DEV Community: Juaria Iqbal

Generic AI Tools Are Rented. When Should Your Business Own Its AI?

Juaria Iqbal — Thu, 08 Oct 2026 11:44:33 +0000

Most AI helpdesk tools now cover the basics. They answer common questions, hand a conversation to a person when they should, and follow the rules you set up in their dashboard. On features alone, renting and owning look closer every year.

The difference shows up somewhere else: on the invoice, in your security review, and on the day you want to leave. The question that matters is who owns the AI, what you pay for, and who controls where your data goes.

What "bring your own AI" means for a business

Usually the phrase describes employees. Microsoft's 2024 Work Trend Index found that 78% of people who use AI at work bring their own tools, ones the company never chose or set up.

For a business, it means owning the AI that speaks for you instead of renting it inside someone else's product. The code, prompts and workflows belong to you. You choose the AI model underneath (the engine that writes the replies, from OpenAI, Anthropic, Google or others), you decide where it runs, and you hold the keys.

Owning your AI means the system, the bill and the keys are yours.

Generic tools are the right place to start

I want to say this plainly before anything else. If your support is mostly common questions and the volume is modest, a generic AI tool is a good choice. It is quick to switch on, cheap to try, and someone else maintains it.

A generic tool is the right answer until one of these four things starts to hurt.

Sign 1: you pay per resolution, not for what you use

Many AI helpdesk tools charge per seat and per resolution, a fee each time the AI closes a conversation. The fee scales with exactly the thing you want more of: the better the AI works, the bigger the bill.

An owned agent is paid for differently. The model provider bills by usage, for the text the AI actually reads and writes, and that bill goes to your own account. Hosting is billed the same way. There is no extra fee per answer on top.

Owning does not make running AI free, but you pay for what you use, at the provider's price, instead of a toll on every answer.

Sign 2: your security review wants more control

Sooner or later someone asks where customer conversations are stored, who can read the logs, and whose account the AI runs under. With a rented tool, the honest answer is usually "theirs", on their terms.

An owned agent can run in your own cloud account, behind your sign-in and access controls, with the model keys in your name and an audit log of what the AI looked up and did. One thing to be clear about: if it calls an external model, that provider still processes the text it is sent, under its own terms. Where that is not acceptable, a self-hosted model is an option worth pricing.

Security is easier to answer for when the system runs in an account you control.

Sign 3: you are stuck with their model

A rented tool runs on whichever model its vendor picked. When a cheaper, faster or more private model appears, you wait for them to switch, or you don't get it.

When you own the system, the model is a setting you can change. You can test a new one on your own conversations and switch when it is better for your business, without rebuilding everything around it.

If a better model comes out tomorrow, the choice to use it should be yours.

Sign 4: you could not leave if you wanted to

Ask your vendor what you take with you if you cancel, and get the answer in writing. Your prompts, your workflows, your conversation history and the tuning your team did over months usually live inside their platform.

If leaving would mean starting from zero, every renewal is already decided for you.

How to decide

Count the signs that apply to you.

None or one: stay with the generic tool. It is doing its job.
Two or more: price what owning it would cost before your next renewal. Compare it with what you pay now, at the volume you expect next year.

Owning your AI does not mean building it alone. MIT's 2025 study of business AI projects, reported by Fortune, found that projects built with specialist partners made it into everyday use more often than purely in-house ones. That matches how we approach our custom AI agent and helpdesk builds: the project-specific system is yours under the engagement agreement, it runs where you choose, and model usage can be billed to your own account.

In my experience, the conversation that matters most is rarely about the AI. It is about who will own it in three years.

Paying more every month for an AI you don't own? BrillMark builds custom AI agents and helpdesks around your own knowledge, systems, rules and channels. Get a free working prototype built on your website and one of your customer journeys.

What Painting Taught Me About Managing Technical Projects

Juaria Iqbal — Tue, 15 Sep 2026 05:36:27 +0000

On attention, patience, and what painting taught me about really listening to people

Custom software work is exciting because nothing about it is fixed. Priorities shift as a project develops. New information reshapes the plan halfway through. A client sees an early version and realizes what they actually need is something a little different from what they first described. It is the kind of work that keeps you sharp, but it also asks you to hold a lot at once, and the instinct is to try to respond to everything at the same time.

