.png)
In recent years, we have seen incredible shifts in the technology landscape – these shifts have opened up a new world of possibilities for what small teams can now accomplish.
As operators that have worked closely in and around these new capabilities, we have published below a set of Operating Principles (OP) that we have developed internally that guide how we operate as a team every day.
OP#1 – With automation, second order effects can have a transformational impact:
When implementing automation, we must consider the second-order effects that result from automating a given workflow. Let’s look at a specific example.
We receive “X” number of inquiries from prospective customers every day. It takes one person about 2-3 hours to respond to these. The impact of automating this is by no means limited to the time saved. Let’s review some of the second-order effects:
- Speed and Quality – Relative to humans, AI is much better equipped to write detailed and thoughtful responses around common questions related to product, pricing or value proposition. AI is faster, responds in real-time, and brings forward responses that are more structured and detailed relative to those written by human counterparts.
- Conversion & Brand Development – As a result of the above second-order effects, we are able to improve conversion and brand development.
In brief, companies and teams that are rapidly automating core workflows can build major advantages versus ones that do not. When evaluating automation, we need to look at second-order effects to prioritise which workflows to automate first.
OP#2 – High-calibre operators need to re-think their roles as innovators:
In the future, high-calibre operators will focus primarily on problem-solving, and not so much on repetitive tasks that need to happen every week. In the foreseeable future, the ability to imagine new and implement new solutions will continue to be a highly valuable skill set.
For high-calibre operators, the core job has shifted to driving innovation – imagine new solutions, implement with urgency, measure the impact, and move into higher fidelity problems. This requires that we move our focus away from doing the daily grind, and toward developing agents that can do the same faster, cheaper and better. Let’s look at a few examples of this:
- Example #1 – Let’s say you spend 30 minutes each week to send a report to keep your team abreast of how things are trending. In the age of AI, you need to spin up an agent that can do that for you with greater precision, consistency and structure. You can also automate the inflow of data into the agent to bring full end-to-end automation, reducing your role to simply reviewing and approving the update before it goes out.
- Example #2 – Let’s say you spend 2-3 hours each week to ask questions from LLMs for education on key aspects of building product features. In the age of AI, you need to train an agent that can ask all the obvious questions, and then summarise everything that you need to know about that topic.
The impact of the above two examples is not limited to the time saved. Instead, it has a huge impact on effectiveness. In the first example, AI will uncover trends and patterns that humans may not catch. In the second example, AI can have a transformational impact on raising the thinking calibre on new product features. In due time, these things add up, and effectiveness translates into better execution.
The core spirit of OP#2 is to rethink your role as one that focuses on developing the capabilities that are required for the Company to innovate. It is a fundamental shift in how you view your role.
OP#3 – The “Founders Office” needs to lead by example:
As a post PMF startup operating in the age of AI, we need to execute with great precision on a broad diversity of topics.
How does our product fit into the broader ecosystem in which we operate? In what areas of our execution do we need to think more proactively? How do we manage the delegation of key priorities and the roll-up of execution on these fronts? At startups, historically, there is limited definition and rigor around some of these core questions and internal workflows.
With LLMs, organisations need to be designed for adaptability, and the Founders Office needs to serve as the driving force behind these adaptations. Each major adaptation is a project in and of itself, and as with most projects, there needs to be a deadline and a clear DRI (Directly Responsible Individual). In large companies, the unit driving adaptations is sometimes referred to as the “AI Transformation” team. At Metal, we refer to this as the Founder’s Office.
OP#4 – Agents needs to be treated as internal products:
On a daily basis, teams across the board need to spin up agents that perform various types of routine work. Each agent is ultimately an internal product that requires architectural thinking. Do we design our agents to be deterministic in nature whereby they produce defined outputs from defined inputs? Or do we design agents for flexibility whereby they can harness the core intelligence of LLMs to self-improve?
In building an agent, we are developing an internal product. And as with any high-quality product, we need to think through the architectural, design and workflow optimisations that are characteristic of great products.
