Archive for November 2, 2026

Polygraf AI and Carahsoft Partner to Help Public Sector Organizations Adopt AI While Maintaining Control Over Sensitive Data, Operational Risk and Compliance Requirements

Posted in Commentary with tags , on November 2, 2026 by itnerd

Polygraf AI and Carahsoft Technology Corp. today announced a partnership. Under the agreement, Carahsoft will serve as Polygraf AI’s Master Government Aggregator®, making the company’s AI security and governance solutions available to the Public Sector through Carahsoft’s reseller partners and NASA Solutions for Enterprise-Wide Procurement (SEWP) V, The Interlocal Purchasing System (TIPS) and OMNIA Partners contracts.

Polygraf AI offers AI security, data governance and sensitive-information protection solutions designed to help organizations adopt AI while maintaining control over data, security and compliance. Its AI Behavioral Control (AiBC) Plane applies real-time policy enforcement across AI interactions, endpoints, documents, email, collaboration tools and virtual meetings, helping organizations identify and protect sensitive information before it is transmitted, shared or processed by AI systems. Polygraf AI’s solutions include controls for AI prompts and outputs, sensitive-data discovery and redaction, AI-use monitoring, audit logging and detection of AI-generated and synthetic content. The platform can be deployed on-premise, in private cloud, at the edge or in air-gapped environments to support organizations with stringent data-residency and operational-control requirements.

Polygraf AI provides Public Sector organizations with capabilities to support the controlled adoption of generative AI while helping protect sensitive Government information. Its solutions can help agencies govern the use of public and private AI models, safeguard personally identifiable and other regulated information, monitor AI activity, secure virtual meetings and identify and redact sensitive content in documents and files. Through these capabilities, Polygraf AI provides an AI-specific security and governance layer that can integrate with existing technology environments and help Public Sector organizations expand the use of AI while maintaining established requirements for data protection and oversight.

Polygraf’s AI security and governance solutions are available through Carahsoft’s SEWP V contracts NNG15SC03B and NNG15SC27B, TIPS Contract #220105 and OMNIA Partners Contract #R240303. For more information, contact the Carahsoft Team at (703) 871-8548 or Polygraf@carahsoft.com; or learn more about Polygraf’s solutions here.

Leave a comment »

The $6.3 Bill Binance ownership failure & 3 Qs CEOs must ask

Posted in Commentary with tags on November 2, 2026 by itnerd

Binance lost about $6.3 billion in a single burst, the largest crypto theft on record. The early finding points to a compromised engineer laptop that reached an internal service able to submit withdrawals, which means the transactions looked authorized at every checkpoint they passed.

The failure wasn’t a missing tool. It was that nobody owned the whole path from a developer machine to a withdrawal. So nobody tested it, fixed it, and tested it again.

Waseem Ahmed, Head of Engineering, Secure.com:

“The $6.3 Billion Binance hack Is a wake-up Call:  the traditional security model needs to change.

“Most of the money left in about 20 minutes. That is the detail security teams should sit with. Binance’s investigation is still open. Early reports pointed to a compromised DevOps machine, but the company has since said the attack involved additional breach points and was not simply one engineer’s permissions. The pattern is still familiar: an action can be technically authorised and still not be legitimate, and monitoring alone will not tell you the difference in time.

“Three questions worth asking this week. What could our most privileged developer identity reach if it were stolen? Would we see that use in minutes, not hours? Is there a human approval step before an irreversible transfer?

“The organisations that close this gap test their own attack paths continuously, and they decide in advance what gets contained automatically and what needs a person.”

Hopefully organizations wake up to the fact that to be fully protected from hacks, everything must be considered. Otherwise you get stories like this.

Leave a comment »

LinearB launches Agentic Studio to give enterprise engineering leaders and coding agents the context to ship better software faster

Posted in Commentary with tags on November 2, 2026 by itnerd

LinearB today launched Agentic Studio, a key part of the next generation of developer productivity platforms.

