A Brief History of LLMs at Northwoods
ChatGPT was released on November 30, 2022. Like other professional services firms, it took over a year before LLM technology hit our radar, and in late 2024, we dipped our toes into the water and began experimenting. Our initial impression: hallucinations, unpredictable output, and enough problems to suggest the technology wasn't ready for prime time. But those impressions didn't last. Our curiosity and intuition nagged, and a few began doing bits of productive work. We also become aware that the tools were improving rapidly. Today, AI has our full attention and is transforming how we work. Here are the stages our team went through, and where we're going next.
Stage 1 — Skeptic / Unaware
This was our state in late 2024. AI tools had entered our awareness, but most of my co-workers were either actively dismissive or passively resistant. Their mental models of AI were shaped by preconceived beliefs (largely science fiction-based) or over-hyped media coverage. In hindsight, we were cautious. I didn't push hard for everyone to gain hands-on experience, the kind of immersion that would have generated faster insights. Our shared belief: "This is interesting, and it might help me with my job, but...." followed by a reasonable list of AI's deficiencies with thinking, tasks, and judgements. We were not using paid models at first, but I'm not sure that would have mattered. The AI models of 2024 had many challenges.
Stage 2 — Curious Experimenter
Triggered by professional curiosity, YouTube videos, and a company directive delivered in early 2025, the team at Northwoods began more actively playing with ChatGPT, Copilot, Perplexity, and a few others. We moved from short, Google-like queries to longer prompts, but we limited exploration to low-stakes tasks. Our curiosity was time-constrained, and usage was sporadic. We still had our day jobs, and productivity gains were not yet obvious. The learning often occurred in off-hours and was not connected to active work output. Common belief: "This is interesting....there might be more here than my first impression."
Stage 3 — Tactical User
By mid-2025, about 2/3 of the team at Northwoods identified specific tasks where AI could reliably save them time: drafting emails, summarizing documents, writing code snippets, finding bugs, proofreading, and writing first-pass copy. A number of us began using LLMs regularly, but narrowly, and our prompting was still largely trial-and-error. A common belief emerged: "This saves me time on certain things."
Stage 4 — Intentional Practitioner
Within six months, the more visionary members of our team began developing more complex mental models of how AI could work. They began writing more complex prompts, implementing projects, and building reusable agents. We developed a deeper understanding of context-setting, iterating on outputs, and knowing when not to trust the tool. One of the key skills gained was learning the "personality" of the AI, which we liken to learning the quirks, strengths, and foibles of a new co-worker. By late 2025, four of our team members had woven AI into their daily workflows across multiple task types. They started sharing tips with colleagues. Common belief: "I know how to get good results if I set it up right."
Stage 5 — Workflow Integrator
We are entering this stage now. A healthy segment of our team is beginning to redesign how they work -- not just inserting AI into existing processes, but restructuring tasks around AI's strengths. Some of our backend developers have stopped coding almost entirely, turning their full attention to writing requirements, test plans, validation criteria, and agentic workflow orchestrations. They think in terms of systems: custom instructions, reusable prompts, and automated sequences. The more advanced users have become informal mentors and coaches on AI practices. Common belief: "I'm rethinking the whole process, not just the steps."
Stage 6 — Multiplier / AI-Native
By mid-2026, our team will enter the AI-native practitioner stage. We will be building tools and agents, training others, restructuring our internal workflows, evaluating new platforms, and contributing to organizational AI strategy. I see my job as pushing Northwoods toward a future where AI is not just a tool, but a trained collaborator with multiple skills. As our speed of adoption increases, we will refocus attention on security, governance, verification, and testing -- re-learning how to best integrate human oversight into an ever-increasing agentic workflow. Our shared belief by the end of 2026 will be: "This changes what our team can deliver."
Ready to Integrate AI into your company?
At Northwoods, we support organizations through every stage of AI – from planning to developing tools and agents that solve your business problems.
If you'd like to chat we’re here to help – please reach out!
