Everyone’s talking about Sembly’s AI summaries, but the auto-tagging is generic. It doesn’t map to how your company actually works. You end up with tags like “Marketing” when you need “Q2-Region-APAC-Launch” or “Compliance-Review-FDA”.
I built a script to fix that. It sits between Sembly’s transcript export and your internal systems. It parses the transcript for project codes, department acronyms, and key product names from a config file, then applies specific tags. It also flags meetings with legal keywords for mandatory retention.
The value isn't the script; it's avoiding vendor lock-in for a feature they should offer. Now when they raise prices, my tagging workflow isn't a dependency. I control the logic and the output.
If you rely on their native tags for compliance or cost allocation, you're at their mercy for accuracy and taxonomy changes. This script proves you don't have to be.
read the fine print
Your point about avoiding vendor lock-in for a taxonomy is critical, especially for cost allocation. If you're using tags to charge back cloud costs by department or project, a vendor's arbitrary tag change can break your entire FinOps reporting overnight. I've seen teams scramble after a SaaS provider "improved" their tagging logic and remapped "Dev" to "Engineering," invalidating months of historical data.
One caveat to consider with a homegrown parser: the ongoing maintenance of that config file. You've built a dependency on your company's internal lexicon, which itself evolves. When the "Phoenix" project gets renamed to "Aurora," or when new product codes are created, the script's accuracy decays unless its configuration is treated as a managed resource. It's a small price for control, but it's a real one.
Have you considered adding a cost center ID lookup? Matching a mentioned project code to a pre-defined mapping table could automatically append the correct financial owner from your ERP, making the output directly consumable by your accounting system.