Call for Speakers – AI Heroes 2027

From Agents to AI-Native Teams

In 2026 agents left the demo stage. They now open pull requests, run tests, triage incidents and talk to other agents through MCP. The hard questions are no longer about what a model can generate. They are about how teams design, secure, measure and run software when part of the work is done by agents.

AI Heroes 2027 is about that shift. We want to hear from the people who are doing it in production.

AI Heroes is the international conference on AI for developers, by developers. A place where builders compare notes, share battle-tested patterns and shape what "AI-native engineering" really means.

Three events, one community

In 2027 AI Heroes is a series of three events in Italy's innovation hubs:

You can submit once and tell us which dates you are available for. Each event is selected separately, about a month before it takes place.

The format

One track, one day, three parts:

• Morning – Hands-on workshops. Two deep-dive workshops of about 90–100 minutes, up to 30 participants each. Attendees bring their own laptop.
• Afternoon – Conference. Six 30-minute talks with case histories, patterns and vision, for about 100 attendees.
• Evening – Networking aperitif with speakers, attendees and partners.

You can submit more than one proposal. You can also submit a workshop and a separate 30-minute talk on the same topic.

Who is in the room

Expert developers, tech leads, platform engineers, architects, product managers and CTOs who build and run software for a living. They come to learn what works, what fails and what to try on Monday morning.

What we are looking for

Real experience only, no sales pitches. We look for production case studies, numbers, failures and the lessons they taught you, patterns and playbooks the audience can apply the next day.

1. Agentic software engineering

How agents are changing the whole software development life cycle, and the teams that run it.

Keywords: coding agents in CI/CD, background and asynchronous agents, spec-driven development, agent skills and reusable playbooks, AI code review and testing, legacy modernization, measuring real impact (DORA, DevEx, cost per change), new roles and skills, onboarding juniors in an agentic team.

2. Building AI-powered products

From prototype to reliable features that users trust.

Keywords: retrieval and knowledge systems beyond basic RAG, voice and multimodal apps, generative and intent-based UI, copilots inside existing products, mobile and on-device AI, latency and cost budgets, rollout and rollback strategies.

3. Multi-agent systems and protocols

Beyond the single chatbot: how agents plan, call tools and work with each other and with people.

Keywords: orchestration patterns, MCP servers and clients, agent-to-agent protocols (A2A), durable and long-running workflows, human-in-the-loop design, when not to use agents.

4. Context engineering and memory

The quality of an agent depends on what it knows at the right moment.

Keywords: context and memory design, token economy and context compression, documentation as context, enterprise knowledge and knowledge graphs, retrieval for codebases, keeping context fresh and consistent.

5. Trust, security and evaluation

How do you know an agent is doing the right thing, and how do you stop it when it is not?

Keywords: evals in CI, observability and tracing for LLM apps, prompt injection and tool abuse, permissions and sandboxing for agents, supply-chain risks of skills, plugins and MCP servers, red teaming, guardrails that do not kill velocity.

6. Sovereignty, cost and regulation

Adopting AI while keeping control of data, budgets and compliance.

Keywords: open-weight and small language models, on-premise and edge deployment, fine-tuning vs. prompting, FinOps for AI, data residency, the EU AI Act in practice (transparency and general-purpose AI obligations, AI literacy, preparing for the high-risk rules).

What makes a strong proposal

• A clear problem and a concrete outcome: what will the audience be able to do afterwards?
• Real experience: architecture, numbers, trade-offs, what went wrong.
• For workshops: prerequisites, what participants build, what to install beforehand.
• No sales pitches. Product demos, marketing decks and "buy our tool" talks will be rejected. Speakers from vendors are welcome only with real engineering experience the audience can reuse, whatever tools they use.

Selection

Proposals are reviewed by a committee of past AI Heroes speakers. Original talks, never presented at other conferences, have priority.

Selected speakers receive:

• a full-day ticket (workshops and conference) for the event where they speak, networking aperitif included.

Travel and accommodation. The organisation cannot cover travel and accommodation in general. Within a limited budget we can reimburse up to €300 for two speakers, with priority to workshop speakers.

Companies that cover their speakers' travel and accommodation become event partners and receive:

1. their logo on the event website and newsletters;
2. five extra afternoon conference tickets for their team;
3. a 20% discount code on all tickets to share with their network.

Important notes

• All talks and workshops are live and in person in Milan, Turin or Bologna. No remote sessions.
• The conference language is English.
• You can edit your proposal at any time. After a closing date, please let us know if you change anything.
• By submitting a proposal you accept our Privacy Policy.

Do you have a story that can help other teams ship better software with AI? Submit your proposal and help us shape AI-native engineering.

Official website: aiheroes.it

Submissions close on 2027-10-12 23:59 (Europe/Rome), 11 months, 3 weeks from now.