A connected assistant prototype.
A local interface and backend brought meeting tools, Google Drive documents, Google Analytics, and Search Console queries into one assistant workflow.
PROJECT STUDY / MARKETING AGENCY
An AI assistant
for agency operations.
Connect what was said, what the data shows, and what needs to happen next. A prototype built around the everyday work of a marketing agency.
- ENGAGEMENT
- Design & prototyping
- FOCUS
- Meetings, knowledge & reporting
- WORKING PRINCIPLE
- Draft, source, human review
THE PROBLEM
The information exists.
The answer takes work.
Agency work lives in many places. A commitment is made on a call. The supporting document sits in a shared folder. Performance data lives in a reporting tool. An account lead has to bring those pieces together before the next conversation.
The project started with a practical question: could an internal assistant help find those commitments, prepare a status draft, and keep the supporting sources within reach?
The aim was to give the team a useful starting point for review, with clear ownership of what gets shared with a client.
SEE THE WORKFLOW
From meeting notes
to a reviewable brief.
Move through the three steps to see how a short meeting excerpt becomes structured commitments and a draft.
PREPARED EXAMPLE / FICTIONAL DATA
THE INPUT
A few lines. Several next steps.
A fictional planning call with two clear commitments and one unresolved dependency.
- 1Account lead
“I’ll send the revised landing-page outline by Friday.”
- 2Strategy lead
“I’ll review that outline on Monday.”
- 3Account lead
“Analytics access is still pending. We don’t have a confirmed access date.”
An illustrative walkthrough using fictional notes and prepared outputs. The project study below describes the prototype work.
WHAT I BUILT
A prototype grounded
in the team’s work.
Reusable meeting workflows.
Instructions for extracting commitments, asking questions across meetings with source references, drafting status reports, and checking transcripts for inconsistencies.
A plan for team adoption.
Client-context templates, a training guide, an adoption plan, and a measurement framework to support the next phase of implementation.
PROJECT STAGE & VALIDATION
Built as a prototype.
Designed for a careful rollout.
The reusable meeting prompts were validated against eight mock meetings. The broader system remained a starter and pilot, with end-to-end orchestration and rollout described as the next phase. Production adoption and business impact are not established by this project study.
WHAT THIS MEANS FOR YOUR TEAM
Start with one
question worth answering.
A useful first scope might be preparing a weekly account brief, finding open commitments, or combining approved documents with performance data. The first step is to choose the question, the sources, and the person who will review the answer.
We can then define a small pilot, test it with representative examples, and decide what needs to be true before your team relies on it.
Explore the Internal AI Assistant engagement Read the agent handoff guide