the work.

We map the routine, the people involved and the numbers that matter, then agree on what the first system should change.

Build

Build in your accounts.

We build in your stack, test with real data and agree on where a person makes the decision.

Hand over

Make it yours.

Your team gets the working software, the documentation and a clear way to run it. Then we build the next one.

AI engineering or GTM engineering?

Both are engineers who build with AI. They point it at different problems.

An AI engineer and a GTM engineer, side by side
Criteria GTM engineerAI engineer
Builds The machine that wins customers.The software your company runs on inside.
Points AI at Revenue coming in: the right buyers, the right message, every reply answered.How the company runs: money, reporting, knowledge and daily operations.
Typical first system A pipeline program: the list, the research, the outreach and reply handling.A finance or reporting system that replaces a spreadsheet and a manual routine.
At SMGP Embedded GTM engineering.Embedded AI engineering, this page.

Read the difference between a GTM engineer and an AI engineer

We run our own company on the same systems: 455 automations built, 194 running today.

See how SMGP runs

Before we build.

What is embedded AI engineering?

Embedded AI engineering is an AI engineer placed inside your company who builds, runs and hands over the internal software your business runs on, from finance and invoicing to reporting, company memory and the tools your team uses every day. The AI engineer works in your accounts and your stack, and everything built stays with you.

What is the difference between an AI engineer and a GTM engineer?

A GTM engineer builds the machine that wins customers: the list, the research, the outreach, the replies and the CRM. An AI engineer builds the software the company runs on inside: finance, reporting, company memory and internal tools. Both build with AI. One points it at revenue coming in, the other at how the company runs. The full comparison is in our blog post on the difference between a GTM engineer and an AI engineer.

Is an AI engineer the same as a software developer?

Not quite. A software developer usually builds a product that customers use. An embedded AI engineer builds the systems your own team uses, and puts AI models to work on jobs a person used to do by hand: reading, sorting, summarizing, drafting and checking. We do not train AI models. We build with the best ones available and connect them to your data, your tools and your people.

What does an embedded AI engineer build first?

Usually the routine that costs the most hours or causes the most mistakes. For many companies that is finance: invoices, payments and a cash view that lives in a spreadsheet. For others it is a report someone rebuilds by hand every week, or the knowledge that disappears after every call. We choose the first system with you on a 30-minute strategy call.

Do we keep the software?

Yes. The software runs in accounts you own, the workflows and the code are yours, and the documentation is written as the systems are built rather than after. If we part ways, nothing turns off and nobody has to reverse-engineer anything.

Can you work with the tools we already use?

Yes. We build on n8n, Claude and Supabase and connect to the tools you already run, such as Stripe, Slack, Google Workspace, Microsoft 365 and your CRM. If a tool already does its job, we connect to it rather than replace it. You do not have to buy anything to start.

Do we need a pipeline program first?

No. Many companies start with pipeline and add AI engineering once the pipeline is running, but a company whose bottleneck sits inside the business can start here. We recommend where to start on the strategy call, and the proposal follows the call.

Who does the work?

Lucas Mack builds and runs the systems and owns the strategy, with the SMGP delivery team behind him. You get one point of contact in your own Slack, and you see the work as it is built.

See what an AI engineer would build in your business

Thirty minutes with Lucas. We find the routine that costs you the most and scope the first system to your team. You leave with a plan either way.

Book a strategy call Contact our team