AI Action Lab

Orient

The first session established a shared starting point. Rather than surveying the whole AI landscape, it focused on two things: how to prompt well, using a simple three Cs framework, and which tools were worth paying for and which were not. The working guidance was pragmatic: use free services for conversation and thinking, and pay for the tool you are going to delegate real work to.

Participants left with homework rather than notes. Each was asked to set up their own foundational files: plain text or Markdown documents describing their business, their services, their clients and their tone of voice. These files became the backbone of everything that followed. Without that context, AI outputs can easily become generic or miss the mark for a particular business or client.

Apply

The second session moved from principles to practice. Workflows were defined in deliberately unglamorous terms: a repeatable sequence of two to four steps, triggered by a specific event. An enquiry arrives. A meeting ends. A trade show finishes. Participants saw how to connect their AI tool to the systems they already use, principally Google Drive and Gmail, and were encouraged to connect one specific folder rather than an entire drive.

Two workflows were demonstrated end to end. The first was mass follow-up: photographing a stack of business cards, having them transcribed into a contact spreadsheet, and generating personalised draft emails for each contact, informed by the foundational files and by publicly available background on the recipient. The drafts produced for the group during the session were shared with participants afterwards for comment, which turned the demonstration into something they could judge on its merits. The second was proposal writing, which surfaced the real state of play across the cohort: Excel templates and manual assembly, itinerary platforms costing between $120 and $200 a month, InDesign for the final document, and complex itineraries taking up to a week to produce.

The session also covered free tools that earn their place, including Google NotebookLM for turning a curated set of source documents into audio briefings and slide decks.

Embed

The final session was about making it stick. The group worked through three workflows together rather than in breakouts: proposal writing, CRM and mass follow-up. Participants saw how to build a custom GPT or agent for a specific, repeatable job, how to use skills to package a repetitive task into a reusable template, and how to feed an AI tool a small selection of their best previous itineraries rather than everything they have ever produced.

Two themes ran through the day that had little to do with tools. The first was data: several participants found that their real constraint was not AI capability but how their information was organised, with client data spread across spreadsheets, Mailchimp, HubSpot, GoHighLevel and, in one case, an office still working in Excel because the booking system would not connect. The second was change management. One participant was candid that her team had absorbed as much operational change as it could take in the past year, and that a new set of tools was not going to land well on top of it. Successful adoption in a small business is at least as much about sequencing and internal buy-in as it is about the technology, which is why the closing advice was to pick one pain point and one workflow, and get that working before adding anything else.

Governance came up too, particularly for participants handling client data across borders. Knowing which tools information is allowed to go into, and having something written down that says so, is not a large-organisation luxury.

What came out of it

Participants left with foundational files they had built themselves, at least one workflow designed against a real business problem, and a set of tools they had watched being used rather than described. Several were already spending on AI tools and wanted to make that spend work harder. Around $200 a month for team-wide access was described by one participant as straightforwardly cheaper than hiring.

Support did not stop at the third session. Each participant was offered a one-to-one follow-up to work through their own implementation.