Nubank is a leading financial technology company in Latin America. Its first product, launched in 2014, is a no-fee credit card that is fully managed by a mobile app. Almost 30 million people have requested the product since launch, and the company has passed the 12 million customer mark.
InsightDLM is a comprehensive framework designed to help organizations develop and deploy domain-specific language models (SLMs and LLMs) tailored to industries such as insurance, retail, banking, healthcare, legal, and manufacturing. It offers an end-to-end pipeline that transforms raw enterprise data into fine-tuned, trustworthy language models within a secure environment.
Key Features and Functionality:
- Data Curation Pipelines: Ingests various data types—including documents, Q&A, glossaries, transcripts, and tickets—while performing parsing, deduplication, PII/PHI scrubbing, classification, and versioning with full lineage tracking.
- Synthesis and Labeling: Generates domain-specific Q&A pairs, instructions, reasoning traces, and adversarial examples using teacher-LLM distillation combined with human-in-the-loop labeling.
- Training and Evaluation: Fine-tunes models using reusable recipes such as SFT, LoRA/QLoRA, DPO/ORPO, and continued pretraining. Each candidate model is rigorously evaluated against domain-specific test suites and red-team probes.
- Deployment and Management: Supports model quantization (GGUF, AWQ, GPTQ), serving via platforms like vLLM, SGLang, and llama.cpp, and includes guardrail SLMs for safety. It also offers monitoring tools for drift, cost, and quality within a unified registry.
Primary Value and User Solutions:
InsightDLM addresses the challenge of creating reliable, domain-specific language models by providing a structured and efficient framework. It enables organizations to leverage their existing data to build models that understand industry-specific terminology and nuances, leading to more accurate and relevant AI applications. By offering tools for data curation, synthetic data generation, model training, evaluation, and deployment, InsightDLM streamlines the development process, reduces time-to-market, and ensures compliance with data privacy standards. This empowers businesses to enhance decision-making, improve customer interactions, and drive innovation within their respective sectors.
Autoscribe Informatics Matrix Gemini LIMS can be used in a wide variety of laboratories, both in size and application. Installations include systems with over 500 concurrent users and single user, single PC systems.
Runbear is a no/low-code integration tool that seamlessly connects custom LLM apps into channels like Slack or HubSpot. This enables customers to focus on building and utilizing AI apps while PlugBear handles its integration with the world. It supports various LLM builders, including OpenAI GPTs, Assistants, and LangChain.
Documind is an intelligent document processing platform that extracts structured data from various document types, transforming unstructured content into LLM-ready formats. It supports a wide range of file types, including PDF, DOCX, HTML, TXT, PNG, and JPG, and offers both open-source deployment and a fully hosted cloud version.
Key Features:
- Structured Data Extraction: Converts unstructured documents into structured JSON outputs based on customizable schemas.
- Format Conversion: Seamlessly transforms documents into Text and Markdown formats.
- Customizable Schemas: Allows users to define extraction schemas tailored to their specific needs, with pre-built templates for common schemas.
- LLM Compatibility: Compatible with OpenAI and custom LLM setups like Llava and Llama3.2-vision.
- Auto-Generated Schemas: Automatically generates schemas based on document content.
Primary Value:
Documind streamlines the extraction of structured data from diverse document formats, eliminating manual data entry and reducing errors. Its customizable schemas and compatibility with various LLMs make it a versatile solution for businesses seeking efficient document processing and data extraction.
Toloka Platform is a self-service data labeling and human-in-the-loop (HITL) infrastructure for AI teams building and evaluating large language models.
Built on the same foundation used by Frontier AI Labs and Big Tech like Microsoft and Amazon, Toloka Platform gives AI researchers, ML engineers, and technical teams immediate access to high-precision human feedback workflows—without sales calls, procurement delays, or enterprise minimums.
Core Capabilities:
AI-Assisted Pipeline Design
Describe your annotation goal in plain language ("I need to rank LLM outputs for safety"), and the built-in AI Assistant generates the complete workflow: annotator instructions, task interface, quality control logic (gold sets, overlap, consensus thresholds), and skill-based routing. What traditionally takes data ops teams days to configure now launches in minutes.
