On our own business first. In client work after.
Everything on this page runs today, and each item has its entry in the Edge Log.
AI helps write the code. What gets built, and how the pieces fit, is decided by a senior engineer.
Automated tests run on each change, and an engineer reviews it before it reaches production.
Production errors become tasks an agent sizes and fixes. Only small, safe fixes go in without a person. Read the entry →
Our internal testing tool: it reads each screen as text instead of sending screenshots to a large model, so an agent walks a whole app in minutes. 56 screens in 25 minutes on a real app.
Recurring invoices go out on their own unless they would bill below the cost of the work, and hourly projects get a draft on the 1st that a person approves. Every day, checks flag books that disagree with themselves for review; they never rewrite a record.
Old addresses are matched to new ones by rules first; AI suggests the rest, and a person approves each suggestion before it is used. After launch, every redirect is checked again each week for 6 weeks, with what broke or got fixed since the last check.
Answers Spanish legal and tax questions from official sources only: BOE law, all 17 autonomous communities and 41,237 binding tax rulings. Every claim carries its citation. Client access is next.
Our assistant answers and acts across finance, CRM, outreach, banking, legal and research with over 100 tools, and a change waits for a person’s approval, with its impact and a preview. Every morning it reads 5 business signals and keeps a short agenda of 3 to 8 grouped tasks.
Agents read receipts in any format and from any country, match them to bank movements and book them only when every check passes. What needs a person goes to one review queue.
Each production error becomes a task; a coding agent sizes it, fixes it and proposes the change, and only small, safe fixes go in without a person. 11 agent fixes went into our platform on 12 October 2026.
Each AI request now carries only the tools it needs: 8 instead of 19 for a banking question, decided in about 249 ms. In a client’s system 2 days after we measured it, under a week from its release.
Monitoring caught a client site going down and handed it to an AI agent. It was back up within 9 minutes, before anyone on the team had looked at it.
A company’s documentation, data and processes, structured in one place so AI can use them for personalised outputs: answers, quotes, follow-ups. Running on our platform; live for a client in November.
Cookies
Help us make this site more useful?
With your OK, we use cookies to see which pages help and which don't, and to measure our ads on Instagram and Facebook. You can change this any time. Privacy policy