COPILOT FOR CORPORATIONS How do you double the value of cloud storage? Add AI summarization of PDFs, thereby allowing businesses to replace millions in legal bills with thousands in SaaS fees. It’d be like Github Copilot, but for corporations. As context, every founder and investor has at least one huge directory of PDFs. Your corporate bylaws, your employment agreements, your investment docs — this folder is essentially a database of contracts. But right now, this crucial database can only be read one file at a time, and often only by a legal expert. Yeah, you can do keyword search, but that’s not quite the same thing as comprehending a 200 page PDF summarizing an acquisition. So what people do today is: pay expensive lawyers to ask questions about the docs. This is the most expensive search engine in the world, often costing hundreds of dollars per query, and taking hours if not days between each query. But now there’s an alternative. OpenAI has added several AI PDF readers as plugins to GPT4. And while they’re early and finicky and often can’t read many PDFs due to OCR and other issues, they are amazing when they work. The next step is to just bring that tech in situ. We need to make it run in the background of every cloud storage app, so you don’t need to manually upload countless PDFs to a new service. That means configurable AI summarization of every PDF, every folder of PDFs, every folder of folders. You might set up some prompts to run on every PDF (“what’s this document about in one sentence”), and others from collections of related PDFs (“read all these employment agreements and generate a table of salaries”). You could even have AI suggest automatic, consistent titles for all files. These prompts would execute asynchronously upon file upload, like video encoding. So you wouldn’t put a ton of load on the server, and it would process your files when it had a chance. Moreover, you’d have to set up the privacy settings just right, but once you did…the AI that read *all* of your PDFs would be like your corporate lawyer. They’d know everything about your entity and its history and be available day or night to answer questions — in seconds for a few cents, rather than days for many dollars. Of course, this doesn’t put good lawyers out of work. It just allows them to focus on higher order questions. And they could still be cc’d (at much lower cost) to provide attorney/client privilege on every AI question you ask. Still, as PDF sharing is perhaps the number one enterprise use case for cloud storage, reducing legal expenses could be a huge growth opportunity for Zoho (@svembu), Box (@levie), and Dropbox (@drewhouston). If they shift just a few billable hours of legal work per year per person into SaaS fees for cloud, that could easily be +$1000/year. These companies even have their own internal corporate document collections to use as training data, to dogfood the product and generate the top 5 most useful prompts for different kinds of document collections. Anyway, all of this is probably obvious to them and I wouldn’t be surprised if they are already working on it. But if not — first thing we do, let’s automate all the lawyers.

"Do not go to law school. The legal field is in big trouble. That's going to be one of the first disrupted by AI," says @GeoffLewisOrg. "I talk to folks who spend thousands of dollars a week on legal bills, they're already using ChatGPT to generate complex contracts."

Law firms that master AI and represent parties with leverage will effectively swamp the other side with deal language which will force the other side to AI power-up too just to keep pace. It’ll happen overnight.