Historically, productivity platforms gave engineering leaders metrics. LinearB now turns everything happening across software delivery into context that people and agents can consume and act on. Engineering leaders ask questions in plain language and get answers and next steps. They define the context surface with role-based access built in, and AI assistants help extend that context and build metrics, artifacts and dashboards. Coding agents receive that context, so they understand their task and environment better and make smarter decisions.

Acting on that context matters more than ever. LinearB’s 2026 AI benchmarks, covering 2.7 million pull requests across more than 250 organizations, show that developers who use AI most heavily nearly doubled their merged pull requests year over year. Yet the software factory faces new problems. In LinearB’s 2026 Software Engineering Benchmarks Report, AI-assisted pull requests were 2.6 times larger than unassisted ones at the 75th percentile, and agent pull requests waited more than five times longer for review.

LinearB’s effort skill, now in early access, matches a coding agent’s effort to the task. It gauges a change’s business relevance, complexity and risk using the code quality in the area, recent service incidents and ownership history. Low risk changes use less effort and fewer tokens. Complex, risky or high priority changes get more effort and careful, tested edits. More skills are on the way, focused on accelerating downstream stages of the software factory, such as review and merge.

Kraken used LinearB’s MCP server with AI to find what was really slowing its code reviews. The answer wasn’t large changes. It was time zones and reviewer availability. Kraken cut review time at the 90th percentile by more than half in four months.

The Agentic Studio is available now to all LinearB customers. Agent skills are available in early access.

Why it matters now

Engineering leaders have more data than ever and less time to use it. Boards and CFOs now ask whether AI spend is paying off, and a dashboard full of numbers doesn’t answer that. At the same time, coding agents write more of the code, and they start every task knowing nothing about the codebase’s history.

Productivity platforms have always been good at measuring. The next step is turning what they measure into action. That means answers for the people running engineering, and context for the agents doing more of the work.

What LinearB does

LinearB collects data from across the software delivery lifecycle and makes it usable. It normalizes data from different Git and project management systems into one model, matches each engineer’s identities across tools, and links every change from ticket to release. It adds AI attribution and industry benchmarks, so leaders can tell whether a number is good or bad.

The engineering context layer delivers that knowledge to people and to agents.

Engineering leaders ask questions in plain language, like where delivery is stuck, how their developers are doing and whether AI is paying off. Instead of reading a chart, they get an answer and a next step. They can build their own metrics and pull the same context into AI tools like Claude through MCP.

Coding agents get the same context through MCP and agent skills. The first skill, now in early access, matches an agent’s effort to the task. It checks how complex and risky a change is, using the code’s rework rate, recent incidents and ownership history. Simple, low risk changes use less effort and fewer tokens. Complex or risky changes get more effort and careful, tested edits, which raises quality where it matters most. The skill works with any coding agent, and leaders can see how often agents were routed to low, medium and high effort.

The platform is built for the enterprise, with role based access and guardrails on what agents can do.

Journey and milestones

LinearB started by giving engineering leaders visibility into how their teams deliver software. Years of integration work across the software delivery lifecycle gave it one of the largest datasets on how engineering teams work. Its 2026 Software Engineering Benchmarks Report drew on more than 8.1 million pull requests from 4,800 engineering teams in 42 countries. As AI coding tools spread, LinearB added AI attribution and connected to more of the tools teams use.

In 2026, Gartner named LinearB a Leader in its first Magic Quadrant for Developer Productivity Insight Platforms, for both its ability to execute and its completeness of vision. Early customer response to the context layer is strong. The AI assistant and MCP access show the highest user retention of any features LinearB has released.

Market trends

LinearB’s 2026 AI benchmarks, built from 2.7 million pull requests and 83,000 developers across 253 engineering organizations between February and May 2026, show why measuring AI isn’t enough on its own. The teams pulling ahead are the ones acting on what they learn.

