Topics
Latest
AI
Amazon
Image Credits:Tom Merton(opens in a new window)/ Getty Images
Apps
Biotech & Health
Climate
Image Credits:Tom Merton(opens in a new window)/ Getty Images
Cloud Computing
Commerce
Crypto
Enterprise
EVs
Fintech
Fundraising
appliance
Gaming
Government & Policy
computer hardware
layoff
Media & Entertainment
Meta
Microsoft
Privacy
Robotics
Security
Social
Space
Startups
TikTok
fare
Venture
More from TechCrunch
Events
Startup Battlefield
StrictlyVC
Podcasts
Videos
Partner Content
TechCrunch Brand Studio
Crunchboard
meet Us
H , the Paris inauguration establish by Google alums , made a expectant splash last summertime when , out of the blue , it announced aseed round of $ 220 millionbefore free a single product . Three month later , still without a product , that splash commence to look like a ruinous rising tide when three of the companionship ’s five co - founder left over “ operational and business dissonance . ”
But the company has kept drown , and today it ’s announce its first Cartesian product : Runner H , an “ agentic ” AI place at business and developers across task like character pledge and process automation . It ’s work up atop the startup ’s own , proprietary “ compact ” LLM base on just 2 billion parameters .
H has set up a waitlist for Runner H on its site . CEO Charles Kantor said that it will be releasing APIs to those on the list over the coming days to use agents “ off the ledge ” that have been pre - built by H , as well as for developers to create their own . Access to the API will also follow along with access to something anticipate H - Studio to essay and deal how these services turn .
Initially , using those APIs will be free , and later there will be a payment exemplar introduced .
Even using compendious LLMs , construction and running AI is not flashy , especially as competition keep to raise money to develop their own production . TechCrunch has also confirmed that H is erect a Series A to build what Kantor trace as part of the second era of AI — with LLM companies like OpenAI being part of the first earned run average .
“ We are lucky to be in the position of building our own model , ” Kantor said . “ But this second era will be as capital intensive as the first one . ”
( think the $ 230 million H has already raise — it appear to have sum up another $ 10 million since announcing it earlier this class — was a commixture of fairness and translatable debt . The long list of investor in that round include individuals like Eric Schmidt , Yuri Milner , and Xavier Niel ; VCs such as Accel and Creandum ; and strategical backers like Amazon , Samsung , and UiPath . )
Join us at TechCrunch Sessions: AI
Exhibit at TechCrunch Sessions: AI
Kantor told TechCrunch that H has softly been working with a handful of customers in areas like eastward - Commerce Department , banking , insurance , and outsourcing , who have been helping it perfect the Cartesian product .
“ Everything [ in H ] is not based on our creative thinking but customer feedback , ” he state .
Runner H will initially focus on three specific use cases : robotic process mechanisation ( RPA ) , quality assurance , and concern process outsourcing .
RPA is an area that has exist for year , using canonic script to automate the most repetitive tasks that world have had to execute — such as reading form , check out boxes , and send single file from one blank space to another . In fact , a lot of RPA has never been built with AI baked in , even after AI start to develop advanced skills . The musical theme with Runner H is that it will be able to unravel RPA across forms , sites , and other templates even when they have been modified ( something that might have broken former script ) , and across a much wider range of a function of author .
Quality assurance can cover a all-embracing range of applications , but Kantor said that one of the most popular so far has been reducing the “ maintenance burdens ” around site examination — validating pageboy availability , model real drug user actions , or ensuring compatibility across requital methods — in exceptional when modifications have been made .
BPO is a catch - all arena that will hide not just fixing and improving billing process , but also speeding up how an agent can use and access data from unlike reservoir , and more .
There has been a race among foundational AI company around how many parameters are going into LLM . ( GPT 4 for good example has 175 billion argument . ) But Runner H is taking a very different approach with just 2 billion parameters , both for its LLM and for its computer - vision - based “ VLM . ” Kantor ’s disceptation is that this lay down them importantly more effective in terminus of cost and operations , key when solve on winning and keeping business deals , and H ’s own operational costs .
“ We are specialists , ” he said . “ We are building for the agentic era . ”
The troupe also claims that it works : It says that its compact model outperforms Anthropic ’s “ Computer Use ” by 29 % ( based on WebVoyager benchmarks ) as well as models from Mistral and Meta .
