Client Overview
A medium-sized design and social media agency, based in Bristol, wanted to move beyond ad-hoc AI use and make AI a safe, consistent, team-wide capability - rather something only a few “power users” could benefit from.
The Challenge
After an initial AI training rollout, one blocker kept showing up - especially among junior content staff:
- Prompting felt like extra work: people wanted to use ChatGPT more, but the time spent writing “good prompts” often cancelled out the benefit.
- No shared knowledge of what worked: teams didn’t know which prompts colleagues were using, so everyone was effectively starting from scratch.
- Inconsistent outcomes: even when AI was used, results varied widely because there wasn’t a consistent starting point for quality, tone, and structure.
The agency needed a way to make good prompting instant, repeatable, and shareable across the whole company.
Our Approach
AgencyTech treated prompting like a workflow asset - something to standardise and distribute the same way you would templates, brand guidelines, or operating procedures.
We focused on three things:
- Reduce friction to near-zero (so using AI feels faster than not using it)
- Make high-quality prompts visible across the business (so knowledge doesn’t stay “in people’s heads”)
- Create a system that scales (so the library improves over time, not just at launch)
The Solution
A custom-built Prompt Library app integrated with Google Workspace and ChatGPT, designed for fast daily use.
Key elements delivered:
- Role/department-based organisation (Creative, Content, Client Success, Finance, Account Management, etc.) so staff can browse what’s relevant instantly
- Search + categories to find the right prompt in seconds
- Launch library of 40 high-impact prompts, curated around real workflows (emails, briefs, reporting insights, captions, proposals, summaries)
- Client-safe tailoring by referencing internal tone-of-voice documents to steer outputs without reinventing prompts every time
- Submission + approval workflow: staff can contribute new prompts, with admin approval to maintain quality and consistency
Net result: AI became a shared capability - not a personal hobby.
The Results
Measured via verified app usage stats and a short internal pulse survey.
- Weekly adoption: 94% of survey respondents reported using the Prompt Library weekly
- Usage validation: app analytics showed consistent weekly active usage across the majority of the 60-person team
- Efficiency gains: users reported reclaiming time primarily through faster first drafts, fewer rewrites, and less “blank page” prompting (aligned to ~4 hours/week per person)
- Quality uplift: respondents reported outputs were more concise, more usable, and closer to “client-ready” on the first pass
- Scalability: contributions increased over time through the submit → approve → publish loop, meaning the library improved as adoption grew
Key Takeaways
We turned AI prompting into a shared, searchable system - making high-quality outputs faster, more consistent, and scalable across the business.
