Connect with us

Technology

Sondera Compiles Natural-Language Rules into Provable Control Over AI Agent Actions

Published

on

Sondera’s autoformalization research was accepted at ICML 2026’s Agents in the Wild workshop and at FLoC 2026’s LLM-Solve workshop, with a related tool demo at Black Hat Arsenal. Sondera transforms natural-language policy into formally verified rules that run on any agent.

NEW YORK, June 30, 2026 /PRNewswire/ — Every organization runs on rules, security, compliance, and business logic written in natural language. As AI agents take on real work, those rules have to bind them too. Sondera today announced that its research on compiling natural-language policy into formally verified controls for AI agents has been accepted at workshops at ICML 2026 and at the Federated Logic Conference (FLoC) 2026’s LLM-Solve workshop, and selected for a tool demonstration at Black Hat Arsenal.

The paper, “Autoformalization of Agent Instructions into Policy-as-Code,” by Sondera’s Adam Mondl, Matthew Maisel, and John Brock, was accepted at the Second Workshop on Agents in the Wild at ICML 2026, one of the field’s leading machine learning conferences, and LLMSolve, a workshop at FLoC on applying large language models to formal and solver-based reasoning. A related tool, “GolemHalt: A Deterministic Reference Monitor for AI Coding Agents,” will be demonstrated at Black Hat Arsenal, Black Hat’s curated showcase of open-source security tools.

In peer-reviewed research using MedAgentBench, an independent benchmark for clinical AI agents published in NEJM AI, Sondera’s pipeline autoformalized more of the 88-rule policy than prior published work had hand-coded (23 of 88), and the resulting rules blocked every adversarial unsafe write attempt (99 of 99).

Autoformalization is critical for enforcing agent policies because real policy and business logic live in natural language — in documents like a HIPAA manual, FINRA guidelines, a standard operating procedure, or an agent’s own instructions. Traditionally, turning natural-language policies into something a machine can enforce has meant hand-coding rules one at a time, which does not scale to how fast agents are being deployed.

Sondera’s pipeline reads the natural language and compiles it directly into formally verified Cedar policy-as-code, with a theorem prover checking every rule and adversarial simulation stress-testing it before production, both to find edge cases and to confirm that legitimate work is still permitted. The approach is neurosymbolic: neural classifiers such as LLMs-as-judges evaluate what an agent is doing probabilistically, while symbolic rules decide deterministically what the agent is allowed to do.

At runtime, a verified deterministic rule returns a decision (such as allow, deny, or escalate) for each agent action. Because the policy-as-code is enforced deterministically outside the context window, rather than relying on an AI-as-judge, issues such as prompt injection, emergent behavior, and model drift cannot talk the policy enforcement layer out of its rules. And because enforcement is stateful, tracking the agent’s full trajectory and the flow of information across it, the same action can be allowed or denied depending on what the agent did earlier in the run.

“The agent incidents we see today aren’t from prompt injection and hijacking. They’re from authorized humans asking authorized agents to do legitimate tasks, like analyzing a financial file or configuring a server. Along the way, the agent reaches the goal with unintended behavior, like leaking or destroying data,” said Josh Devon, co-founder and CEO of Sondera. “Even beyond that, enterprises are struggling to apply business logic at scale to their agents, rules that aren’t security or compliance but just standard operating procedure, like a coding agent that should stand up a server on the company’s approved cloud account rather than an unapproved vendor. Our research is focused on letting organizations turn on the most powerful, long-running agents possible while having the confidence that they will follow the rules.”

Sondera’s policy and agent control plane is currently in private beta. Teams interested in participating can visit sondera.ai.

The full paper “Autoformalization of Agent Instructions into Policy-as-Code” is available at arxiv.org/abs/2606.26649.

Sondera’s open-source harness and SDKs are at github.com/sondera-ai.

About Sondera

Sondera gives enterprises provable control over AI agents, in natural language. Its approach is neurosymbolic: neural classifiers detect what an agent is doing, symbolic rules decide what it is allowed to do. Through a process called autoformalization, Sondera compiles an organization’s natural-language rules into formally verified Cedar policy-as-code, checked by a theorem prover and enforced on every action an agent takes, on any agent runtime or harness. Because symbolic enforcement runs outside the model and across the full run, prompt injection and drift can’t bypass it, and the same action can be allowed or denied based on what the agent did before it.

