Discussed · Episode 1 · 14 August 2026
Where AI Stands in Software Development
The first 3 AImigo conversation compares three practical lenses: AI4SDLC, AI-Assisted Engineering adoption, and management of a large engineering organisation.
3 AImigo
Topic map
Follow the connections between questions we have discussed and topics we want to cover next.
Updated: · The order and dates of planned topics may change.
- Discussed
- Next episode
- Planned
- In reserve
Delivery and engineering systems
Models and AI economics
People, roles and skills
Agent autonomy and control
The agent as a customer
AI-native companies and states
Monthly AI digests
Arrows show which topic continues an earlier one; dashed lines connect topics that share a question.
Delivery and engineering systems
Models and AI economics
People, roles and skills
Agent autonomy and control
The agent as a customer
AI-native companies and states
Monthly AI digests
Discussed · Episode 2 · 21 August 2026
AI Writes More Code. Why Doesn't Delivery Speed Up?
AI can make an individual engineer write code much faster, yet a higher volume of changes does not by itself shorten end-to-end delivery of value to users.
Connections
Next episode · November 2026
Infrastructure for agent-driven development
What happens to CI, version control, merge queues and permissions when thousands of agents prepare changes, and how to count the cost of an accepted change including failed attempts, checks and rollbacks.
Connections
Planned · November 2026
Who reviews the code an agent wrote
What approving a change means when nobody has read all of it, and who owns the evidence of correctness and the risk decision.
Connections
In reserve
Spec-driven development: discipline or waterfall 2.0
When a specification for an agent pays off, and when it turns into an extra tax on every change.
Connections
In reserve
Legacy and migrations: rewrite it over a weekend?
Agents made translating code cheaper, but not acceptance and ownership: what counts as the correctness criterion for a migration.
Connections
Planned · November 2026
Inside the model: the minimum an engineer needs
How weights, fine-tuning and inference work, and why the cost of a solved task is not the price of a token.
Connections
Planned · December 2026
Token budgets: FinOps for AI
How to plan and defend model spending: the cost of an accepted change, request routing and soft limits.
Connections
In reserve
Your own models or the frontier
How to choose between open weights and frontier models by the cost of an accepted task and the ability to switch vendors.
Connections
Discussed · Episode 4 · 4 September 2026
AI Hires AI. How Do We Redesign Tech Interviews?
AI now participates on both sides of hiring, so banning a tool cannot restore the old signal: résumés and familiar online stages increasingly fail to predict performance in the role.
Connections
Discussed · Episode 5 · 11 September 2026
A Junior Without Easy Tasks. How Do You Enter Tech in the AI Era?
Building a product is easier, but landing a first engineering job is harder: AI takes over the small tasks that once trained new engineers.
Connections
Discussed · Episode 6 · 18 September 2026
How to Keep Up with AI Changes Without Draining Your Energy
Models, agents, and research arrive faster than anyone can evaluate them. Chasing every update takes time away from practice and independent thought.
Planned · December 2026
Do we need team leads when the team is people and agents
How the manager's role changes when agents grow the amount of started work faster than the amount of verified work.
Connections
In reserve
Will we forget how to program
Whether AI helps build a skill or only amplifies an existing one, and what that means for people who are just starting.
Connections
Discussed · Episode 9 · 9 October 2026
What Can We Trust an AI Agent With? Boundaries of Autonomy
What does it mean to delegate a task, a role, or an end-to-end outcome to an agent? The co-hosts examine autonomy through content releases at Flo Health, a game migration between platforms, and the creation of podcast covers.
Connections
In reserve
An agent made a mistake: investigating incidents
How to investigate an incident when the actor is an agent that can alter the evidence and switch off monitoring.
Connections
In reserve
Who owns the agents in a company
How to split responsibility between the business sponsor, the platform and security — and who may stop an agent.
Connections
Discussed · Episode 10 · 16 October 2026
Agents with Payment Authority: Would We Trust AI with Our Money?
How can we delegate an entire purchase to an agent? The co-hosts begin with family groceries, Vinted searches, and unnecessary extension leads in an order. Discovery, selection, negotiation, order approval, and payment require different permissions; a budget cannot replace knowledge of personal preferences.
Connections
Planned · December 2026
Loyalty when an agent does the buying
Where loyalty moves when an agent makes the choice: into machine-readable benefits, model preferences or the agent's owner.
Connections
In reserve
Your next user is an agent
How a product should accept traffic from agents acting on someone else's authority, and what trust in them should rest on.
Connections
Discussed · Episode 8 · 3 October 2026
AI-Native: What Does It Mean for a Company?
What distinguishes an AI-native company from a team where everyone uses AI? The co-hosts compare three perspectives: agents working across employees, a path from intent to outcome with fewer manual handoffs, and redesigning the overall system in which people and agents work together.
Connections
Discussed · Episode 11 · 23 October 2026
The AI-Native State: What It Is and Where the Boundary Lies
Can the AI-native company model transfer to government? Evgeny Sergeev, Aleksey Litvinov, and Alexander Polomodov discuss agency goals, the citizen journey, and the difference between employee assistance and autonomous execution of rules.
Connections
Discussed · Episode 3 · 28 August 2026
August 2026 AI Digest: Cheaper AI, Harder Adoption
Six connected stories from August reveal the same shift: access to models is getting cheaper while building a working AI system is getting harder.
Connections
Discussed · Episode 7 · 25 September 2026
September 2026 AI Digest: Cheaper Generation, Costlier Verification
The September digest moves the conversation from the models themselves to the systems around them: OpenAI sells the Codex harness as the Agents API, providers push cloud agents and new plans, and agents gain traces, policies, and outcome metrics like production services.
Connections
Discussed · Episode 12 · 30 October 2026
October 2026 AI Digest: Anthropic IPO, New Models and Agent Security
Alexander Polomodov, Evgeny Sergeev, and Aleksey Litvinov discuss October's AI news: faster vulnerability discovery, the economics of AI labs, and new models.
Connections
Planned · November 2026
November 2026 AI digest
The month's main news on AI in software development and what it means for engineering teams.
Planned · December 2026
December 2026 AI digest
December's news and the year in review: what changed in AI for software development in 2026.