Qodo
@QodoAI

The AI code quality and governance platform for developers.

qodo.ai/
Joined December 2022
600 Photos and videos
Pinned Tweet
Qodo
@QodoAI
Oct 1
Introducing Qodo 3.0: the quality and governance layer for the software factory. Engineering teams are building agentic software factories. Output is up, but code review hasn't kept up. And a factory with no quality control is just a faster way to ship defects. Qodo 3.0 governs your software factory from the first line an agent writes through review and into production. Instead of individual PRs, group your code into work packages as connected changes across repos. Reviewed, ranked, and measured together, with a clear map of impact.
34
15
94
414,673
Where should we draw the boundary for language models? @honeycombio's Liz Fong-Jones joins @nnennahacks and @itamar_mar on The Agentic Review podcast to discuss why lessons from coding don't transfer to every field.
Quote card for The Agentic Review with Liz Fong-Jones, Technical Fellow at Honeycomb. Liz says to keep language models for coding and cautions against applying them to every field just because it resembles software engineering.

ALT Quote card for The Agentic Review with Liz Fong-Jones, Technical Fellow at Honeycomb. Liz says to keep language models for coding and cautions against applying them to every field just because it resembles software engineering.

1
1
4
328
Watch or listen: lnk.to/x8PAp9TW

The Agentic Review

The Agentic Review by Qodo is where technology and engineering leadership meet to unpack real-world lessons from teams adopting AI into their workflows, exploring the gap between high-speed AI coding...

lnk.to
2
89
Qodo retweeted
Nnenna
@nnennahacks
20h
I want prompt changes evaluated like production code. They change which findings developers see. What I like about Qodo's approach: cost, latency, and missed findings measured separately. That's the standard I'd want before shipping a prompt change.
Qodo
@QodoAI
20h
Which code review findings will your team act on? We used GEPA to improve Qodo’s recommendation agent. In a production A/B test, its average cost fell roughly 50% and latency 25%, with comparable precision. Ravid Cohen explains how we tested it.
Article

Using GEPA to Make Code Review Recommendations More Useful

Adapted from Ravid Cohen’s engineering article for Qodo. A code review asks developers to spend attention on a change. Every finding needs enough context to explain why that attention is worthwhile.
1
3
339
Which code review findings will your team act on? We used GEPA to improve Qodo’s recommendation agent. In a production A/B test, its average cost fell roughly 50% and latency 25%, with comparable precision. Ravid Cohen explains how we tested it.
Article

Using GEPA to Make Code Review Recommendations More Useful

Adapted from Ravid Cohen’s engineering article for Qodo. A code review asks developers to spend attention on a change. Every finding needs enough context to explain why that attention is worthwhile.

5
2
13
1,157
When does fine-tuning a model make sense? Gil Elbaz (@gil_elbaz_) discusses latency, scale, and domain knowledge as reasons to train, alongside the expertise that requires. Listen to The Agentic Review podcast with @nnennahacks and @itamar_mar.
2
1
4
358
Watch Gil Elbaz on The Agentic Review podcast: qodo.ai/podcasts/why-ai-cont…

Why AI Control is the New Security Standard

qodo.ai
2
93
31.7% of engineering leaders in our 2026 survey say quality depends on exhaustive manual work. Which controls rely on someone remembering to intervene? Record why a retry was disabled or who owns a data contract, and keep that reasoning current.
Bar chart redrawn from Qodo’s 2026 report, Figure 7. Engineering leaders report fully guardrailed human and AI review, 47.3%; quality maintained through exhaustive manual work, 31.7%; AI review tools missing context-specific or architectural violations, 18.7%; little or no meaningful protection, 1.3%.

ALT Bar chart redrawn from Qodo’s 2026 report, Figure 7. Engineering leaders report fully guardrailed human and AI review, 47.3%; quality maintained through exhaustive manual work, 31.7%; AI review tools missing context-specific or architectural violations, 18.7%; little or no meaningful protection, 1.3%.

1
3
9
646
Compare leaders' reported review environments in our 2026 State of AI Code Quality report, pp. 12–13. Download: qodo.ai/state-of-ai-code-qua…

State of AI Code Quality Report

Qodo’s 2026 State of Code Quality report surveyed 500 U.S. developers and 300 engineering leaders. Get the latest insights on AI and code quality here.

qodo.ai
2
71
What breaks when AI agents reach production? Tomorrow in NYC, we're bringing builders and OSS maintainers together to talk reliability, code review, agent loops and observability. Agentic AI in the Wild Oct 22 · 5–8 PM EDT Corgi Cafe, 121 E 27th St
Event flyer for Agentic AI in the Wild: Open Source Agents in Production, Thursday, October 22 at 5 PM in New York City. Speakers: Dana Fine, Ryan Hamilton, Baruch Sadogursky, Jai Bhagat, Jay Messina and Miriam Lebowitz, plus a surprise speaker. Community partners and supporting organizations appear below.

ALT Event flyer for Agentic AI in the Wild: Open Source Agents in Production, Thursday, October 22 at 5 PM in New York City. Speakers: Dana Fine, Ryan Hamilton, Baruch Sadogursky, Jai Bhagat, Jay Messina and Miriam Lebowitz, plus a surprise speaker. Community partners and supporting organizations appear below.

