AI software delivery, in practice.
Notes from our engineering team on shipping software with AI. We write about what works in production: automation patterns, testing with LLMs, delivery process, and the trade-offs behind real projects. No hype, no vendor scoring, just what we learned building and running systems for clients.
What an AI agent costs to build, and to keep running
Staged costs from proof of concept to regulated production, every figure sourced, and the model run rate worked out from published API prices.
Learn more →Why AI agent pilots fail to reach production
Only 11% of agentic use cases reach production. The reason is not model capability. What a pilot never tests, and where our own agent actually broke.
Learn more →DPDP Act compliance: what your product has to do by May 2027
DPDP Act compliance is due by May 2027. Here is what has to change inside your product, and why most posts quote the deadline wrongly.
Learn more →AI-generated code technical debt: what the evidence shows
The two most-quoted statistics about AI code quality do not survive their sources. What 302,600 AI-authored commits actually show, and how to find the debt.
Learn more →A software development RFP template you can actually use
A free software development RFP template you can download and send today, including the AI-usage and code-ownership sections most 2026 templates leave out.
Learn more →Article 50 is in force. Here is what your chatbot must do.
Article 50 applied from 2 August 2026 and the Omnibus did not defer it. What to disclose, the grace period closing 2 December, and the gap in the Code.
Learn more →Finishing what the AI builder started
The app demos beautifully but has no real auth, no data model, and no observability. What production-ready actually requires, and retrofit versus rebuild.
Learn more →What custom software actually costs in 2026
Cost ranges by project class, rate bands by provider type, the maintenance line everyone forgets, and what AI-augmented delivery does and does not change.
Learn more →Questions to ask an AI development company, with our answers
The standard vendor-vetting checklist, plus the 2026 questions incumbents miss, plus our own answers, on the record, including the weak ones.
Learn more →IT budget planning 2027 software and AI teams actually need
A worksheet for 2027 software and AI budgets: build-vs-maintain ratios, the line everyone forgets, and a compliance deadline landing inside the year.
Learn more →A year running a WhatsApp AI agent for lead qualification
We built and run Kavya, a nine-language WhatsApp assistant that qualifies our own leads. Here is the architecture, the guardrails, and what broke.
Learn more →Who owns the code your agency's AI wrote?
Purely AI-generated output may not be copyrightable at all. What that means for your agency contract, and the exact clause wording that decides who owns what.
Learn more →