Agents in the Loop — AI Agents Across the Software Lifecycle
By now we all generate code with AI. This meetup covers everything around that: the context that has to exist before a spec, the path from spec to shipped, the long tail of production, and whether your team is actually better for it.
Four practitioners — from monday.com, bit, Flare, and LinearB — each take one stage of that lifecycle. Production systems, not demos — with humans still in the loop where it matters.
Run of show
18:00 Pizza, beer & mingling
18:30 Welcome
Natalie Goldstein · VP of R&D · Flare18:35 How Multiplayer AI Builds the Shared Context Behind Every Spec
Lotem Fridman · Internal AI Lead · monday.com18:50 Spec-Driven Development: Building with AI Agents
Tal Zion · Head of Mobile & Web · bit19:05 Break
19:20 The Unglamorous 80%
Michael Litvin · AI Engineer · Flare19:35 Dev Productivity in the new AI SDLC
Yishai Beeri · CTO · LinearB19:50 Mingling & open discussion
Full talk abstracts at the bottom ↓
About Flare, your host
At Flare we lead that change. Our agents not only write code, but also ship and run our software. Inside the product, the agents also work on real legal cases spanning from divorces, custody, and immigration matters. The agents take the mundane work. People take the judgment calls — which is exactly the work worth hiring for. Learn more at helloflare.com
The talks
➔ How Multiplayer AI Builds the Shared Context Behind Every Spec
Building the Context, not the Factory
Lotem Fridman · Internal AI Lead · monday.com
Before spec-driven development can start, someone has to curate the context — what customers said, what the data shows, what design already knows, what the codebase allows and how it should be changed. That context doesn't collapse into one spec; it branches into as many as the work needs, with each discipline pulling the slice that's theirs. PM, Designer, Analyst, and Engineer work in parallel, each paired with their own agent, all reading and writing into the same virtual context window — so knowledge compounds instead of decaying as the work progresses.
➔ Spec-Driven Development: Building with AI Agents
How specs become production code through agentic workflows.
Tal Zion · Head of Mobile & Web · bit
Explore modern AI-driven development workflows, from GitHub Spec Kit to BMAD and custom agentic approaches. Learn where AI excels, where humans stay in control, and how to build production-ready software faster.
➔ The Unglamorous 80%
AI for the ongoing work of running software.
Michael Litvin · AI Engineer · Flare
Writing a feature is a small slice of a system's life. Most of it is monitoring, triage, maintenance, cross-system planning, and keeping institutional knowledge current — the work that never ends. This talk is about handing that work to AI agents. Two systems doing it at Flare: Hermes watches production and wakes an agent only when it's worth it; Donna holds the institutional memory that lets agents reason across systems. All of it with humans still holding the approvals.
➔ Dev Productivity in the new AI SDLC
Measuring AI's real impact — beyond the vibes.
Yishai Beeri · CTO · LinearB
AI is changing the SDLC and with it how we think about productivity. This session will dive into how and what to measure in the new world, to understand AI's impact on your dev productivity.