Shared drives may be a thing of the past, given the rapid proliferation of AI reference documents housed by marketing teams.
TL;DR: maintain content approval workflows as a priority.
Will marketing teams start moving their AI working files to GitHub by next year? A consensus seems to suggest so, with shared drives unable to cope with the volume of AI skills, prompts and audit documents.
And given that these files are dependent on one another, any edit to the position document quietly changes the files that rely on it.
With a B2B business cross-referenced across a GitHub repository, every member’s AI assistant reads from the same brand voice and messaging notes, every change gets logged, and wrong edits can be reversed in one step.
GitHub is largely a developer’s tool, built by them, for them. Whether it remains a solution for less tech-savvy marketers is only a projection, and can even be detrimental in some ways to niche sectors like asset management.
So, what does this mean for fund marketers?
Approved language is a staple requirement across all fund marketing documents: strategy descriptions, risk warnings, fees, ESG claims, and prescribed disclaimers. Once a fund gets repositioned, or when regulatory guidance shifts, this language changes.
If AI tools are used to draft RFP answers or commentaries from a library of reference films, every IR, distribution lead and marketer must know which is the approved working version, and when it was used.
This is a record-keeping dilemma not helped by changing shared ‘areas of working’, and concerns the SEC’s books-and-records rule that expected advisers to retain advertisement and supportive material, plus the FCA’s expectation to keep records of promotions and approvals.
Change history shows exactly what an AI was told, by whom, and when, and mimics what a compliance officer would design. The exact opposite is a folder of prompts on somebody’s laptop. Another hurdle is that many firm’s IT security policies restrict tools such as GitHub, while anything holding unpublished performance and client data needs proper access controls.
What this GitHub idea encapsulates is that marketers should follow a design pattern for their pre-approved systems, no matter what they currently use, as follows:
- Create a ‘single source of truth’ library of compliance-approved documentation that AI tools can read from – not copying and pasting it via individual chats.
- Any shared prompt or reference file must have version controls and named authors to trace output back to specific changes.
- Any edits to working in the library needs a sign-off; the same as any factsheet.
- GitHub or SharePoint or any governed content system can all work, so long as versioning is embedded and that IT is involved in selecting the platform.
- Most importantly, never use unpublished internal documentation or customer names in prompt libraries.
The future of shared working through AI tools relies on platforms able to absorb voluminous amounts of data. Fund marketers will favour certain workflows to do so, so long as the solution they choose is compliant and approved at all levels.
Source
The Irreplaceable Marketer, “Your marketing team will be on GitHub within a year” (via TLDR Marketing)
