The Agentic Sprint
A human-gated operating model for autonomous software delivery.
Software, systems & the people behind them
Siri Dalugoda
I build software, study how engineering systems behave, and write about what survives contact with production.
Writing and research on software engineering, autonomous systems, AI governance, security, technical delivery, products and the organisations around them.
Selected writing
A small collection across essays, research and engineering practice.
Read
The Agent Never Logged Off
Persistent AI agents are about to break one of security's oldest assumptions, that activity has boundaries. A case for continuous agent authority, authority leases and memory provenance.
Building a Persistent AI Engineering Agent
How to build an agent that picks up a ticket, works across six repositories for days, survives restarts and produces reviewable pull requests. Make the work persistent, not the model.
Human Delegation Provenance
Credited research and open-specification work on keeping human authorisation verifiable across chains of autonomous agents.
Helixar research / open specification
Autonomous Loop
An open-source engineering experiment exploring resumable, verifiable and bounded autonomous software execution.
The Agentic Sprint methodology
The Agentic Sprint is published in full at theagenticsprint.com, where it can be read, implemented and argued with.
The essay on this site introduces the model in plain language. The methodology that defines it precisely, a manifesto, a normative specification, the standards around it and the templates a team can put to work on Monday, now lives on its own site. Every document there states what it requires, what it merely recommends and what remains an open question.
Latest writing
Essays, research, concepts and shorter field notes from the archive.
- I built an AI office staff for trade businessesHow I built ONYX1 Office: six AI office roles for small trade firms that run on the owner's own machine, with an animated office view and scheduled routines.
- AI brain rot: what happens to engineers when the hard part becomes optionalWhat the evidence says about AI, deskilling and the next generation of engineers: cognitive debt, the apprenticeship gap, and what is worth still doing by hand.
- We may have already created a new kind of AGIThe original AGI vision imagined an artificial human. What emerged may be distributed, non-biological and organism-like. Have we already crossed a threshold?
- Leadership isn’t about knowing; it’s about navigatingLeadership is less about having every answer and more about setting direction, listening well, and moving through uncertainty.
- The momentum mindset: move, learn, adjust. Don’t wait for perfectTeams create clarity by moving early, learning from evidence, and adjusting before small problems become large ones.
- Not every engineer can deliver every projectSoftware delivery improves when leaders match people’s strengths and working styles to the demands of each project.
Explore by topic
Browse the questions that connect the writing and research.
- Agentic systemsAutonomous agents, orchestration, memory and bounded execution.
- EngineeringArchitecture, quality, tooling, reliability and software delivery.
- AI governance & securityAuthority, provenance, evidence, controls and accountability.
- Delivery & organisationsOperating models, teams, decision-making and technical leadership.
- Products & operationsB2B software, workflows and systems built around real work.
- Notes & experimentsShort observations, prototypes, questions and things learned while building.
Research & projects
Writing, research records and software, with ownership and attribution stated at source.
Some work here is independent writing or open-source engineering. Other work was developed with an organisation or published elsewhere. Each relevant page explains the relationship and links to the canonical source.
About the author
I work across software delivery, autonomous systems, AI governance, security and operational products. This site is where I make the reasoning, evidence and unfinished questions behind that work available to other people.
Read more about Siri