Teams use Sondera to apply complex business logic, security, and compliance rules to coding agents, the agents they build, and the agents they sell. Open-source harness at github.com/sondera-ai. Learn more at sondera.ai.

Contacts
Media Contact
pr@sondera.ai

View original content to download multimedia:https://www.prnewswire.com/news-releases/sondera-compiles-natural-language-rules-into-provable-control-over-ai-agent-actions-302814775.html

SOURCE Sondera, Inc.

Continue Reading
Click to comment

Leave a Reply

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

Technology

Mobile Communications America Introduces Managed Cellular DAS Solution for Hotels Impacted by Wireless Carrier Transition Notices

Published

on

By

MCA’s SecurePlan® + Signal Source combines monitoring, operational support, and new signal source management to help hospitality organizations maintain reliable cellular connectivity.

SPARTANBURG, S.C., June 30, 2026 /PRNewswire/ — Mobile Communications America, Inc. (MCA), a national leader in integrated communication, connectivity, and security solutions, today announced an expanded managed services offering that combines Signal Source as a Service (SSaaS) with its SecurePlan® support program to help hospitality organizations maintain reliable in-building cellular coverage as carrier support models continue to evolve across the country.

As wireless carriers increasingly shift away from supporting legacy 4G distributed antenna system (DAS) deployments, many hotels, resorts, convention centers, and other venues are left responsible for critical wireless infrastructure previously supported through carrier-led programs. Without a clear path forward, organizations risk loss of coverage, which impacts the guest experience, staff productivity, and day-to-day operations.

Hospitality leaders seeking to understand the impact of these industry changes and explore available options can learn more at https://info.callmc.com/hospitality-das.

To help organizations navigate these challenges, MCA combined SecurePlan®, a best-in-class managed service to keep existing DAS operational and optimized, with Signal Source-as-a-Service, delivering a fully managed approach to existing in-building wireless infrastructure, helping organizations maintain reliable cellular connectivity through:

Signal source management and ongoing system oversightProactive monitoring and issue resolutionMaintenance, spares, repairs, and operational support from wireless expertsReduced burden on internal IT and facilities teamsA predictable managed services model designed for long-term reliability

“As organizations navigate changing carrier support models, they need a dependable path forward,” said John Bramfeld, Sr. Director of In-Building Wireless Solutions at MCA. “SecurePlan® + Signal Source provides the management, monitoring, and support required to maintain reliable connectivity, protect existing DAS investments, and extend the useful life of critical in-building cellular infrastructure. Losing carrier signal sources isn’t the end of the world, but it can create a functionality gap in your DAS if not addressed quickly.”

Under the SecurePlan® + Signal Source model, MCA helps hospitality organizations respond to the loss of carrier support by providing a managed path for signal-source continuity, system monitoring, technical support, and lifecycle planning. Rather than requiring hotel owners and operators to manage complex DAS infrastructure on their own, MCA’s technical teams continuously monitor performance, support ongoing operations, and help organizations protect their investment and the remaining useful life of their existing DAS.   

Reliable cellular connectivity has become an essential component of the modern guest experience, supporting everything from mobile check-in and digital room access to staff communications, public safety, and day-to-day operations. As demand for seamless connectivity continues to grow, organizations are increasingly seeking partner-led solutions that provide greater visibility, flexibility, and operational control.

SecurePlan® + Signal Source reflects MCA’s continued investment in helping customers navigate complex connectivity challenges through managed services, technical expertise, and lifecycle support. By combining industry-leading network management solutions with nationwide service capabilities, MCA helps hospitality organizations maintain the connectivity experiences their operations and guests expect.

About Mobile Communications America

Mobile Communications America (MCA) is a national leader in wireless communication, data, and security solutions. MCA serves more than 65,000 customers nationwide across public safety, commercial, education, healthcare, utilities, and government sectors. From two-way radios and network infrastructure to access control and video surveillance, MCA delivers comprehensive systems that ensure safety and operational efficiency. Learn more at www.callmc.com.

