Sonarly: The AI On-Call Engineer Turning Sentry Noise into Fixed PRs

2026-03-16 | ProductHunt | Official Website | YC Profile

The homepage shows the core integration flow: Monitoring tools (Sentry/Datadog/Grafana) → Sonarly Smart Categorization → GitHub PR + Slack Notification

30-Second Quick Take

What is this app?: Connect your Sentry, Datadog, Grafana, and other monitoring tools to automatically filter alert noise, investigate root causes, and submit fix PRs directly to GitHub/GitLab. Simply put, it's a 24/7 AI on-call engineer that never sleeps.

Is it worth watching?: Worth watching, but keep expectations in check. This is a 2-person YC W26 team product addressing a very real pain point (too many engineers drowning in Sentry alerts), but the product is still in its infancy with only 11 PH votes. Great for small teams to experiment with, while larger teams might want to wait and see.

Three Questions to See if It Fits

Is this for me?

Target Users:

• Small teams (2-10 people) bombarded by Sentry/Datadog alerts
• Startups without dedicated SRE/DevOps
• Indie developers / Full-stack engineers who have to write features and handle production issues

Is it me? If you open your inbox every day to a pile of Sentry alerts, 80% of which are duplicates or false positives, but you're too afraid to ignore them all—you're the target user.

Scenarios:

• 3 AM Sentry error → Sonarly has already investigated and submitted a PR; you just review it in the morning
• 50 alerts a day but only 5 matter → Sonarly filters the noise
• Want to know the root cause but no time to dig through logs → Sonarly correlates logs/traces/metrics for you

Is it useful?

ROI Judgment: If you're a small team, the entry cost is near zero (free + 3 minutes), and the benefits are obvious. It's worth a shot.

Is it well-received?

The "Aha" Moments:

• Alert Noise Reduction: Merges 50 alerts into 5 actionable ones, so you don't have to hunt for signals in the noise.
• PRs with Evidence: It doesn't just change a line of code; it includes the full investigation process (logs, traces, metrics screenshots).

Real User Feedback:

"really helps to cut through all the emails I get from Sentry" "spending much less time on alerts and fixes now" — ProductHunt User

"the alert deduplication alone sounds like it would save a ton of time" — Hacker News Comment

"most teams just drown in Sentry noise and end up ignoring half of it" — Hacker News Comment

For Indie Developers

Tech Stack

• AI/Model: Claude Code + Opus 4.6 (Anthropic's strongest model)
• Integration Layer: Connects via OAuth and Webhooks to Sentry, Datadog, Grafana, CloudWatch, Vercel
• Code Platforms: GitHub (via GitHub App), GitLab (including self-hosted)
• ChatOps: Slack, Discord
• NPM Package: @sonarly/tracker

Claude Code in action: AI automatically investigates Vercel and Sentry logs, correlating multiple data sources to pinpoint the root cause

Core Feature Implementation

Sonarly's core logic follows three steps:

1. Alert Ingestion + Deduplication: Receives alerts via webhooks, uses semantic analysis to merge duplicates (understands different symptoms of the same root cause, not just string matching).
2. Root Cause Investigation: Uses Claude Code as a coding agent to access codebases, logs, traces, and metrics to build a "system runtime map."
3. Fix Generation: Generates code changes based on findings, submits PRs to GitHub/GitLab with a full evidence chain in the description.

Key Technical Point: It doesn't just look at code—it understands the runtime context. This is the fundamental difference between Sonarly and pure code AI like Cursor or Copilot.

Open Source Status

• Is it open source?: No, closed-source SaaS
• Similar Open Source Projects: DrDroidLab/sample-debug-agent (similar concept but rudimentary)
• Difficulty to build yourself: High. The challenge isn't the AI call; it's the multi-source correlation + continuously learning system representation. Estimated 2-3 people, 6 months.

Business Model

• Monetization: SaaS subscription (pricing not yet public)
• Current Status: Free to use (early user acquisition phase)
• Discount Code: SONARLYHN —— 2 weeks free

Risk from Giants

High risk. Sentry is already working on AI Autofix (now called Seer), and it's doing well:

• Sentry Autofix uses Claude 3.7 Sonnet + Gemini Flash 2.0
• 95% root cause accuracy, 54% fix success rate
• Already testing "auto-trigger Autofix" features

PagerDuty has also launched a full AI Agent Suite (SRE Agent + Scribe Agent + Insights Agent).

Sonarly's moat lies in being a cross-platform middleware. Sentry's Autofix only fixes errors caught by Sentry; PagerDuty's AI doesn't generate code fixes. Sonarly connects multiple monitoring sources + codebases simultaneously. This is a differentiator, but also the easiest gap for any giant to close.

For Product Managers

Pain Point Analysis

• Problem Solved: Engineering teams are overwhelmed by alerts; 80% is noise, and the remaining 20% takes massive time to investigate.
• How painful is it?: The founders' own experience—a 2-person team getting 50 Sentry alerts a day. This is extremely common in small teams.
• Frequency: Daily, high-frequency necessity.

