How we think about deploying this.
Sessions we have run, written up. Less about what the technology can do and more about what it takes to get it past a security review and into production.
Ship in days, not quarters
How Linkt collapses the build–test–deploy loop so teams move from idea to production in a fraction of the time — without trading away quality or control.
Agents your security team signs off on
Deploying autonomous agents without handing over the keys — scoped permissions, full audit trails, and guardrails that hold by default.
Go-to-market that runs itself
Agents that research accounts, draft outreach, and keep the pipeline moving — so your team spends its time closing, not coordinating.
Three arguments we keep having.
Every session here started as a conversation we had already had four times that month with four different teams. Writing them down was self-defence.
Getting past the review board.
The hardest part of an enterprise AI deployment is rarely the model. It is scoped access, audit trails, and a believable answer to what happens when the system is wrong.
Shipping in days without regretting it.
Speed and rigour are usually posed as a trade. They are not — most of what makes a build slow is rework caused by skipping the parts that made it correct.
Where agents actually earn their keep.
Not everywhere. Agentic systems pay off where the work is high-volume, judgment-heavy and already has a standard someone enforces. Elsewhere they are an expensive way to do a job a script could do.
Written for the person who has to make it work.
There is enough AI commentary aimed at people who will never deploy anything. These are notes for the team that has to.
No predictions.
We write about what we have already put into production, not about where the industry is heading in eighteen months.
The failures are included.
The approaches we abandoned are usually more useful than the ones that worked, because they are the ones you are about to try.
Specific enough to argue with.
If a piece cannot be disagreed with by somebody who knows the domain, it was not saying anything.
About the library.
Are these sessions recorded live?
Most began as a working session or a talk and were written up afterwards, which is why they read like an argument rather than a whitepaper.
Can you run one of these for my team?
Yes — we run them as private working sessions fairly often. It is also the lowest-commitment way to find out whether we are useful to you.
Do you publish a newsletter?
No. When there is something worth saying it appears here. We would rather write four things a year that are worth your time than forty that are not.
Prefer to argue in person?
Most of these positions were sharpened by somebody pushing back on them in a room. Bring the objection — it makes a better first meeting than a pitch.
Book a call