The Runbaze platform is opening to early teamsJoin the waitlist →
Platform
Solutions
Company
Who we are
About Careers Contact
Follow along
LinkedIn ↗ Substack ↗
Newsroom
Book a demo →

Newsroom

Essays, working papers, and field notes, published as we go.

EssayNov 2026 Before automating a workflow, agree which data to trust An agent does not ask which number is right. It takes the one it is given. The one-page data agreement to write before you automate: one version of record, a named owner, and a rule for conflicts. EssayNov 2026 AI transformation for service providers Service firms sell hours, so growth has always meant more people. Four steps that break that link, and the foundation that makes every agent after the first one faster. EssayOct 2026 Which process should get your first agent? Repetition is the wrong test. Three tests that find the right first process, one worked example, and why the first agent should be a capability on the platform, not a script next to it. EssaySep 2026 Making legacy applications AI-ready An AI layer inherits the application underneath it. How to modernise a legacy system one slice at a time: data first, critical rules off the model, agents last. EssaySep 2026 From loops to graphs Single agents in loops demo well and fail at scale. Connecting specified work into an explicit graph is the fix, and it is exactly how Pulse is built. Case studySep 2026 Proof of origin, proof of process Open Food Chain builds product passports for the food industry. From validated idea to live product: a foundation, an audit, and development beyond the chat interface, on the record. EssayAug 2026 The four tiers of company memory Agents write now, and a knowledge base pollutes by default. Canon, memory, archive, bin: up is a decision, down is disuse. Open sourceAug 2026 Open-source anomaly detection for wildfires NASA satellite data, anomaly detection with a second opinion, and a person at the gate. Released worldwide, MIT-licensed, free to self-host. ExplainerAug 2026 What does the EU AI Act require of AI deployers? What applies already, what arrives in December 2027, and why the deployer obligations read like the Principal Firm written into law. Working paperAug 2026 The Principal Firm The paper behind the platform: why AI collapses the cost of execution, and what the firm that remains must be able to answer for. PostJul 2026 What is a forward deployment engineer? The role at the centre of how we deliver: embedded at the client, until the operation runs on what was installed. Press releaseJun 2026 Pulse 1.0: the first release of the Runbaze platform Built for software delivery teams and installed in your own environment: agents on the board, safety rules, cost budgets, evaluations, and a record of every run. Case studyMay 2026 Letting AI run the building Coolnomix: an energy platform that adjusts real buildings on its own. What it takes to trust automation with physical consequences: validation before action. Case studyMay 2026 Moving a decade of live data without stopping the business A mission-critical telco ERP rebuilt and a decade of live data migrated while the operation kept running. The Strangler Fig method, and why the estate comes first. EssayMay 2026 The transformation to the Principal Firm A form defined by subtraction: the operating core, middle line, and technostructure go. What remains specifies and answers. PostApr 2026 Hello, world. Why Runbaze exists: to help the next generation of firms into existence, and to document the work as we do it.
Press

Press contact.

Interviews, comments, or background on the thesis: talk directly to the founder.