Build vs buy:
AI agents.
Buying an off-the-shelf AI agent is fast, but it is generic and usually ties you to someone else's tool and roadmap. Building your own takes a little more up front, captures your context, and leaves you owning the system and the skill.
Commodity, or
your edge?
Buy when the task is a commodity that everyone does the same way. There is no point rebuilding a generic transcription or scheduling tool.
Build when your knowledge or process is the edge, because that is exactly what an off-the-shelf tool cannot replicate. An agent grounded in your digital brain answers and acts like your business, not like a template.
Here is the part people get wrong: building sounds slow and risky, but with someone who has done it before you can ship the smallest useful version in days. You can see this in Sentinel, which started from one outcome and grew from there. The win you build stays yours.
What I have built.
The agents worth building are the ones that hold your edge. I built Sentinel with Vantum from their own methodology, and an agent for a client's legal team from their own material. Neither could be bought off a shelf, because the value was context only that business had. The commodity parts you can rent. The parts that are yours, you build and keep.
Common questions.
Is a built agent harder to maintain?
Not when you build it together and understand each part. You own it and can fix or extend it, instead of waiting on a vendor's roadmap.
How long does it take to build one?
Often days, not months, when you start from the one outcome that matters and build the smallest useful version first.
Build the parts
that are yours.
Bring the task to a free session. If buying is smarter, I will tell you. If building wins, we will start.