How to get your company cited by ChatGPT
Most advice on this starts with how to write. That is the third thing that matters. The first two are whether a crawler can reach you, and whether anyone else mentions you.
Notes from the workshop
What we've learned building software with remote teams — speed, quality and everything between.
GEO vs SEO: what actually changes, and what does not
Most of GEO is SEO you should already be doing. The genuinely new parts are few, specific, and worth separating from the noise.
The best AI software development companies, and how to pick one
Every list like this is published by a company that appears on it, including this one. Here is how to read them anyway, and what to ask.
The agentic SDLC: from SRS document to production
One person writes the spec and approves the release. Between those two moments, a chain of agents plans, builds, reviews, tests and ships. Here is the wiring.
Agents on call: DevOps, QA and incidents without a rota
Shipping is the easy half. The expensive half is the three years afterwards, when something breaks at 4am and a human has to be awake for it.
How small can a SaaS team be in 2026?
The honest answer is smaller than it was and larger than the marketing says. Here is which work actually moved, phase by phase.
How much does it cost to build an AI app?
The model call is the cheapest line on the invoice. What costs money is everything required to make its answers trustworthy.
Your development agency isn't working out. What now?
Secure the accounts before you raise the problem. Everything else in this is recoverable; losing access to your own code is not.
Rebuild or refactor? How to decide when software is failing
Everyone wants to rebuild. The existing system is the only honest record of everything your customers turned out to need.
Is AI search optimisation worth paying for?
We sell this service, so here is the uncomfortable part first: for a good number of businesses it is not worth buying yet.
Agency, freelancer or in-house: who should build your MVP?
Most people pick the option they are most comfortable with and justify it afterwards. Here is what each one genuinely costs, and when it is the right answer.
How long does it take to build an MVP?
Three to six weeks for most first versions. The interesting question is not the estimate — it is which five things move it, because four of them are yours.
How to hire an AI engineer, and what it costs
The title covers at least three different jobs, and hiring from the wrong one is the expensive mistake. Here is how to tell which you need.
Offshore, nearshore or local: what actually changes
We are a remote team, so we have an interest here. That is exactly why this sets out the real risks rather than waving them away.
The questions to ask before you sign with a development agency
Most of these are uncomfortable to ask and cheap to answer honestly. That asymmetry is exactly what makes them useful.
What belongs in an MVP, and what to cut
Cutting scope is the highest-return activity in an early build, and almost nobody does enough of it. Here is a test that settles most arguments.
Build vs buy: when writing it yourself is the wrong call
The build case is almost always argued on licence fees versus development hours. Both numbers are the wrong ones.
How to write a spec an agency can actually quote from
You do not need a technical document. You need eight sections that remove the guesses — and a guess in a quote is always priced in your disfavour.
Why is my app slow? Find the cause before you rewrite anything
The rewrite is the most expensive possible response to a problem you have not measured. Here is the order to look in, and what each answer means.
Inheriting a codebase nobody documented
The instinct is to rewrite it. The first two weeks should be spent making it legible instead — and the database will tell you more than the code does.
Prompt injection: the hole in every AI feature you ship
There is no prompt that fixes this. What there is: a set of design decisions that decide how much damage a successful injection can actually do.
What it costs to run a SaaS, after you have built it
Everyone asks what the build costs. The number that decides whether the business works is the one that arrives every month afterwards, forever.
pgvector vs Pinecone: which vector store, and when
Start with the database you already run. The interesting question is not which is better, but what specifically would make you move.
What does an MVP actually cost to build?
Most MVP quotes are wrong for the same reason: they price a feature list instead of a decision. Here are the real ranges and what actually moves them.
Adding AI to a SaaS product that already works
The hard part is not the model. It is choosing which job to give it, and building the parts around it that decide whether anyone trusts the output.
How much does a RAG chatbot cost to build and run?
A demo costs a weekend. A version you can put in front of customers costs considerably more, and almost none of the difference is the model.
How to choose a software development company
Everyone claims senior engineers and agile delivery. Here is how to tell the difference, using things you can check rather than things you are told.
llms.txt: what it is, and whether your site needs one
A one-file proposal for telling language models what your site is about. Cheap to add, genuinely useful — and widely misunderstood.
Should you block GPTBot? Check what your CDN already did
We allowed every AI crawler in robots.txt. Cloudflare was still returning 403 to all of them. Here is how to find out if that is happening to you.
BullMQ vs RabbitMQ vs Kafka: how to actually choose
Most teams asking this question do not have a queue problem yet — they have work happening inside the request. That changes which answer is right.
MVP, prototype or proof of concept — which do you actually need?
Ordering the wrong one is the most expensive mistake in early product work. Each answers a different question — and only one of them is meant to have users.
RAG or fine-tuning: how to choose, and why it is usually RAG
Teams reach for fine-tuning when they want the model to know something. That is the one job fine-tuning is worst at.
Breaking up a monolith without a rewrite
The rewrite is the failure mode, not the plan. Here is how to extract services one seam at a time while the thing stays live.
Why your MVP breaks the moment it works
Success is the failure mode. The shortcuts that got you to launch are exactly the ones that break when launch works.
GEO: getting cited by ChatGPT, Perplexity and AI Overviews
Ranking first is worth less when the answer appears above the results. Being the source that answer cites is the new position one.