Welcome back to the EUVC Podcast where we dive deep into the craft of building and backing venture-scale companies in Europe.
Modern software doesn’t fail quietly.
It fails on Black Friday.
It fails while the CFO is in a board meeting.
It fails when your biggest customer is mid-way through a critical workflow.
And when it does, there’s one brutal reality:
The data is there but nobody has time to interpret it.
Today we’re exploring one of the most under-discussed yet mission-critical parts of building modern software: reliability in production.
Joining Andreas are:
👩🏻💻 Poone Mokari: CEO & Co-Founder, ewake
Paris-based startup building AI agents for software production reliability, fresh off a $2M pre-seed led by Connect Ventures.
💥 Pietro Bezza — Managing Partner, Connect Ventures
Europe’s most product-obsessed early-stage investors (Aikido, Typeform, TrueLayer), backing ewake as their next agentic AI investment in observability.
We unpack why observability is overdue for a rewrite, how AI agents finally provide the “reasoning layer” that logs & metrics never could, and how ewake is building a global devtools company out of Paris.
Here’s what’s covered:
01:12 | What ewake does — AI agents for software production reliability that reason across logs, metrics & code to cut through observability overload
02:32 | Why Connect backed them — trusted intros, a massive category (post-cloud, multi-$B), and founders with rare insider insight into reliability engineering
05:18 | The shift AI enables — from reactive data dashboards to an intelligence layer that correlates structured + unstructured data and finds root causes
07:48 | The hidden layers of tech — why deep, unglamorous infrastructure (observability, reliability, SRE workflows) is a massive opportunity for new entrants
08:52 | The wedge — LLMs as reasoning engines over infrastructure data: not more dashboards, but an operator that collaborates with engineers in critical moments
11:48 | Production ≠ code on your laptop — the real-world complexity: business context, urgency, multi-team coordination, and why semantic reasoning matters
14:38 | “Can we trust AI?” — why agentic workflows differ from ChatGPT, how ewake constrains context, guards against hallucinations & enforces “don’t know” responses
16:38 | Founder–market fit — living the pain at Criteo, deep SRE experience, and product instincts that made ewake’s pitch compelling pre-product
17:16 | Connect’s thesis — product-first founders, problem insight over pedigree, and why product is the highest leverage driver of venture-scale outcomes
22:31 | Product-led ≠ PLG — clarifying the difference between product-first strategy and the specific go-to-market motion of product-led growth
26:02 | How Awake raised $2M pre-product — insight clarity, storytelling from lived experience, fast-moving investors, and a clear “teammate, not dashboard” vision
30:40 | What Connect looks for — opinionated founders with singular insight, UX instincts, and a tinkerer’s mindset for frontier-tech categories
38:20 | Why build in Paris — deep AI talent pools, strong engineering culture, global problem space, and a shift toward France as a magnet for AI founders
42:15 | Geography myths — why great companies emerge anywhere, Europe’s deep industry advantage, and dual-hub (EU + US GTM) playbooks
47:23 | Where ewake is now — out of stealth, hiring, in design partnerships, building alongside early users, and stress-testing agents in real incidents
51:52 | Final reflections — design-led vs. tinker-led founders, why ewake fits the frontier-tech profile, and what the next wave of AI infra looks like
🎧 Listen on Apple or Spotify — chapters ready to go.
✍️ Show Notes
ewake: The AI Teammate for Reliability Engineering
ewake builds AI agents that help engineers understand, diagnose, and resolve production issues faster — without drowning in dashboards.
Not another observability platform.
Not another data pipeline.
Not another widget to configure.
A reasoning layer.
On top of Datadog, Grafana, Prometheus, Sentry, GitHub, and Slack.
The Pain Today
Software teams operate in chaos during incidents:
Millions of logs
Hundreds of metrics
Dozens of dashboards
Slack threads exploding
Code diffs deployed minutes earlier
Alerts half the org doesn’t understand
Everything is “visible.”
Nothing is explainable.
What teams truly lack is interpretation — the meta-layer that answers:
What actually went wrong?
Where should I look first?
What changed right before?
Is this pattern anomalous?
Who should be paged?
Will this get worse?
ewake’s AI Agent
ewake reads all signals — logs, metrics, traces, deploy diffs, configs, threads — and reasons over them to:
highlight root-cause candidates
surface suspicious diffs
cluster related anomalies
explain issues in natural language
summarize evolving incidents
provide next-step suggestions
offer context to non-engineering stakeholders
And it does this where engineers already collaborate: Slack.
“Developers don’t need more dashboards.
They need a teammate that understands the data.” — Poone
This is a UX shift as big as the move from monitoring to observability — from data → reasoning.
Why Connect Ventures Backed ewake
Pietro breaks it down:
1. The intros were from world-class filters
One design-led angel + one observability founder → same intro.
That’s the Connect Ventures bat-signal.
2. The market is enormous
Observability is one of enterprise AI’s largest whitespace opportunities.
3. ewake has founder-market fit
Poone and her co-founder Omid spent years on 3 a.m. on-call rotations at Criteo — one of Europe’s toughest production environments.
4. AI is the wedge incumbents can’t react to
They have the dashboards.
ewake builds the intelligence layer above them.
5. A product-first, tinker-first founding style
ewake isn’t building a five-year roadmap.
They’re iterating at the frontier, where models, agents, and capabilities evolve monthly.
“We overweight insight, product obsession, and problem truth over prior founder experience.” — Pietro
AI Agents ≠ ChatGPT (And Why This Matters in Production)
The #1 reliability question:
“Will it hallucinate?”
ewake’s approach prevents this:
bounded context (specific issue windows)
deterministic workflows
guardrails
“I don’t know” scoring
multi-step reasoning
no generative fiction
ewake doesn’t build a generalist assistant.
It builds an operator.
One that can say:
“Here’s what I know. Here’s what I don’t. Here’s what you should check.”
Trust is the oxygen of reliability.
ewake’s design is built around that principle.
Why Build in Paris?
Paris has become Europe’s AI powerhouse:
Mistral effect
Deep ML research bench
Top engineering schools
A growing DevTools ecosystem
Cross-border talent magnet
ewake is built in Paris but designed for the world.
Its customers are global.
Its engineering culture is global.
The problem is universal.
Where ewake Is Today
ewake is:
out of stealth
deployed with design partners
used internally at ewake during incidents
iterating with human-in-the-loop feedback
expanding agent reasoning workflows
embedding deeper into Slack-based incident response
Their north star?
Trust.
An AI teammate that engineers actually rely on when the fire alarms go off.
💡 Investor & Founder Takeaway
ewake is building the missing layer in observability:
the reasoning layer.
The shift mirrors what Copilot did for software creation — but for production operations:
from “more data” → actionable, contextual intelligence.
It’s an enormous market, a painful problem, and a founding team with rare insider knowledge.
