Shared agent workflows
Agents your team can work with.
A useful agent setup often starts with one person. Getting the rest of the team to use it means agreeing the inputs, access, review steps and who looks after it. I work with the people doing the job to build that shared workflow and test it on real examples.
AI agent
Talk agent workflows with my AI agent.
Ask how I choose a task, connect the tools, set review points and help your team run the workflow.
Three conversations a day. Two minutes with my AI agent on Google Gemini. No sign-up. How it’s handled
Alongside 100+ client engagements.
A few teams I’ve worked with:
We start with a recurring task the team understands well enough to judge: preparing a report, researching an account, reviewing a release or handling a support queue. Existing examples tell us what a useful result looks like and where a person needs to stay involved.
I turn that process into a brief the agent can follow, connect the tools it needs and make the limits explicit. The workflow records what happened and brings exceptions back to its owner. We test it with the team before deciding which parts can run with less supervision.
Adoption includes the handover. Your team learns to run the workflow, review its output and change the instructions. We compare quality, time spent and usage with the existing process, then decide whether to improve it or take on another task.
What gets built
Work we can take on together.
A repeatable team workflow
Shared inputs, instructions and checks for work that currently depends on one person’s setup.
Agents connected to the work
Access to the relevant tools and data, with permissions limited to the task and approval points for consequential actions.
Review and exception handling
A clear owner, a record of the work and a way to catch incomplete, uncertain or incorrect output.
A team that can maintain it
Working sessions and documentation so the workflow can be used and improved after the engagement.
How I work
From the first decision to handover.
Choose a task worth changing
Map the current process with its owner. Establish representative examples, time spent and what counts as an acceptable result.
Build the first workflow
Write the brief, connect the required tools and define access, approvals, review steps and stopping points.
Test with the people using it
Run ordinary cases and exceptions. Compare the output with the existing process and keep human approval where the work requires it.
Hand over and assess adoption
Document how to run, review and maintain the workflow. Track usage, quality, exceptions and time spent before expanding its scope.
Recent client projects
See the work in use.
Recent client projects with published build accounts. The How I work chapters explain the briefs, reusable instructions, reviews and release checks behind my use of agents.
The work in detail
Read the decisions behind it.
- Project notes and methods
How I organise work with agents
Seven chapters on briefs, shared instructions, review and release checks, with examples from my projects.
Read the account → - Project notes and methods
Agents replacing recurring tools
An account of turning repeated work into agent workflows.
Read the account →
Ways to work together
Choose a scope that fits the work.
A working session, a defined release or ongoing delivery. We agree the responsibilities, dependencies and fee before starting.
- Engagement
Team adoption
Build and test one workflow with the team that will use it, including preparation and documentation.
£2,500 fixed → - Engagement
Delivery sprint
Implement a workflow that needs deeper integrations, operational checks or a production rollout.
From £10,000 per project → - Engagement
Ongoing partnership
Develop adoption across several workflows, reviewing their use and results as the team learns.
From £12,000 a month →
Common questions
Working together.
Scope, responsibilities and what happens after the release.
Contact meWith the workflow that is useful but difficult to share or trust. We examine its inputs, access, instructions and review process, then agree the next improvement. There is no need to start again with a generic introduction to AI.
We agree that per task. Early runs are reviewed, and actions such as publishing, spending or contacting customers keep the approvals the business requires. Any increase in autonomy follows evidence from the workflow and a decision by its owner.
I start with the systems you already use. We check access and integration options during scoping, and agree any new software or usage costs before adding them.
Compare it with the current process using representative work: quality, errors and exceptions, human review time and whether the team actually uses it. Time saved is something to measure, rather than a figure to promise before the workflow is tested.
Discuss your priorities
What needs to change for your team?
Bring the problem, the existing setup and the decisions you have already made. We can work through the next useful step on a free 30-minute call.
Other areas of work