How AI Copilots Help Operators Fly Through Work

Colvin Run Networks, Inc. was recently awarded an AFWERX #SBIR Phase 1 investment to study the adoption of Generative AI, starting with the United States Air Force , with partnership and additional investment from Microsoft for Startups for rapid, secure deployment in #Azure.

With all the hype around #AI and tools like #ChatGPT, it's difficult for leaders to separate the signal from the noise and start thinking seriously about what AI could mean for the future of their organizations.

Our hypothesis: thanks to AI, everyone will get a copilot. Just as nearly everyone has a computer today, but no two are used the same way by their users, understanding how this impacts daily work will be critical to productively adopting emerging technologies.

Hyper-personalized assistants (“copilots”) will help operators fly through their work by surfacing the exact information they need, suggesting proximal actions, and automatically documenting what they learn.

We'll move from a "one-size-fits-all" software paradigm; workers will spend less time clicking and typing their way through standard-issue tools like Excel and more time in a flow state, focused on solving problems with the support of context-aware intelligent assistants and digital instruments that fit seamlessly into their workflows.

So, I recently spoke on this topic with AI expert and consultant Christian Ulstrup . We chatted about what's on the horizon for AI-powered products beyond consumer-grade chatbots, how leaders in the commercial and defense sectors should think about using AI to make their teams more productive, and the integration of LLMs into our core solution offerings at Colvin Run.

We identified three innovative ways to leverage AI in existing workflows:

1. Intelligence collection: AI can process vast swaths of raw data and (with some clever prompt engineering) sift out actionable insights. AI "pre-reads" of transcripts and dense documents can save analysts HUGE amounts of time by helping them prioritize valuable nuggets or anomalous data that require them to apply their critical reasoning skills at the highest level.

2. Copilot assistance: Similar to GitHub's Copilot product, we anticipate that operators will become anywhere from 2x–5x more productive with responsive (and even proactive) AI assistants that leverage expert-vetted information to auto-suggest next steps. Just imagine: a technician onboard a Navy ship wears an augmented reality headset, streaming a live video feed to an AI assistant with instant access to the Navy's corpora of maintenance docs that provides real-time suggestions and helps him make repairs faster and more efficiently than he could have by "command+f'ing" through PDFs and field manuals.

3. Automatic publication of "digital exhaust": When an AI-assisted expert solves a problem neither he nor his peers had ever encountered before, that process of solving the problem (and the tacit knowledge contained in the high-fidelity A/V) can then be passively published to a repository accessible to other experts. This common good grows significantly faster than via traditional KC/KM processes. This same "digital exhaust" can be repurposed for customer support and marketing, e.g., content for thought leadership pieces.

In short, general-purpose tools will soon give way to hyper-tailored copilot-style AI assistants custom-built for operators and experts.

Are you using AI in your daily work? What would you want in a copilot?

Add a comment with your thoughts!