NineTwoThree AI Studio

If you’re waiting around for Google or Microsoft to solve your enterprise’s AI problems, or because “AI is expensive”, let us set you on the right path: Your company's most valuable asset is sitting untapped in your databases. The oil rig is built. The drill is ready. But you haven't flipped the switch. All we heard in the 2010s is “Data is the new oil”...and now it’s finally true. You can use your proprietary data to transform your company with AI. Every enterprise is sitting on a reservoir of AI gold: - Pharma data that could accelerate drug discovery - Clinical trials that hold patterns no human could detect - Hospital records that could transform patient outcomes - Manufacturing insights that could eliminate waste - Supply chain data that predicts disruptions before they occur - Shipping information that optimizes your logistics - Customer histories that reveal what they'll want next And the winners aren’t waiting. Early adopters are seeing ROI in the tens of millions. The question isn't "Can we afford AI?" It's "Can we afford to let our competitors process 100 clinical trials while we do just one?" The market will consolidate soon - and the winners will be the ones who tapped into the oil flowing from their databases.

Documents are the lifeblood of business, but most AI solutions stumble when processing complex, specialized content. For Prisonology, a legal services company that helps reduce prison sentences and security classifications, their experts were spending valuable time manually analyzing court documents to create persuasive scorecards for judges. Our solution? We built an AI system that actually works like their human experts do: - A secure knowledge base of their 20+ years of successful legal motions - A specialized reading module that processes court PDFs through branching questions - A reasoning module that generates accurate severity scores based on document analysis The result is dramatically reduced research time, allowing lawyers to focus billable hours where they matter most – on their clients. If your organization's documents are too complex for off-the-shelf AI, perhaps the answer isn't forcing standard models to work. Sometimes innovation means building systems that truly reflect your experts' workflow.

This project remains the blueprint for adding a proper AI agent to your product. Protect Line is one of the UK’s largest insurance providers. They wanted a state-of-the-art AI agent to sit between customers and the support / sales team. Its constraints are pretty standard: - Talk like a human - Comply with all regulations - Pass high-quality leads to sales We designed the perfect system to accomplish this: - RAG-backed knowledge base the agent could draw from - Custom prompts for 10+ scenarios - Chain-of-thought reasoning to follow the sales script - (MOST IMPORTANT) Know when to pass off the conversation to humans Steal this process if you want to do the same. Obviously, there are more details, and we'd be happy to discuss them. But this is the 80/20 for building an AI agent to act as the first touch for your customers.

How to be AI-first without actually being AI-first: - Buy the .ai domain for your business - Read 17 different AI newsletters every week - Sign up for ChatGPT, Claude, Perplexity, and Gemini - Also sign up for a random startup that is a wrapper for the above 4 LLMs - Never use any of the models besides ChatGPT - Put “AI expert” in your LinkedIn header - Use ChatGPT to write a LinkedIn post or two, then realize it’s really bad at that - Add a Jira card at your job to “Decide on our AI strategy” (forever in TODO state) Congrats, you’re AI-first without actually doing anything! Now, how do you ACTUALLY become AI-first? Understand where AI can meet the needs of tomorrow’s entrepreneurs.

We have a few openings to build AI models for companies starting in April. Here’s why you should jump at the opportunity: We have dozens of clients that have returned their AI investment with us within the first year. - We helped a consultancy firm go from 500k → 1M with a 200k Investment in the AI Model. - We helped reduce maintenance orders by 90% for a massive mechanical shop creating 8 figures of savings. - We created an entire new million-dollar AI product for a startup, and helped hire their entire data science team to run it. Now they are getting Series B interest. Our product managers are MIT / Harvard?/ MBA trained, and they only need 5 weeks to roadmap your AI project. We are 1/10th the cost of *Big Consulting Firm.* (We've competed and won bids for clients against them.) All of our AI projects have been put into production in less than 5 months. It feels to us that these companies are finding a slot machine that they know will hit 7 - 7 - 7 this year. Our process takes them from idea through execution and into production. By the time we are training their team on how the AI model works the company is seeing massive returns. There are only a few slots - are you next? (And if you recommend someone, we offer 5 figure cash payouts for the introduction.)

This is a reminder that OpenAI has a family of tools, not just ChatGPT ??

Our team stopped using ChatGPT overnight, without a single complaint. And we didn't lose a step. And this trend will continue until ChatGPT releases hardware or reverse partners with Microsoft. Yes, OpenAI has built the best model, and it's still the best model we use for all of our clients today under the hood. (In fact, we are turning 6 figure investments into 7 figure returns for dozens of clients.) But ChatGPT, the app, is no longer used by our team. We got rid of it with a single announcement in our general Slack channel. Why? We found something better for our use case. Google released an LLM that works securely inside our Google Workspace, and is 80% as good as ChatGPT. - Finally, we can prompt questions about our clients, securely, in an LLM (we couldn't before.) - Finally, we can copy transcripts from phone calls into Gemini’s AI studio and ask it to summarize what the call was actually about while remaining as secure as an email. - Finally, we can take a product call that I missed, run the transcript into AI studio, and extract the purpose, meaning, and 5 takeaways without worrying about data leakage. We can search through emails faster, find data in our drive from a semantic reference, or search previous proposals to get ideas on the paragraph we are stuck on.

A lot of industry leaders are worried about adopting AI - they're feeling the pains of the cloud shift that took 5x too long. But AI is here to stay. It won't take as long as you think to get on board, and you'll be ahead of competitors (who are starting to look at AI already) Invest in your future!

"Wow, that new AI model performed better in one irrelevant edge case, we'd better switch"

An AI privacy wish list, if you will.