This AI already writes 20% of Salesforce’s code. Here’s why developers aren’t worried

[ad_1]

Join our daily and weekly newsletters for the latest updates and exclusive content in the industry’s leading AI coverage. Learn more


When Anthropical CEO Dario Amodei announced the EU Write 90% of the code During the six months, the world of coding is wrapped in the mass out. But inside RighteousnessA different reality has already been formed.

“About 20% of all Apex code Written in the past 30 days Agent“Jayesh Govindarajan, Sale Vice President of Salesforce AI does not know me during the last meeting.

However, Salesforce’s developers do not disappear. Develops.

“The vast majority of development – at least what I called on the first project of the code – will be written by AI,” said Govindaraj. “However, this first project is based on the root of which developers.”

Up to strategic control from code lines: developers become technology pilots

Program engineering is always mixed with creativity with Tedium. Now the AI ​​is managed by the last and developers to the past.

“You move from a more strategic role than a strategic role,” said Govindaraj. I have something to build “only” so I’ll build, ‘but’ but what ‘should we build? What does the client actually want? ‘”

This change reflects other technological disorders. When the calculator was replaced by hand calculation, mathematicians did not disappear – they solved more complex problems. When the digital cameras were killed in dark rooms, the photography expanded more than a contract.

Salesforce thinks the code works in the same way. As the AI ​​software scattering the costs of the program, developers are always obtaining the missing: time.

“Once you create a working prototype, now it takes hours, now you need hours,” he said. “Instead of showing a document explaining what customers can build, you simply hand it over to the processing program. Then you repeat according to their reaction.”

‘Vibe Coding’ Here: Why Software Engineers close more AI than writing each order

Codes began to master the calls “Vibe coding“- A term developed by Openai co-founder Andrej Karpathy. The experience processes what produces higher direction than accurately, then produces.

“You are just a kind of high-level direction and allow AI to use your creativity to create the first draft,” said Govindarajan. “It will not work as you want but gives something to play.” It looks good, do more “or ‘these buttons Janky, I don’t need them.” “

The process compares to musical cooperation: “AI, develops the developing melody, determines the rhythm.”

Although AI is superior in the right job app, Govindarajan admits that there are restrictions. “Will you set up the next generation database with vibe coding? Unable. But can you make a really cool ui with a base caller and a fantastic job application?”

New quality Imperative: AI creates more production code

AI not only writes the code in a different way – requires different quality control. Salesforce has developed it Agentforce test center This machine demanded the new inspection approach after discovering the code created.

“These are stochastic systems,” Govindarajan explained. “Even with very high precision, there are scenarios where the scenarios can fail. Perhaps in step 3 or in step 4 or from step 4 or 17 to 17 to 17.

Nature of AI’s performances is not a non-detail of the nature, developers to be a specialist in the border test and guardianship. What to write the code is not to say, but you do not need to evaluate it.

Start the code: How does the AI ​​squeeze the life of all software development

Transformation extends outside the initial coding to cover the life life of the program.

“In the structure stage, the tools understand the existing code and expand everything in acceleration,” said Govindarajan. “Then comes the regression tests that generate test situations for a new code that AI can handle.”

This comprehensive automation creates “significantly harsh loop” between Govindarajan’s opinion and application. The faster developers can be tested and cleaned, how ambitious can be.

Algorithmic thinking is still important: why computer science bases remain important in the AI ​​era

Govindarajan often provides concerned questions about software engineering.

“I always ask people to learn computer science.” “The answer is completely yes, because algorithmic thinking remains important.

How these skills change. Instead of directing each solution character, developers to make optimal results in AI tools. The person gives judgment; The machine gives speed.

“You still need good intuition to give the right instructions and access the output,” Govindarjan said. “It is necessary to enjoy what EU works and what it works and whatever it does or know.”

Strategic altitude: developers become more business partners than technical practices

The coding itself is also recognized, the developer has to join more work strategies.

“Developers accept the roles of control, agents working with their names,” Govindarajan explained. “But they are responsible for their placement. Buck still stands with them.”

These upgrades are more approaching to decision makers and applications – a promotion – not to eliminate.

Salesforce supports this passage with the tools prepared for each stage: agentforce code code, agent builder customization and agentforce test center provides reliability. Together, the developers form a platform to grow into this extended roles.

The company’s vision offers a tougher contrast from the narrative “developers are doomed.” Program engineers who enter the worn and adaptable can be considered more important than ever.

In an area where the investment is daily, the most powerful compiler represents the most powerful compiler, not how the code is written, and why he writes. For those who want to improve their mental models for developers, the future will be terminated and less like transsrendence.


[ad_2]
Source link

Leave a Reply

Your email address will not be published. Required fields are marked *