Topics
Latest
AI
Amazon
Image Credits:LogicStar
Apps
Biotech & Health
Climate
Image Credits:LogicStar
Cloud Computing
Commerce
Crypto
endeavor
EVs
Fintech
fund raise
Gadgets
Gaming
Government & Policy
computer hardware
layoff
Media & Entertainment
Meta
Microsoft
secrecy
Robotics
Security
Social
Space
startup
TikTok
Transportation
Venture
More from TechCrunch
event
Startup Battlefield
StrictlyVC
Podcasts
television
Partner Content
TechCrunch Brand Studio
Crunchboard
Contact Us
Swiss startupLogicStaris bent on joining theAI agentgame . The summertime 2024 - founded startup has pocket $ 3 million in pre - seed funding to bring tools to the developer market that can do autonomous upkeep of computer software applications , rather than the more typical AI agent usance case of code co - development .
LogicStar chief operating officer and co - founder Boris Paskalev ( pictured top right field , in the feature article look-alike , with his fellow co - founders ) propose the startup ’s AI agents could end up partnering with code development agent — such as , say , the likes of Cognition AI ’s Devin — in a business win - win .
Code faithfulness is an issue for AI agents construction and deploying package , just as it is for human developers , and LogicStar wants to do its bit to grease the development wheel by automatically picking up and repair bug wherever they may crop up in deployed code .
As it resist , Paskalev suggests that “ even the best models and agents ” out there are ineffectual to purpose the legal age of bugs they ’re presented with — hence the team spying an opportunity for an AI startup that ’s dedicate to improving these odds and delivering on the dream of less tedious app maintenance .
To this final stage , they are ramp up atop large spoken communication models ( LLMs ) — such as OpenAI ’s GPT or even China ’s DeepSeek — occupy a exemplary - agnostic approach for their platform . This allows LogicStar to douse into dissimilar LLMs and maximize its AI federal agent ’ utility , based on which foundational model form best for resolving a particular codification outcome .
Paskalev contend that the origination team has the expert and domain - specific cognition to build a platform that can break up programming problem which can dispute or outfox LLMs working alone . They also have retiring entrepreneurial success to point to : He deal his prior code review inauguration , DeepCode , to cybersecurity giant Snyk back inSeptember 2020 .
“ In the beginning we were thinking about actually building a big spoken communication exemplar for code , ” he told TechCrunch . “ Then we realized that that will quickly become a commodity … Now we ’re building sham all those large speech exemplar are there . Assuming there ’s some really enough [ AI ] agentive role for codification , how do we take out the maximum business value from them ? ”
Join us at TechCrunch Sessions: AI
Exhibit at TechCrunch Sessions: AI
He state that the idea built on the team ’s understanding of how to analyze software program applications . “ compound that with large speech framework — then focalise into earthing and affirm what those large lyric models and the AI agent really evoke . ”
Test-driven development
What does that think of in practice ? Paskalev says LogicStar performs an analysis of each program that its tech is deployed on — using “ classical computer science method ” — in orderliness to build a “ knowledge root word . ” This devote its AI agent a comprehensive single-valued function of the software package ’s inputs and outputs ; how variables link to functions ; and any other linkages and dependencies , etc .
Then , for every bug it ’s presented with , the AI broker is able-bodied to make up one’s mind which parts of the app are touch on — allowing LogicStar to narrow down the social function needing to be simulated in parliamentary law to screen scads of possible fixes .
Per Paskalev , this “ minimized execution environment ” allows the AI agentive role to persist “ thousands ” of tests aimed at regurgitate bugs to identify a “ failing test , ” and — through this “ mental testing - driven development ” approach — ultimately put down on a fix that stick .
He confirm that the literal hemipteron fix are source from the LLMs . But because LogicStar ’s platform enables this “ very firm executive surround ” its AI agents can work on at musical scale to separate the wheat from the chaff , as it were , and wait on its users with a shortcut to the best that Master of Laws can extend .
“ What we see is [ Master of Laws are ] gravid for prototyping , screen thing , etc . , but it ’s absolutely not great for [ code ] production , commercial applications . I consider we ’re far from there , and this is what our political program delivers , ” he argued . “ To be able to take out those capabilities of the models today , we can actually safely extract commercial-grade note value and actually save fourth dimension for developers to really focus on the authoritative stuff and nonsense . ”
enterprise are set to be LogicStar ’s initial target . Its “ atomic number 14 agent ” are intended to be put to piece of work alongside corporate dev team , albeit at a fraction of the salary required to hire a human developer , handling a range of app maintenance labor and freeing up engineering talent for more creative and/or challenging work . ( Or , well , at least until LLMs and AI agentive role get a lot more capable . )
While the startup ’s pitch vaunt a “ fully autonomous ” app maintenance capableness , Paskalev confirms that the platform will give up human developers to review ( and otherwise oversee ) the fixes its AI agents call up . So trust can be — and must be — earned first .
“ The truth that a human developer return pasture between 80 to 90 % . Our goal [ for our AI agents ] is to be on the button there , ” he add together .
It ’s still other days for LogicStar : An alpha version of its technology is in testing with a number of unrevealed companies which Paskalev refers to as “ design partners . ” presently the tech only supports Python — but expansion to TypeScript , JavaScript , and Java are bill as “ add up soon . ”
“ The primary goal [ with the pre - seed backing ] is to actually show the technology works with our excogitation partner — rivet on Python , ” adds Paskalev . “ We already spent a yr on it , and we have lots of opportunity to actually expand . And that ’s why we ’re trying to focus it first , to show the value in one case . ”
The inauguration ’s pre - seed raise was lead by European VC firm Northzone , with angel investor from DeepMind , Fleet , Sequoia sentry , Snyk , and Spotify also joining the bout .
In a statement , Michiel Kotting , mate at Northzone , articulate : “ AI - driven code generation is still in its early stages , but the productivity gains we ’re already seeing are revolutionary . The potential for this engineering to streamline evolution processes , deoxidize costs , and quicken innovation is immense . and the squad ’s vast proficient expertise and demonstrate racecourse disc post them to deliver real , impactful results . The futurity of software development is being reshaped , and LogicStar will play a all-important role in package maintenance . ”
LogicStar is operating awaiting listfor potential customers want to express sake in getting early accession . It enjoin us a genus Beta tone ending is plan for by and by this year .
