←
Centific1,261 posts
Centific avatar

Centific

@CentificGlobal

Building good AI.

◷ Joined January 2020▧ 606 Photos and videos
Centific Applied AI Research avatar
⟳ Centific retweeted
Centific Applied AI Research@Centific_CAIR·Nov 17

One speech-to-speech model got a speaker’s gender wrong 2% of the time when voice and topic matched a stereotype. When they clashed, it got it wrong 90% of the time. We tested five models on clips that paired a male or female voice with a stereotypically male, female, or neutral topic. Their output voices stayed consistent. How they referred to the speaker didn’t. One model called a woman “he” after she talked about rebuilding an engine. If you build or buy these models, test cases where voice and topic disagree. #NeurIPS2026 workshop “Can We Trust AI Evaluation?”, Sydney, Dec 12. arxiv.org/abs/2609.09263 Co-authored by @harshitr (Centific) and Tanu Mitra (University of Washington)

Centific Applied AI Research avatar
⟳ Centific retweeted
Centific Applied AI Research@Centific_CAIR·Nov 12

Centific Applied AI Research is presenting eight papers at #NeurIPS2026 workshops this year. Each one asks a version of the same question: how do we know if AI is ready for the work we ask it to do? In one study, the strongest model got the final answer right on about 80% of healthcare tasks. Its complete deliverables met the professional standard on just 18%. A good score can hide the failures that decide whether people can rely on AI for the whole job. Read our new article by Shamsheer Rahiman (@5ham5h33r), Aastha Bisen (@DesignAastha) and Sasa Phanitsombat (@KaoSasakorn) to see what the eight studies found and how to design evaluations that catch the mistakes that matter: centific.com/blog/ai-model-e…

Centific avatar
Centific@CentificGlobal·Nov 9

We’re excited to share that we will present research at the Human-Centered Robot Learning and Interaction (HumanRLI) workshop at #CoRL2026. Our study asks what happens when several people teach a robot by correcting it as it works, and one person’s influence later needs to be removed. If someone teaches the robot an unsafe behavior, can you undo what that person taught it while preserving what it learned from everyone else? If you’re attending CoRL, join us at HumanRLI on November 12 from 2:00 to 6:00 p.m. CST in Room 202 at the JW Marriott Austin.

Post attachment
Centific Applied AI Research avatar
⟳ Centific retweeted
Centific Applied AI Research@Centific_CAIR·Oct 21

Centific Applied AI Research was at #BayLearn2026 at Santa Clara University. Great to see @KarthikCentific present our work alongside the Bay Area ML community.

Had a great time presenting at #BayLearn2026 at Santa Clara University, representing Centific Research! Inspiring keynotes from Andrew Ng, Ion Stoica and Azalia Mirhoseini, a great AI for Science panel, and exciting work from the Bay Area ML community. Thanks to my co-authors!

Post attachmentPost attachmentPost attachmentPost attachment
Centific avatar
Centific@CentificGlobal·Oct 19

You are not short of AI. You are short of certainty. At @LocWorld 56, we will reveal the next evolution of Flow. AI that puts your enterprise knowledge to work. Human judgment and trust built in. Book a meeting with our team. We look forward to seeing you! multilingualai.centific.com/… #LocWorld56 #TrustByDesign #Centific

Swetha | Reinforce Labs avatar
⟳ Centific retweeted
Swetha | Reinforce Labs@SwethaSunkara·Oct 13

Last week during #COLM2026, #ReinforceLabs and @CentificGlobal hosted a fireside chat and a panel on agent safety in production — with leaders from @Google , @Meta , @glean , @Salesforce, and Across AI. Our founder and CEO, @AnishDasSarma led the fireside chat that covered everything from model safety to agentic AI. The conversation moved through the biggest safety challenges facing the industry today, how to design evals that actually catch issues before production, and how teams stay current in a field that moves this fast. There was no shortage of tough questions and good debate - exactly the kind of evening we wanted to host. #AISafety #AIEvaluation #COLM2026 #ReinforceLabs #AIAgents

Post attachment
Centific Applied AI Research avatar
⟳ Centific retweeted
Centific Applied AI Research@Centific_CAIR·Oct 8