  • Heavy AI users are pulling away. The typical developer who uses AI on at least 75% of coding days merged 96% more pull requests per week than a year earlier. The typical developer using no AI merged 3.6% fewer.
  • Adoption is shallower than the hype suggests. Even at elite organizations, AI assists only 54% of pull requests.
  • Agents still need people. At the 90th percentile, autonomous agents open 4.7% of pull requests. At elite organizations, 79% of agent pull requests merge within 30 days, compared with 92% of human only pull requests. At fair organizations, agent yield drops to 37%. The likely reason is ownership. When no developer owns an agent’s pull request, it often never merges.
  • AI code review is the fastest win. Pull requests with an AI review merge at the highest rate of any type and lift yield by up to five percentage points.
  • AI costs are climbing fast at the top. Token spend at the 90th percentile reached $35.20 per developer per coding day, up nearly 75% in three months, while the median barely moved.

Engineering leaders already use LinearB to answer the questions their boards are asking about AI.

Forward-looking vision

The effort skill is the first in a library of agent skills. Each one turns LinearB’s context into an action an agent or a leader can take. Next, LinearB will focus on helping teams merge and govern AI generated code safely, and on cutting what engineering organizations spend on AI. The goal is the one productivity platforms have always promised, helping teams deliver better software faster. LinearB now does it by giving people answers and giving agents the context to act.

Leave a comment »

Datadobi Receives SOC 2 Type I Attestation from KPMG for Its Software Engineering, Quality, and Security Processes

Posted in Commentary with tags on November 2, 2026 by itnerd

Datadobi today announced that it has once again received its Service Organization Control (SOC) 2 Type I attestation covering its software engineering, quality assurance, security, operations, and support processes, following an independent audit by KPMG. Developed by the American Institute of Certified Public Accountants (AICPA), SOC 2 is a widely recognized framework for assessing how service organizations protect customer data. The attestation confirms that Datadobi’s internal controls and processes continue to meet the highest standards for the integrity, security, and confidentiality of customer data.

With unstructured data growing exponentially in recent years, the challenge of managing it is accelerating at an ever-increasing pace. IT and business organizations are moving data for system upgrades, to take advantage of the cloud, for cost or compliance reasons, and to prepare data for AI, analytics, and research. The race to get unstructured data ready for AI raises the stakes further: data feeding AI models, retrieval pipelines, and agents must be complete, uncorrupted, and properly governed, because errors or exposures at this stage are amplified in every output that follows. Throughout these processes, data integrity must be protected against human and machine error as well as malicious attacks. That puts the spotlight not only on the software organizations rely on to manage their data, but on the engineering, security, and operational discipline of the vendor behind it.

The SOC 2 Type I report reaffirms Datadobi’s company-wide commitment to data integrity, security, and software quality, giving organizations confidence that the data they move, manage, and prepare for AI is handled under rigorous, independently audited controls. It provides an in-depth look at the internal controls and processes that govern how Datadobi builds, tests, releases, and supports its software, including:

Secure, Defensive Software Engineering

A defensive engineering methodology is embedded across Datadobi’s software development lifecycle. All code is deployed following a four-eyes principle, and every code change is peer reviewed and passes extensive automated testing before it is included in a release. An agile release methodology means Datadobi’s software can be adapted quickly to customer-specific needs or interoperability challenges, without compromising quality or control.

Independent Quality Assurance (QA) Function

Datadobi maintains a QA function that operates independently of development to verify the quality of every release. If an issue is identified in a customer environment, it is investigated, and a software patch is provided to resolve it. A root cause analysis then determines whether any process, test, or architectural change is needed, so that lessons learned feed back into how all Datadobi software is built.

Risk Assessment and Governance

Datadobi applies a formal risk-control matrix across the business to evaluate risks and the controls in place to mitigate them. The matrix is regularly reviewed by Datadobi’s management and product management teams, ensuring security and data protection remain leadership priorities embedded in day-to-day operations.

Training, Support, and Customer Guidance

Datadobi trains customers and partners to use its software effectively in complex, multi-vendor environments and supports them at every stage of a project, from planning and pre-cutover testing to dry runs, cutover, and verification. Customers also have access to an extensive knowledge base of storage interoperability guidance and best practices, along with chain-of-custody reporting that documents how their data has been handled.

To learn more about Datadobi’s service commitments and obtain the available SOC 2 Type I report, visit https://datadobi.com/kpmg/

Leave a comment »