View original content to download multimedia:https://www.prnewswire.com/news-releases/mobile-communications-america-introduces-managed-cellular-das-solution-for-hotels-impacted-by-wireless-carrier-transition-notices-302814839.html

SOURCE Mobile Communications America

Continue Reading

Technology

Atera Extends Enterprise Lead with Top G2 Rankings

Published

on

By

Ranked No. 1 across 15 G2 Summer 2026 reports, Atera outperforms platforms including ServiceNow, Salesforce, and Microsoft Copilot as enterprises increasingly prioritize outcome-driven AI

NEW YORK, June 30, 2026 /PRNewswire/ — Atera, the pioneer of the Autonomous IT platform for the modern enterprise, today announced a series of enterprise milestones with No. 1 rankings across 15 G2 Summer 2026 Index reports spanning five categories and evaluating customer preferences across Implementation, Usability, and Results. The G2 rankings place Atera ahead of other enterprise platforms such as ServiceNow, Salesforce Agentforce, Microsoft Copilot, and Jira, reflecting growing enterprise demand for its patented Autonomous IT agent, Robin.

Atera took the top spot in multiple Enterprise categories, including AI Agents, Agentic AI Software, AIOps Platforms, AI IT Agents, and Service Desk. Several of the reports evaluate highly competitive markets, with some covering more than 100 vendors. G2’s Index reports are widely used by enterprise software buyers. Atera’s No. 1 rankings highlight a shift in how organizations evaluate both AI-driven and enterprise IT platforms, with greater emphasis on customer-validated outcomes and verifiable ROI.

At the center of this momentum is Atera’s Autonomous IT agent, Robin, which ranked first across the three core AI Agents indices: Results, Implementation, and Usability. With scores of 9.55 in Implementation, 9.40 in Results, and 9.31 in Usability out of a possible 10, Atera achieved the highest overall performance in the category. Ranking No. 1 across all three indices demonstrates consistent strength from initial deployment through measurable business outcomes.

“These milestones reflect a broader shift in how enterprise teams adopt and evaluate AI,” said Gil Pekelman, co-founder and CEO of Atera. “Organizations are moving beyond the theoretical promise of AI and focusing on measurable impact in practice. Robin was designed to remove the everyday technology friction that gets in the way of work and help people stay focused on higher-value work that moves the needle for their organizations. That’s what these rankings ultimately reflect.”

Consistent performance across the full lifecycle

G2’s AI Agent Index reports evaluate platforms across three critical phases of adoption:

Implementation Index: Speed and ease of deploymentResults Index: ROI and business impactUsability Index: Ease of use and administration

Atera’s No. 1 ranking across all three phases reflects strength across the full AI agent lifecycle as enterprise buyers increasingly evaluate AI agents based on end-to-end performance rather than isolated capabilities.  

Autonomous execution that delivers results

Atera’s patented AI agent, Robin, is designed to detect, diagnose, and resolve technical issues end-to-end, without a technician needed in the loop. Unlike platforms that primarily route tickets or surface recommendations, Robin takes action to resolve IT incidents autonomously across devices, servers, mainframes, and networks, enabling IT teams to focus on higher-value work while employees continue working uninterrupted.

About Atera

Atera is defining the category of Autonomous IT, in which AI agents resolve technical incidents end-to-end, without a technician in the loop. Robin by Atera is the world’s only AI technician that detects, diagnoses, remediates, and verifies incidents by taking action directly on devices, servers, networks, and mainframes—resolving technical Tier 1 and complex Tier 2 incidents in full.

Built on a patented architecture, Robin operates as a system of action, working within existing enterprise environments and adhering to the highest enterprise-grade security and compliance standards. It is governed by configurable guardrails, complete audit trails, and approval workflows.

More than 13,000 organizations across 120+ countries rely on Atera to keep their people working, uninterrupted. To learn more, visit atera.com.

View original content to download multimedia:https://www.prnewswire.com/news-releases/atera-extends-enterprise-lead-with-top-g2-rankings-302814848.html

SOURCE Atera

Continue Reading

Technology

Basecamp Research brings EDEN’s antibiotic and vaccine design models to Claude Science

Published

on

By

Scientists can now design potent antibiotics and rapidly prioritise vaccine targets through Claude Science, thanks to integration with Basecamp Research’s EDEN models.

LONDON and CAMBRIDGE, Mass., June 30, 2026 /PRNewswire/ — Basecamp Research today announced that its antibiotic design and vaccine target prediction EDEN models are now available through Claude, including Claude Science, Anthropic’s AI workbench for life sciences research. This allows researchers to generate and prioritise therapeutic candidates through a conversational interface in a matter of minutes.

By combining Claude’s reasoning with EDEN’s biological design capabilities, researchers can now go straight from a target to a shortlist of high-performing antibiotic or vaccine candidates. This capability is available in Claude.ai, Claude Desktop, Claude Mobile, Claude Code, Cowork, and Claude Science through Anthropic’s connectors directory.

