42% of larger consultancies say they've "invested in custom AI solutions." At first glance, that may seem bad for companies like Glossa AI - if you build in-house, you probably won't also buy a dedicated AI tool. But I'm finding the opposite to be true, actually - many of my current customers started out building in-house, found building everything to be quite difficult, and decided to buy at least a few dedicated tools.
The Consultancy Growth Network report that the 42% comes from explains why, too: the #1 blocker to AI returns is risk, quality, and the inability to trust the output. It's not hard to make AI do things, it's hard to make AI do things that you TRUST. Outputs need to be consistent, traceable, and verifiable to be trusted, broadly speaking.
(Note: some things ARE better built in-house! Low-stakes, internal, single-party, reversible things. Like internal knowledge search or internal dashboards.)
