FeaturesSolutionsSecurity & ComplianceTeamBlog
Talk to Sales
Request a Demo

What Did You Mean? Ambiguity Can Be Fun – And AI Can Help When It's Not

Stephen Balzac
January 28, 2025

Ambiguity in wordplay can be the source of a great deal of humor, as in the classic Ronnie Corbett sketch, My Blackberry Is Not Working.  

The sketch takes place in a grocery store with a customer complaining that his blackberry is frozen—the fruit, not the once-popular mobile phone. The puns continue from there, and I won’t spoil it for you (you can watch it at the link above). Ambiguity is what makes it all funny. 

However, that sort of ambiguity is much less enjoyable in a business context. While it’s funny to watch a comedian ask about Apple and get the fruit instead of the computer (or vice-versa), it’s not so much fun when you’re asking a serious question and getting back the wrong answer.  

One of the challenges in finding the right data for any given task is that the computer is not all that good at figuring out what you mean. Even AI systems often struggle to figure out context and thus may not correctly differentiate between the meanings of Apple—and they are not nearly as funny about it as Ronnie Corbett. At least he’s just pretending to be that naive. AI actually is that naïve. It can search based on the meaning of your words, but it struggles to understand the intent of your query. 

And that’s a problem because searching already returns too much useless information. Ambiguity only makes the problem worse, putting the burden on the user to sort through results and figure out what’s worthwhile and what isn’t. 

Fortunately, SWIRL solves this problem. SWIRL determines context, making your AI less naive so that you get the results you want rather than feeling like you’re the straight man in a comedy sketch. SWIRL does this by taking advantage of two techniques: one-shot query rewriting and interactive qualification. With one-shot query rewriting, SWIRL rewrites your query to incorporate such clues as your role, location, or previous queries to determine your intent with the current query. With interactive qualification, SWIRL asks you questions to determine intent and then incorporates that information into the query. 

In the end, you won’t get a classic comedy routine. However, you will get the information you need when you need it because, let’s face it, most of us would rather watch a comedy routine than be trapped inside one.   

Tired of feeling like you’re trapped in a bad comedy routine when you search for information? Contact SWIRL today. 

SWIRL delivers secure, federated AI search across all your systems, re-ranked with AI and kept within your tenant’s security boundary.
Connect Your Systems
Link iManage, NetDocuments, M365, SharePoint, email, research tools, regulatory sources, SQL databases, and other systems. No data lake required. No second index to secure.
Search Everywhere at Once
SWIRL runs a federated search across all connected systems simultaneously. Native permissions ensure lawyers only see documents they're authorized to access.
Re-Rank with Your LLM
Results are re-ranked by your firm's chosen large language model to surface the most relevant items first. Provenance, citations, and source systems preserved.
Feed Your Assistant or UI
Results flow via APIs and connectors to M365 Copilot, ChatGPT, or other assistants. Or lawyers use SWIRL's own legal search interface directly.
Request demo
Ready to See SWIRL in Action?
Schedule a demo with your IT, knowledge management, and practice leaders. See how SWIRL delivers secure, federated AI search across all your firm's systems—in your own tenant, under your control.
Request a Demo
Talk to Sales
SWIRL Corporation 235 Bear Hill Rd | Waltham MA 02451
© Copyright 2026
Terms & Conditions
Privacy Policy