The world needs new antibiotics

Drug-resistant infections play a role in nearly 5 million deaths per year, but the pharmaceutical industry has largely retreated from antibiotic development. New antibiotics are badly needed, particularly for the pathogens spreading fastest in lower-income countries where last-resort drugs are hardest to access.

“Microbes have been producing antibiotics and evolving resistance to each other for billions of years,” said Glen Gowers, Co-founder and CEO of Basecamp Research. “EDEN learned from that history, and now, through Claude, researchers all over the world can design successful new antibiotics in minutes, not years.”

In collaboration with University of Pennsylvania researchers, Basecamp Research demonstrated that 97% of the antibiotic peptides designed by EDEN are active against World Health Organisation (WHO)  priority pathogens when tested in the lab. Fleming Prize winner and Presidential Associate Professor César de la Fuente led the work by UPenn’s Machine Biology Group.

One candidate, EDEN-7, was tested in mice infected with multidrug-resistant Acinetobacter baumannii – a pathogen associated with hospital outbreaks worldwide – and showed efficacy in the same range as a last-line antibiotic, despite being generated zero-shot, meaning the model produced it without subsequent optimization or iterative engineering.

“This collaboration shows how frontier biological foundation models can be paired with rigorous experimental validation to accelerate antibiotic discovery,” de la Fuente said. “Antimicrobial resistance is one of the greatest existential threats facing humanity and collaborations like this between academia and industry are critical.”

Finding vaccine targets in minutes

Developing a vaccine against an emerging pathogen is a race against time. Which part of the pathogen to target is often determined empirically, which can take months of laboratory work.  This delay often costs lives.

EDEN’s vaccine design model identifies which proteins are most likely to trigger a protective immune response, outperforming comparable genomic foundation models. By integrating it into Claude, researchers can describe a problem in plain language and have Claude run a prioritisation workflow against the pathogen’s genetic sequence. This can reduce several weeks of research per pathogen into a single conversation.

A growing collaboration

“The antibiotic crisis and the need for new vaccines are two of the most important public health challenges of our time,” said Jonah Cool, Head of Life Sciences Partnerships and Deployment at Anthropic. “Making EDEN available through Claude Science gives researchers a new way to explore and prioritise treatments for some of the most dangerous pathogens on Earth.”

Built on the world’s largest biological dataset

Most biological AI models are trained on a narrow set of well-studied organisms – the ones scientists have already catalogued. In contrast, EDEN is trained on BaseData, the largest, fastest-growing and most information-rich biological database on Earth.

To build it, Basecamp Research has run expeditions to over 200 locations across more than 30 countries, sampling the places life is strangest and least understood, including thermal springs, deep-sea sediment, polar ice, remote high-altitude plateaus. In the process, it has documented more than a million species new to science. The result is over 10 billion new genes and roughly ten times the content of every public database combined.

Basecamp Research aims to scale BaseData 100-fold over the next two years through the Trillion Gene Atlas, a partnership with Anthropic, NVIDIA, PacBio and Ultima Genomics designed to generate genomic data at the trillion-gene scale for AI-driven drug discovery.

Diversity is what drives EDEN’s performance across a wide range of tasks. Every sample is collected under informed-consent and benefit-sharing agreements so that the countries and communities who steward this biodiversity share in the value it creates, with each sequence traceable to one of hundreds of country-specific permits. This allows a portion of revenue to be fed back to the country and community where the data was originally sourced, setting a standard of data provenance that the rest of the field has yet to match.

About Basecamp Research
Basecamp Research is dedicated to solving major challenges in healthcare and life sciences by exploring Beyond Known Biology™. The company trains frontier AI models on BaseData, the world’s largest biological dataset, collected through partnerships with more than 200 organisations across more than 30 countries. Basecamp Research is developing a pipeline of therapeutics and works with commercial and academic partners worldwide to accelerate therapeutic discovery and development.

BaseData™, Beyond Known Biology™, EDEN-GLM™ and aiPGI™ are trademarks and technologies of Basecamp Research.

Logo – https://mma.prnewswire.com/media/3002092/Basecamp_Research_Logo.jpg

View original content to download multimedia:https://www.prnewswire.com/news-releases/basecamp-research-brings-edens-antibiotic-and-vaccine-design-models-to-claude-science-302814440.html

SOURCE Basecamp Research

Continue Reading

Trending