Workflow and AI

Not every repetitive task needs AI. Some of it just needs to stop.

Most marketing teams are carrying work nobody chose. A report that took an hour once and takes four now. A handoff that exists because someone left. A tool three people log into.

I help teams see how the work actually happens and take out what isn’t earning its place. Then I build the automations and AI tools that genuinely save time, and train the team so the thing still gets used a month later.

Show me what the team is doing See what I’ve built

The work about the work

None of this is dramatic. It just quietly takes a day a week off a team that doesn’t have one to spare.

  • The same information gets typed into three places.
  • Reporting takes hours every week and still doesn’t answer the question.
  • Nobody is certain which document is the current one.
  • Approvals add days without adding judgment.
  • Tools were bought that nobody fully uses.
  • People are using AI on their own, with no shared approach.
  • Work moves between marketing and sales by hand.
  • Automation has been layered on top of a process that was already broken.

Most workflows carry work nobody needs, and nobody has been told they can stop.

What the work covers

Roughly in this order, because automating a process you haven’t simplified just makes the wrong thing faster.

  • Workflow improvement

    • Map how the work happens
    • Remove unnecessary steps
    • Improve handoffs
    • Clarify ownership
    • Simplify approvals

  • AI and automation builds

    • Internal AI assistants
    • Content and campaign workflows
    • Research and reporting tools
    • Marketing and sales handoffs
    • Reusable prompt systems

  • Team enablement

    • Practical AI training
    • Workshops by role
    • When to use it, when not to
    • Keeping the brand voice
    • Quality control

  • Adoption and rollout

    • Documentation
    • Testing
    • Rollout with the team
    • Feedback once it’s in use
    • Refining what people actually keep

Fix the process before automating it.

AI helps where the work repeats. Everywhere else it can just add a step.

So before anything gets automated, we go through the same short list:

  • Should this work exist at all?
  • Does it need to happen this often?
  • Is there already a source of truth for this?
  • Is the bottleneck really the tool?
  • Or is ownership just unclear?
  • Will the team actually use this once it’s built?

The test is whether the team still uses it after I leave.

Sometimes the answer needs to be built.

Plenty of people can tell you where AI might help. Fewer will sit down and build the thing, test it against real work, and stay while the team gets used to it.

I go from spotting the opportunity to designing the workflow to building it, and then to the part most of these projects skip.

  • Internal assistants for research, competitor questions, or first drafts.
  • Content and campaign workflows that run monthly without being rebuilt each time.
  • Reporting and analysis that assembles itself before the meeting.
  • Lead routing and marketing-to-sales handoffs that stop being manual.
  • Reusable prompt systems so the team gets the same quality without starting over.

None of that is software you buy. It’s built around how a specific team already works, using the tools they already have.

Then the team learns it: what it’s for, where their own judgment still matters, how to keep it sounding like the company rather than like everyone else’s AI.

The goal isn’t to make everyone use more AI. It’s to help people use it well where it actually improves the work.

How the work goes

  1. See how the work actually happens

    Not the documented process. The one people are running, including the spreadsheet nobody mentions in meetings.

  2. Find the friction

    Where things wait, get re-entered, get chased, or get done twice because two people weren’t sure who owned it.

  3. Simplify first

    Steps come out, ownership gets clearer, and some work stops entirely. This is usually where most of the time comes back.

  4. Build what earns it

    Then connect the systems, build the automation, or build the AI tool for the parts that genuinely repeat.

  5. Enable the team

    Training on their actual work rather than a demo, so people know what it’s for and where their own judgment still applies.

  6. Refine once it’s in use

    What people actually do with it is never quite what you designed. That part is normal, and it’s where the thing becomes usable.

What should be different

Mostly measured in hours the team gets back.

Before
  • Reporting takes hours.
  • Everyone experiments with AI differently.
  • Tools get built and abandoned.
  • The team keeps adding software.
After
  • A repeatable workflow does most of the lifting.
  • Clear use cases and a shared way of working.
  • The workflow is documented, adopted, and part of the process.
  • Existing tools and better workflows do more of the work.

This is probably a fit if…

  • Your team spends too much of the week on work that repeats.
  • People are already experimenting with AI and there’s no shared approach.
  • You want AI training tied to the work your team actually does.
  • You want practical AI built into existing workflows rather than another platform to buy.
  • You need someone who can spot the opportunity and then build the thing.
  • You want automation people will actually adopt.
  • You care about staying authentic and keeping the quality up while using it.
  • You suspect the answer is simpler than more software.
It probably isn’t a fit if…

You want an AI strategy as a headline rather than a change to how the work happens. The useful version of this is unglamorous, and most of the gain comes from the steps that get removed before anything is built.

The engagement

A workflow audit. A process redesign. Mapping where AI actually fits. Building the tools and automations. Workshops and training for the team. Documentation and rollout. Or ongoing improvement as the team and the tools change.

Most engagements are some combination, because a build nobody was trained on and training with nothing built are both half the job. Team training here is a service, not a course. The courses are a separate thing you can buy on their own.

Ask about a workflow review

This comes from doing the work.

I build these for the teams I lead, which is mostly why I know which ones get used and which ones quietly get abandoned.

  • Tools I’ve built

    AI systems for go-to-market

    A content engine that runs monthly planning, assets, email and compliance checks. A messaging agent for on-brand posts and prospect email. A sales assistant for fast competitor and messaging answers.

    See the builds
  • Document technology

    Scan-Optics

    HubSpot CRM and automation implemented from scratch alongside the demand generation rebuild, so the pipeline could be seen as well as grown.

    Read the case study
  • MarTech

    PERQ

    Marketing operations and sales enablement rebuilt through a company-wide restructure, on a budget cut by 34%.

    Read the case study
See more of my work

If that isn’t quite it

Not sure the problem needs AI at all?

Show me what the team is doing today and we can work out whether the answer is automation, a better process, a different tool, or simply less work.

Walk me through it