2
1
2
315
The event's full. Join the waitlist here: luma.com/l7dhbis5

After All Things Open Community Meetup: Agentic AI in the Wild: Open Source Agents in Production ·...

AI and Open Source Builders Meetup, New York City AI agents are moving into production, and more and more of them are built on open source. The question is no…

luma.com
1
69
A reviewer needs to challenge the agent writing the code. Inside Qodo's Dark Factory, the Reviewer never edits the implementation. Shared context lets it question the plan while keeping that responsibility separate.
Qodo's Dark Factory architecture: an engineer, software engineering agent, independent Reviewer, and externally hosted SRE agent collaborate through a shared BAND room. Qodo review, deterministic tools, GitHub, and Notion support the workflow.

ALT Qodo's Dark Factory architecture: an engineer, software engineering agent, independent Reviewer, and externally hosted SRE agent collaborate through a shared BAND room. Qodo review, deterministic tools, GitHub, and Notion support the workflow.

1
1
3
324
The architecture: qodo.ai/blog/from-delegation…

From Delegation to Collaboration: Inside Qodo's Dark Factory

Explore how Qodo’s Dark Factory uses AI agent collaboration to plan, build, and review code with shared context, independent review, and human control.

qodo.ai
1
88
In our 2026 survey, developers use AI for code generation (60%), review (57.6%), and tests (51.4%). Code and generated tests can share a misunderstood requirement. Give verification an agreed acceptance criterion it can check.
Bar chart redrawn from Qodo’s 2026 report, Figure 1, showing selected activities among 500 U.S. developers: code generation, 60%; code review, 57.6%; test generation, 51.4%; security review, 42.2%.

ALT Bar chart redrawn from Qodo’s 2026 report, Figure 1, showing selected activities among 500 U.S. developers: code generation, 60%; code review, 57.6%; test generation, 51.4%; security review, 42.2%.

1
1
4
418
Explore how assumptions can carry across AI-assisted development stages in our 2026 State of AI Code Quality report, pp. 5–6. Download: qodo.ai/state-of-ai-code-qua…

State of AI Code Quality Report

Qodo’s 2026 State of Code Quality report surveyed 500 U.S. developers and 300 engineering leaders. Get the latest insights on AI and code quality here.

qodo.ai
88
I don't read code anymore. The abstraction layer for humans has changed. AI agents are frontline coders now. I define tasks and constraints, and I work with my team lead agent, but my role is that of a product engineer, a general of my robot army: designing logistics, providing provisions (context) and offering judgement for key decisions. I use multiple models to help me, from different model families: - A team lead agent that can either implement work, or assign it to other agents - Automated reviewers to check for biases - Multi-model cross family review for key changes Intentionally, I create smaller iteration loops within the overall code review loop: - Cheap + fast checks and verification from the implementing coding agent can send the change back for additional work - Standardized reproduction of findings - Re-review of final commits post fixes by lead models - Evidence backed merge decisions You need a code review loop that unblocks you as a human, keeping your judgement for the most important decisions, and unblocks your AI factory to move at agent speed. Agents code. Humans lead. Full breakdown featuring @coralbricksdev, @QodoAI, examples from my OSS project @avoidaiwriting , insights from @AnushElangovan, @kimblgn & much more below👇
22
4
13
1,188
Nuance is one of the hardest frontiers in agentic engineering. Making it available to agents is work practitioners will need to do: explain it, codify it, and embed it into the software factories we’re building. In her latest article, @nnennahacks explores how to codify that nuance, and why practitioners who learn to do it well will become increasingly valuable. Read “How to Codify Nuance in Agentic Engineering”.
1
1
4
437
Read “How to Codify Nuance in Agentic Engineering” by @nnennahacks: qodo.ai/blog/how-to-codify-n…

How to Codify Nuance in Agentic Engineering

Make engineering judgment usable by agents with evidence-backed context, scoped memory, standards, and evaluations that test when a decision should change.

qodo.ai
2
102
Liz Fong-Jones of @honeycombio on voice-driven coding: review the transcript before it becomes an agent's instructions. Listen to The Agentic Review podcast with @nnennahacks and @itamar_mar.
1
1
5
393
Watch or listen: lnk.to/x8PAp9TW

The Agentic Review

The Agentic Review by Qodo is where technology and engineering leadership meet to unpack real-world lessons from teams adopting AI into their workflows, exploring the gap between high-speed AI coding...

lnk.to
2
105
When AI writes a change, waiting until the pull request to review it leaves the rework for later. Qodo's Agentic Toolbox brings review into the coding session, using the same review engine and codebase context as PR review. The pull request remains a final gate.
Webinar artwork titled From Code Generation to Code Confidence: Building Quality into the Agentic Workflow, with a code window, review checklist, and Qodo mascot on a dark background.

ALT Webinar artwork titled From Code Generation to Code Confidence: Building Quality into the Agentic Workflow, with a code window, review checklist, and Qodo mascot on a dark background.

2
3
376
Read the session recap: qodo.ai/blog/building-qualit…

Building quality into the agentic workflow

Read about Building quality into the agentic workflow in our blog.

qodo.ai
1
86
Load more