A climate consultancy found time savings of 18 hours a month per person.
Two in-person training sessions helped three colleagues use Claude Cowork on specific steps in their emissions measurement process, with reusable skills and a practical method for building more.
The firm
The firm is a 14-person climate consultancy helping clients measure and reduce their emissions. Its delivery process runs from agreeing the work and collecting client data through to calculating emissions, preparing reports and reviewing the results with clients.
The problem
The team was already using Claude, but at different starting points. Some work still happened in browser chats; some colleagues had experimented with more advanced features, while others were relatively new to it. They wanted to build the AI literacy and practical capability to automate recurring work themselves.
They already had a map of their emissions measurement process. What they needed was a way to turn individual steps in the process into automations that could work with their own information and tools.
Three areas were prioritised:
Launching a new client engagement, from sales handover through to contract, project set-up and kickoff.
Turning notes from a client kickoff into updates to the client-facing project page and tasks for the internal delivery team.
Producing an emissions report from information already held in the consultancy’s carbon-accounting system.
What we did
01 Interviewed each attendee using the existing process map
Before the training, each attendee completed a short interview in the CairnWorks app. The existing process map gave the interviews a clear structure: which steps they worked on, where their time went, which tools they used and what they had already tried with Claude.
The answers helped shape the training and prepare a personal second brain for each attendee. These were folders of editable files covering the business, their role and colleagues, the tools they used, the mapped process and instructions for how Claude should work with them. Each folder also had places for client work, projects, writing style and examples to be added.
In the first session, the group added detail and scored steps by value, effort and time spent. That made it easier to choose specific automation use cases within the same process.
02 Built the foundations in the first session
The first session covered Claude’s desktop app, connectors, voice prompting, second brains, skills, scheduled tasks, artifacts and browser use. Each topic included a practical exercise using the attendees’ own setup.
They connected their working tools, opened and checked their second brains, and tried reusable skills, scheduled tasks and an email-and-calendar dashboard. This gave the group a shared understanding of how the features worked before applying them to client delivery.
03 Built and tested skills for three steps in the process
In the second session, each attendee took responsibility for building a skill for one step in the emissions measurement process: client launch, kickoff follow-up or emissions reporting. They worked with past client material and their existing tools, with guidance from Michael as they built.
The method was to study previous work, build the skill, test it against a separate example and use the corrections to improve the instructions. The group also worked through access problems, inconsistent inputs and the points where a person needed to review or approve an output.
What changed
18 hours
Monthly time savings per person
Including colleagues beyond the three attendees who could use the skills developed in training. Across the three steps, the team put the total at 89.5 hours a month.
7 hrs → 2 hrs
Client launch, per month
Saving: 5 hours a month. The launch skill used the agreed proposal and sales information to prepare the service agreement and scope of work for review.
10 hrs → 6 hrs
Kickoff follow-up, per month
Saving: 4 hours a month. The kickoff skill used the recorded client call to prepare updates to the client-facing Notion page and suggest tickets for the internal tracker, with human approval before creating them.
96 hrs → 15.5 hrs
Emissions reporting, per month
Saving: 80.5 hours a month across the reporting team. Starting with a downloaded export of a client’s calculated emissions, the reporting skill prepared tables, charts and commentary for a report in Claude Design. The two small-client test runs took around 20 minutes compared with roughly two hours previously.
The monthly figures are the team’s estimates recorded during the sessions.
Alongside the three skills, the group left with personal second brains, connected tools and a shared board of 13 scored steps to work through next. They had practised how to build, test and improve a skill, giving them a method they could keep applying to the rest of the process.
In their words
“We were already using Claude, but the training helped us use its features to automate real work. We came away with automations set up, time savings measured, and the ability to build and improve more ourselves.”
Senior leader
A 14-person climate consultancy