Faster code generation is not the flex everyone thinks it is if you can't tell your board what percentage of it actually shipped. π
Mridhula Venkat shares why the measurement problem starts way before the commit.
Measuring productivity by token usage is like measuring engineering output by lines of code.
It just keeps burning for no reason. π€¦
Harish Doddala, shares why outcome maxing is replacing token maxing as the standard that actually matters. #FinOpsX https://t.co/KdomNvQdkf
The metric that matters isn't tokens consumed.
β¨ Its features shipped. β¨
Introducing AI DLC Insights, from prompt to production, every step visible https://t.co/gRWce46Tzb
Mumbai, we're otw. ππ¨
Catch Umasankar Prakash Mukkara, Head of Resilience Testing at Harness, speaking at KubeCon India on June 18 and find the LitmusChaos team at the booth on June 19.
Full schedule and details: https://t.co/CZe5o0UDTh
Nobody got into engineering to spend their afternoon copy-pasting logs into an AI assistant.
So we fixed that. π€
The Harness VS Code Extension is live and your pipelines, approvals, and AI debugging tools now live in the same panel as your code. https://t.co/DzLpdzo4zW
Turns out immersive art and AI infra go really well together.
Who knew? Us. π
Our collab event with @Sentry and Neon by @Databricks at NYC Tech Week was a success!
Until next time, NYC. π½
Your customer support agent didn't cost $28,000 last month.
It cost $0.60 per resolved ticket.
That's the insight Cloud & AI Cost Management gives you. Unit economics for every production AI agent, in the same platform you already use for cloud cost. https://t.co/WmohZS7j44
How do you experiment at scale without losing control? π€
How do you go from manual releases to governed ones? π€ π€
How do AI agents change who builds and how they ship? π€ π€ π€
Still thinking? Join us at Flagship by Harness. https://t.co/V4WvymL26I
The energy you put into culture is the energy that ends up in your product.
Trevor Stuart, SVP at Harness, on what it really means to build a company people grow with.
DigiCert's SVP of Engineering put it simply β when they go down, their customers feel it instantly. Four nines wasn't a goal. It was a requirement.
Here's how they got there: https://t.co/PUsy2bJ9eV
AI can write the code. It can't own the outcome.
We're hiring for the things no tool can automate. Judgment, ownership, problem solving. Rajni Masoom shares what that looks like in practice via ET GCC World.
π: https://t.co/i4WNhuCzXU
Cleanup on aisle AI. π§Ήπ¨
Most organizations are still treating cost as a retrospective metric. Wait for the damage, file a report, and do it all over again.
Bala Venkatrao shares why that model is broken and what replaces it. https://t.co/O8r4z0z7tQ
Cleanup on aisle AI. π§Ήπ¨
Most organizations are still treating cost as a retrospective metric. Wait for the damage, file a report, and do it all over again.
Bala Venkatrao shares why that model is broken and what replaces it. https://t.co/O8r4z0z7tQ
We're not going to tell you to measure your AI ROI and then not measure our own.
One year of real data from our engineering team. These are the numbers. https://t.co/RRZgUfBZaP
AI busy work has a name now: β¨token maxingβ¨.
Looks like productivity, but isn't.
If you can't say what percentage of AI-generated code reached production this week, this one's for you.
@Codecov is now part of @Harnessio. π
Agents are writing more code than your team can review. Coverage is one of the only signals that still tells the truth about what's tested β and it belongs in the delivery pipeline.
Now it's there. https://t.co/MozFcVBl2B