Northbound Digital

Growth-stage B2B SaaS. Clients in the US and Europe.

An AI operating model built inside your marketing team, then handed over.

Northbound works with growth-stage B2B SaaS marketing teams of three to fifteen people. It maps the work, builds the workflows and the checks inside the tools the team already uses, then hands over the documentation, the decision rights and the training, so the team runs it without Northbound.

1 Map 2 Build 3 Transfer decision rights documented, taught the team runs it 1 Map 2 Build 3 Transfer decision rights documented the team runs it

What you end up with

Work your team can run on an ordinary Tuesday

A small number of workflows the team runs itself, with the checks written down and the approvals named.

Built and running

Three of the systems behind the offer

These were built inside one B2B SaaS marketing team, where they run today. They are the source material for the model, not a record of past Northbound engagements. Status labels have not been rounded up.

AI-first content pipeline

Signed product messaging goes in and channel-ready content comes out. A second model races the lead draft, a third red-teams it, and the drafting model never reviews its own output. A claims register is the only source for numbers, and a person signs off before anything ships.

Live

Agentic advertising pipeline

A three-agent creative stage produces the variants, and a human approval gate sits between that stage and any ad account, so nothing reaches a platform without a person approving it.

In build Not yet live

Capability map

Every activity the team runs, grouped by capability area and mapped as it runs today and as it would run with the model in place, with costs attached. It is the first thing an engagement produces.

Live

map build transfer handover map build transfer handover

How the engagement runs

Three stages, in order

Map, build, transfer.

The order is fixed, because building before mapping produces polished workflows for the wrong work.

  1. Map

    Every activity is listed with the people, tools and hours it takes, then set against how it would run with the model in place. The map also names what the model will not touch, which is where the decision rights start.

  2. Build

    Workflows, instruction files and guardrails go into the team's own stack, built alongside the people who will own them. A workflow nobody in the team has touched is one nobody will keep running.

  3. Transfer

    Documentation, the libraries, the decision rights and the QA routines move across, and training runs until each owner has run their workflow unaided. The engagement is designed to end.

Who you would be working with

Northbound is Robbie Maltby

I built this model because I needed one. Covering a whole acquisition engine alone only became possible once models could do the drafting, and only became safe once the checking was designed in. I run the engagements myself, so the person who maps the work is the one who builds it and teaches it.

Fit and fee

Whether this suits your team

Engagements from

$18,000 a month

A defined scope, priced monthly, running for as long as the map, build and transfer take. The whole fee buys senior time inside your team: there is no junior layer and no bench.

It suits a team that

  • runs marketing for a growth-stage B2B SaaS company, usually three to fifteen people
  • has bought AI tools already and cannot point at work reaching customers through them
  • wants to own the result rather than keep an agency on retainer to operate it

It suits you less well if

  • you run a hundred-person marketing organisation, where procurement, legal review and regional variants make the coordination problem bigger than one operator can solve
  • you want the work run for you indefinitely, which is the opposite of what a transfer is for