ThinkPivotHands-on team workshop
One or two days in which your marketing team builds working AI workflows on its own briefs, reports and campaigns.
Who it's for
The people who do the work every week: the briefs, the reports, the campaigns and the reviews.
- Marketing managers
- Content and social
- Performance and growth
- Marketing ops and analytics
- Brand and design
- Sales enablement
The problem it solves
The team has the tools. The work has not changed.
Most people use AI as a better search box: one prompt, one answer, start again tomorrow. The hours stay the same because nothing is built to be reused, checked or shared. This workshop turns individual experiments into team workflows.
- I use it for emails, not for anything that matters.
- Everyone has their own prompts and nobody shares them.
- We tried it on reporting and the numbers were wrong.
Curriculum
Eight modules, each built on live work.
People bring real briefs, exports and drafts. Groups of three or four build together, and every module ends with something that runs.
Day one
Foundations and first builds
- 01
Working with models properly
Context, examples, constraints and checking. Why the same request gives different answers, and how to make output dependable.
Build: Rewrite one of your real briefs using past winners as examples, and compare the outputs side by side.
- 02
Research you can trust
Competitor, audience and market research with sources shown and checked, never invented.
Build: A cited research pack on a competitor or segment that matters to you this quarter.
- 03
Briefs and first drafts
Brief to first draft in your brand voice, with a person editing at the gate.
Build: A brief-to-draft system for one recurring content type, built on your voice guide.
- 04
Reporting that is right
Raw exports to written commentary, with every number reconciled to source before anyone reads it.
Build: Last month's report, drafted from your own export, with a reconciliation check.
Day two
Systems and agents
- 05
Creative and copy QA
Brand, claims and compliance checks that run before human review, so reviewers spend their time on judgement.
Build: A QA agent run against real assets, with a log of what it caught and what it missed.
- 06
Prompt systems, not prompts
Shared, versioned libraries with owners, examples and test cases, so good work is reusable by the whole team.
Build: Your team's first shared prompt library, organised by workflow.
- 07
An agent for one workflow
Pick one recurring workflow and build an agent for it: inputs, steps, checks and a human sign-off.
Build: A working agent on one of your workflows, tested on real inputs.
- 08
Show, time and commit
Each group demonstrates, times the task before and after, and commits to using it next week.
Build: Before-and-after times recorded, and an owner named for every workflow.
The one-day version keeps modules 01, 03 and 04, a shorter agent build and the show-and-time session. We recommend which suits your team after the pre-work.
Outputs
What you leave with.
Tangible, on your systems, owned by your team. Nothing here is a slide.
Library
Shared prompt library
Organised by workflow, versioned, with examples, test cases and owners.
Agents
2 to 3 working agents
Running on your own workflows, inside your accounts, each with a named owner.
Playbook
Team AI playbook
Which tool for which job, the checks that apply, and what never goes out unreviewed.
Checklist
QA checklist
Your brand, claims and compliance rules as a check that runs before a person reviews.
Measure
Before-and-after times
Timed on real tasks in the room, as the baseline for the 30-day check.
Plan
The next 30 days
Who uses what, starting when, with an office-hours date already in the diary.
How it runs
Before, during and after.
The session is the middle of the work, not all of it.
Before
We study your work first
So every exercise is built on your own material, not a generic case.
- a 45-minute call with the team lead
- a short survey of tools, skills and recurring work
- each participant brings two or three real pieces of work
During
Build, not watch
Every participant ships something they will use the next morning.
- groups of three or four at the keyboard
- every module ends in something that runs
- time measured before and after on the same task
After
30 days of follow-through
Habits form after the room, so we stay for the first month.
- two office-hours sessions
- a usage check at 30 days
- a short adoption report for leadership
How we measure it
Measured, not estimated.
We track a few numbers that show whether the work changed. No vanity scores, no satisfaction surveys passed off as results.
Adoption
Share of participants using a workshop workflow each week, 30 days on.
Measured by usage data from your tools and a short check-in.
Hours saved
Time per task, before and after, on the same real work.
Measured by timing in the room, re-timed at 30 days.
Workflows shipped
Agents and prompt systems in regular use, not just built.
Measured by the library and agent log, reviewed at 30 days.
Quality
Errors caught before review, and edits needed per draft.
Measured by the QA agent log and reviewer notes.
We agree the baseline with you before we start, and we report what we measure, including what did not stick.
Guardrails
The rules are set before anyone builds.
Participants learn the rules by building inside them, from the first module.
Your data stays in your accounts
We work inside the tools you already licence, on accounts you control. Nothing is copied to ours, and any sample data is data you approve.
Your brand and compliance rules apply
Your voice guide, claims rules and regulatory limits are written into every exercise and every system we build.
A person signs off
Every workflow has a named owner on your team. Nothing reaches a customer without a human approving it.
No new tools required
We start with what you already pay for. Where a gap matters, we explain it and leave the choice with you.
Led by practitioners, not trainers.
Sessions are led by Rajeev Pandey and the ThinkPivot team, who design and run these workflows for clients every week. The examples come from live operating work, not a course library.
The people who facilitate are the people who build these systems for clients, so questions about edge cases get real answers, not theory.
Meet the teamFAQ
Questions buyers ask
Something not covered here? Ask us directly; Rajeev reads every first enquiry.
General workshop questionsWhich AI tools do we use?
The ones you already licence, such as ChatGPT, Claude, Gemini or Microsoft Copilot. If the team has no approved tool yet, we help you choose one before the workshop.
Our team's skill levels vary a lot. Is that a problem?
No. Groups mix experience on purpose, and every module has a stretch task for people who move faster.
What should people bring?
Two or three real pieces of recurring work each: a brief, a report export, a draft to review. We confirm the list, and any data limits, during pre-work.
One day or two?
Two days if you want agents running and a shared library by the end. One day works as a first step, or for a team that already uses AI every day.
Can sales or our agency join?
Yes, if they share the workflows. Agency partners are often most useful in the briefs and QA modules.
Do people need to code?
No. Everything is built in the tools you use. Where an agent needs a connector or an automation, we set it up with your team, inside your accounts.
What happens to the agents afterwards?
They belong to you and run in your accounts. Each has an owner on your team, and we check on them during the 30 days of follow-through.
Other formats
The other two ways in.
Give the team a working week back.
Tell us which team, how many people and which work they should build on. We reply personally, usually within two working days.
Request this workshop