Two Centific papers have been accepted at #BayLearn2026! 🎉 The Centific Applied AI Research team will be presenting at BayLearn, the Bay Area Machine Learning Symposium, on October 15 at Santa Clara University. 📄 GUIDE: Governed Unified Intelligence for Document-to-Artifact Generation in Enterprise Settings "A governed multi-agent framework that turns complex enterprise guidelines (text, tables, images) into validated rules and persona-tailored artifacts. On 120 real-world documents: 96% document success, with turnaround cut from 2–3 days to 40–125 minutes." 🔗 arxiv.org/abs/2608.12133 📄 GVD: Governed Versioning and Deduplication for Document Repositories "Document repositories evolve continuously as guidelines and policies are revised, superseded and re-uploaded. GVD groups documents into version families and checks their rules for duplicates, contradictions, refinements and new content. It reaches an F1 of 0.97 for version-family construction and 0.94 for rule-level consistency." 🔗 arxiv.org/abs/2609.17696 ✍️ Authors: @mrsd89587483 , @shivalidalmia10 , @Sharma28Sumukha , @KarthikCentific and @mukherjiab 🎤 At BayLearn? Come meet Sumukha Thoppanahalli, Abhishek Mukherji and Srinivasa Karthikeya Reddy Kovvuri. They’d love to chat about the work! Congratulations to the team 👏 #MachineLearning #AIResearch #EnterpriseAI

Post attachment
Centific Applied AI Research avatar
⟳ Centific retweeted
Centific Applied AI Research@Centific_CAIR·Oct 12

Most egocentric robot training data fails quality checks. A smaller set of verified, deeply labeled human demonstrations can deliver more training value than a larger one. That's the idea behind Humanoid IQ, Centific's platform for training humanoids, vision-language-action models and world action models. Its curated task library includes verified successes, failures and recoveries, so robotics teams can train on data they can trust. Read the new blog by @krishnacentific of the Centific Physical AI team: centific.com/blog/egocentric… #PhysicalAI #Centific

Centific Applied AI Research avatar
⟳ Centific retweeted
Centific Applied AI Research@Centific_CAIR·Oct 12

COLM was a week of learning, sharing, and connecting with the AI research community! On October 8, Prasanna Desikan (@DesikanPrasanna), Head of Healthcare AI at Centific, and Anish Das Sharma (@AnishDasSarma), CEO of Reinforce Labs, hosted a conversation about what it takes to build AI agents people can trust. We’re grateful to the leaders who joined us: Vinay Rao, VP, Google; Nilou Salehi (@nilou_salehi), CEO & Co-founder, Across AI; Panagiotis Papadimitriou (@ppapadim), Senior Director, Meta; Nilesh Dalvi , Engineering Leadership, Glean; and Manjeet Singh (@CoachManjeet), Senior Director, Salesforce. At a COLM workshop, we also presented new research on agent behavior, exploring why agent marketplaces should be evaluated as behavioral systems rather than simply ranked by performance scores. Read the research here by Shivali Dalmia (@shivalidalmia10), Ashi Jain (@jainashi03) and Abhishek Mukherji (@mukherjiab): openreview.net/pdf?id=OoYMkv… Throughout the week, we exchanged ideas with researchers across disciplines about how to better understand, improve, and critically evaluate AI agents. Thank you to our team, collaborators, and everyone who made COLM such a valuable experience! #COLM2026 #Centific

Post attachmentPost attachmentPost attachmentPost attachment
Centific Applied AI Research avatar
⟳ Centific retweeted
Centific Applied AI Research@Centific_CAIR·Oct 2

Opus 5.5 ranks first on our enterprise sales benchmark, and it's strongest where the work is hardest: Revenue Ops. It's also unusually reliable. Across 330 runs, it went in circles once and never left an error unhandled. 10 models. 33 tasks. Replicas of enterprise applications. A ranking hides where a model is actually strong. We report by role and by task. Benchmark your own enterprise workflows: centific.com/solution/rl-env…

Post attachment
Centific Applied AI Research avatar
⟳ Centific retweeted
Centific Applied AI Research@Centific_CAIR·Sep 29

During COLM week, @Centific and Reinforce Labs are bringing researchers, builders, and product leaders into one room to talk about what it takes to build agents people can trust. Fireside: Vinay Rao (VP, @Google) × Anish Das Sarma (CEO, Reinforce Labs) Panel: - Nilou Salehi: CEO & Co-founder, Across AI - Panagiotis Papadimitriou: Sr. Director, @Meta - Nilesh Dalvi: Engineering Leadership, @glean - Manjeet Singh: Sr. Director, @salesforce - Moderated by Prasanna Desikan, Head of Healthcare AI, Centific Evaluation before deployment, guardrails that hold in production, accountability when an agent goes off-script, and what it takes to earn trust. 🗓 Oct 8, 2026 · 6:00–9:00 PM 📍 San Francisco (address after you register) Capped at 50. RSVP early: luma.com/hv3vd3aa #COLM2026