On the days that get to be a lot, I close my laptop and open a paint set instead.

I move between mediums and styles rather than settling into one, acrylic, watercolor, heavy texture, abstract work, often circling back to nature, and I like that no two pieces have to be built the same way. I did not start painting to get better at my job, I started because it gave my mind somewhere quiet to go. But a few years into holding both identities, managing technical projects on one side, painter on the other, I have stopped thinking of them as separate rooms in my life.

The pause is the point

A blank canvas does not care how many messages are waiting in your inbox. It asks for exactly one thing at a time: this color, this stroke, this shape.

That constraint takes some getting used to, especially when your day is built around holding several things at once. But getting used to it is worthwhile. Painting builds a kind of attention that a busy day rarely gives you room to practice on its own: sustained, single-pointed focus on one thing, held long enough to actually see it clearly.

This is not really a story about art as stress relief, though it is that too. It is a story about what happens when you take that same quality of attention and point it at a person instead of a canvas.
You can find more of my paintings on Instagram at @jbeans.art

Listening versus scanning

I noticed something in myself once I started paying closer attention to how I paint. In a fast-moving client meeting, it is easy to listen in a way that is really just scanning, matching what someone says against a list of solutions you already brought in with you. It is efficient, and most of the time it works well. But every so often it means missing something in what a client is actually trying to say, because I was listening for a familiar shape instead of the one actually in front of me.

Painting nature scenes retrained that instinct, in a good way. You cannot render how light falls across water by remembering how light generally falls across water. You have to look at this water, in this moment, and notice the details that do not match what you expected, an unexpected greenish shadow, the place where a reflection breaks. The whole practice is built around resisting the shortcut of "I already know what this looks like."

Client conversations reward the same discipline. When someone says a first draft "is not quite it" yet, that is not vague, it is a specific signal, and it usually holds the real requirement if you are willing to sit with it a little longer instead of reaching for the nearest familiar fix. The pause I have practiced hundreds of times in front of a canvas is the same pause that, in a real conversation, lets the actual need surface before I jump to solving it.

Sitting with ambiguity

Custom software work almost always starts without a fixed reference. A client knows the business problem, not yet the exact solution, and the right shape of it sharpens through iteration and conversation rather than arriving fully formed on day one. That is part of what makes the work interesting, though it can also feel unsettled if you are expecting a project to behave like a copy of something you have already built.

Abstract and textured painting trains you to be comfortable with exactly that kind of unsettled feeling. You rarely start with a finished image in mind. You lay down a layer, see what it suggests, respond to it, and let the piece take shape through that back and forth instead of through a plan decided in advance. None of it is careless, it just trusts that the shape will emerge from the process instead of arriving before it.

That is also, in practice, how the best custom solutions get built. Treating early ambiguity as useful information, rather than a gap that needs to be closed immediately, is what an abstract painter does with a first layer too: not a lack of a fixed answer yet, but material for the layer that comes next.

Layers take time, and rushing them shows

Watercolor and glazed acrylic work in thin layers, each one needing time to dry before the next goes on. Rush it, and the piece turns muddy instead of faster.

Good technical work rewards the same patience. Understanding a stakeholder's constraints before jumping to a build takes time, but it is what makes the work hold up. The projects I am proudest of were not the fastest, they were the ones where I let each layer set before adding the next.

The practice, not just the metaphor

None of this is really about art as a metaphor for work. It is about attention as a skill you can train, and painting happening to be where I train it. If you do not paint, the equivalent might be something else entirely, running, cooking, an instrument, anything that asks for your full attention on one thing and gives you nowhere to hide from it.

What I would actually suggest, if any of this is useful, is smaller than "take up a hobby." Find one practice that asks for single-focus attention on something with no urgency attached to it, and protect that time the way you would protect anything else that matters. Not because it makes you more productive, though it might. But because the muscle you build there, the ability to really look at one thing before deciding what it is, is the same one that makes you someone people trust with their hardest, least well-defined problems.

I still open the paint set on the busy days. It has never once solved the problem waiting for me afterward. But it makes me someone who can actually hear what that problem is, instead of the version of it I assumed going in.