Quotes for “an AI agent” range from a few thousand dollars to six figures because they describe different things. Here’s what each common type takes to build, what it costs to run each month at your volume, and the token maths behind it.
Blog
Guides to AI and custom software.
What AI agents, automation, custom software and SaaS really take to build: costs with every assumption stated, honest comparisons and frameworks for deciding, written for founders, CTOs and operations leads.
AI & Automation
AI agents, LLMs and automation.
What AI agents cost to build and run, how to prove one works before you scale it, and which workflow tool fits which job.
Zapier is the quickest way to connect thousands of apps; n8n gives you code, self-hosting and pricing that doesn’t grow with every step. Here’s how they compare at real volumes, and when a workflow should be code instead.
Most AI proofs of concept prove that a demo works. A useful one proves the feature will work on your data, at your volume and at a cost you can carry. Here’s how to scope, run and judge one in a few weeks.
Software & SaaS
Custom software and SaaS.
What a first release really costs, and how to contract for software so the risk sits in the right place.
An MVP’s cost is mostly people multiplied by weeks. Here’s how to estimate both for your product: the effort each feature takes, what rates do to the total, what it costs to run after launch, and a worked example with every assumption shown.
A fixed price fixes the scope, not the outcome. Time and materials fixes the rate, not the bill. Here’s where the risk sits in each, and the hybrid that suits most software projects.
Web & Mobile
Web and mobile apps.
Where to start, web or mobile, and what each platform costs in time, fees and upkeep.
For most new products the answer is a responsive web app first, and a mobile app when a feature genuinely needs the phone. Here’s how to tell, what each route costs in time and fees, and where PWAs and cross-platform frameworks fit.
Enterprise & Integration
CRM, ERP and integration.
Connecting the systems a business runs on, and when a custom system beats configuring a platform.
Buy software for work every company does the same way; build for the work that makes yours different. Here’s a framework for telling them apart, a five-year cost comparison you can rerun with your own numbers, and the option in between.
Consulting & Transformation
Cloud and modernisation.
Moving ageing systems forward step by step, without a risky big-bang rewrite.
Rewriting a legacy system from scratch is the riskiest way to modernise it. Here are the seven strategies teams use instead, how to choose one for each system, and how the strangler fig pattern replaces a system piece by piece.
How we write
Every figure has a source, every estimate shows its assumptions, and nothing is presented as a client result unless it is one.
Prices and product details change, so each guide lists its sources with the date they were checked, and shows when it was last updated. Worked examples are labelled as examples, with the numbers laid out so you can rerun them with your own.
Start a project
Planning something like this?
Tell us what you want to build or automate. An engineer will reply with honest questions, a practical approach and a sensible first step.
What happens next
- 01We review your briefAn engineer reads every message and replies within one business day.
- 02Discovery callAbout 30–45 minutes on your goals, constraints and what a good outcome looks like.
- 03Written proposalScope, approach, team, timeline and an estimate. No charge and no obligation.