User Persona

Feature Breakdown

Competitive Landscape

Takeaways

1. DX design with a "3-minute onboarding": No SDK installation required; connect via OAuth to lower the barrier to entry to the absolute minimum.
2. Include investigation evidence in PRs: Don't just provide the fix; show why it was changed to build trust.
3. Continuously learning system representation: A flywheel effect where it gets more accurate over time, increasing stickiness.

For Tech Bloggers

Founder Story

Two French teens: Dimittri Choudhury and Alexandre Klobb started coding and freelancing at 16. They moved from a small French village to Paris for CS, co-founded Meoria (an edtech tool helping French students choose universities), grew it to 100k+ users in 6 months, and then—dropped out.

They moved to SF to join YC W26. The motivation for Sonarly was very practical: while running Meoria, their 2-person team got 50 Sentry alerts a day and couldn't keep up. So they thought, "Can we let AI be the on-call engineer?"

Controversy/Discussion Angles

1. The trust issue of "AI automatically modifying production code": The biggest debate on HN—AI fixes might treat symptoms rather than causes, and one bad fix could make a team lose trust in the whole system.
2. The Autonomy Paradox: Too conservative = just a smart alert dashboard; too aggressive = one crash and it's over. How do you find the balance?
3. Sentry is already doing this: Sentry Autofix (Seer) has been running for over a year with 95% root cause accuracy. How does a 2-person team compete with Sentry's 800+ employees?

Hype Data

• PH Ranking: 11 votes (low, early launch phase)
• HN: Hit the front page; Launch HN post has substantial discussion
• Twitter: Founder @dachoudhury's tweet has 38 likes / 6052 views
• daily.dev: Featured in a report

Content Suggestions

• Angle: "Sentry built its own Autofix, so why is anyone building Sonarly? The opportunity for cross-platform monitoring middleware."
• Trend-jacking: AI Agent automation in DevOps is a hot narrative for 2026; do a comparison of PagerDuty AI + Sentry Seer + Sonarly.

For Early Adopters

Pricing Analysis

Hidden Costs: None. No SDK installation or code changes required; connect via OAuth.

Getting Started Guide

• Setup Time: Official claim is 3 minutes
• Learning Curve: Extremely low
• Steps:
  1. Visit sonarly.com and sign up.
  2. Connect your Sentry/Datadog/Grafana via OAuth.
  3. Connect your GitHub/GitLab repositories.
  4. Configure Slack/Discord notification channels.
  5. Wait for alerts; Sonarly starts working automatically.

Pitfalls and Gripes

1. Don't blindly trust AI fixes: PRs must be manually reviewed. AI might fix the symptom but miss the root cause.
2. Don't enable Auto-merge: Most teams are cautious about AI auto-merging code; keep a human-in-the-loop.
3. The degree of autonomy: The founders admit this is the hardest part to nail—too conservative is just a dashboard, too aggressive risks a disaster.
4. Early-stage risk: 2-person team, $500K funding; the product could pivot or shut down at any time.

Security and Privacy

• Data Access: Requires high-level access to codebases, logs, and monitoring data.
• Code Storage: Need to confirm if code is stored on their servers (not explicitly stated).
• SOC2/ISO: Early product; likely no compliance certifications yet.

Alternatives

For Investors

Market Analysis

• Sector: AIOps (AI for IT Operations)
• Market Size: $19B-$47B (2026, depending on the analyst)
• Growth Rate: 15%-27% CAGR
• Drivers: Rising cloud-native complexity, microservice alert explosions, high engineering hiring costs.
• Key Data: 64% of large enterprises have deployed automated incident response systems.

Competitive Landscape

Timing Analysis

• Why now?:
  1. LLMs like Claude Opus 4.6 have reached practical levels of code understanding.
  2. The Coding AI agent wave (Cursor, Devin, etc.) has educated the market.
  3. But existing coding agents don't understand runtime context—this is Sonarly's entry point.
• Tech Maturity: Just at the tipping point of usability. AI can read code and logs, but the reliability of auto-fixes still needs validation.
• Market Readiness: Medium. Developers are more open to AI writing code, but there's still a trust barrier for AI modifying production code.

Team Background

• Dimittri Choudhury: Started freelancing at 16, full-stack + AI engineer.
• Alexandre Klobb: Started freelancing at 16, technical co-founder.
• Team Size: 2 people.
• Past Success: Meoria (French edtech), 100k+ users in 6 months.
• Education: CS in Paris (dropped out).

Funding Status

• Funding: $500K (1 round)
• Investors: Y Combinator
• Date: January 2026
• Valuation: Not disclosed
• Source: Tracxn

Risk Assessment

1. Giant Squeeze: Sentry Autofix is already here and has a data advantage.
2. Trust Ceiling: Many teams won't pass security audits for AI automatically modifying production code.
3. 2-Person Team: Can they build enough technical moat and user base before the giants catch up?

Conclusion

The Bottom Line: Sonarly hits a very real pain point (alert fatigue + lack of manpower) by automating the "monitor -> investigate -> fix" loop with an AI Agent. However, as a 2-person YC early-stage team facing direct competition from Sentry Autofix, its window of opportunity is short.

Resource Links

2026-03-16 | Trend-Tracker v7.3 | Research Time: 3-stage layered search completed