Access to Expert Annotators
Route tasks to a vetted workforce of 10,000+ contributors, including domain specialists in STEM, medicine, engineering, and technical fields, as well as PhD-level annotators for complex evaluation work. Continuous performance tracking ensures data quality at scale.
Production-Grade Quality Controls
Every annotation project includes built-in validation: 100% LLM-based QA coverage, customizable gold standard tests based on your own examples, inter-annotator agreement scoring, and real-time quality dashboards. RLHF and safety evals require ground truth you can trust to train on—Toloka Platform is designed for that precision.
Fully Self-Service Workflow
Sign up, upload unlabeled data (model outputs, prompts, or raw datasets), configure quality thresholds in the browser, and launch to annotators immediately. Results export as JSON. No API integration required. No vendor onboarding process. Pay per completed annotation with transparent, upfront pricing.
Who Uses Toloka Platform:
Teams working on reinforcement learning from human feedback (RLHF), constitutional AI, red-teaming evaluations, prompt engineering, model safety testing, and dataset generation for fine-tuning. Built for organizations that iterate on human feedback as rapidly as they iterate on code.
Toloka Platform eliminates the data tax—the operational overhead that typically delays model evaluation and alignment work. From 10-task pilots to 10,000-task production runs, the platform scales with your needs without requiring contract negotiations or dedicated data ops headcount.
Hermes 3 is a neutrally-aligned, generalist instruction-tuned language model designed to excel in reasoning and creative tasks. Its largest variant, Hermes 3 405B, achieves state-of-the-art performance among open-weight models on several public benchmarks.Hermes 3 addresses the need for a versatile and unbiased language model that excels in both reasoning and creative tasks. By providing state-of-the-art performance, it empowers users to tackle complex problems, generate innovative content, and interact effectively with various tools, thereby enhancing productivity and decision-making processes.
Backplain is a multi-model AI workspace built for teams where data privacy is non-negotiable. It gives your organization access to 47 frontier models across 9 providers — including GPT-4, Claude, Gemini, Llama, and more — all from a single governed interface, at flat-rate pricing with no token metering.
At the core is the patent-pending AI Firewall, which sits between every user and every model. It detects sensitive entities at prompt time and enforces your configured policy — Hard Block, User Warning, or Silent Reconstitute — so your data never reaches a model without authorization. Every session and every prompt is logged for full auditability.
Backplain ships on day one with SSO/SAML, SCIM provisioning, RAG file upload, an iOS mobile app, an admin console with policy controls, and a contractual guarantee that your data is never used for model training.
Whether you're in legal, finance, healthcare, or government, Backplain is the only AI workspace where your CISO doesn't have to choose between capability and control.Backplain is a multi-model AI workspace built for teams where data privacy is non-negotiable. It gives your organization access to 47 frontier models across 9 providers — including GPT-4, Claude, Gemini, Llama, and more — all from a single governed interface, at flat-rate pricing with no token metering.
At the core is the patent-pending AI Firewall, which sits between every user and every model. It detects sensitive entities at prompt time and enforces your configured policy — Hard Block, User Warning, or Silent Reconstitute — so your data never reaches a model without authorization. Every session and every prompt is logged for full auditability.
Backplain ships on day one with SSO/SAML, SCIM provisioning, RAG file upload, an iOS mobile app, an admin console with policy controls, and a contractual guarantee that your data is never used for model training.
Whether you're in legal, finance, healthcare, or government, Backplain is the only AI workspace where your CISO doesn't have to choose between capability and control.
Flowcode is the fast, simple and secure Way to Use QR Codes.
All codes last forever, and no limit to the number fo codes you can make! Full design customization available for you or your brand. All codes are fully backed with real time scan data analytics.
Kotzilla is a mobile observability and AI remediation platform that helps Android and Kotlin Multiplatform (KMP) teams identify, diagnose, and resolve production issues faster. Built on Koin, the #1 dependency injection framework for Android, Kotzilla automatically captures runtime data at the architecture level, giving developers the structural context to understand not just *what* broke, but *why*.
Unlike general-purpose APM tools that require manual instrumentation, Kotzilla leverages Koin's embedded dependency graph to map component lifecycles and relationships automatically. This architecture-level intelligence makes AI-assisted fixes accurate and targeted, giving LLMs the unique context and knowledge base needed for reliable fixes, even on complex issues.
A single SDK covers Android, iOS, Desktop (JVM), and Web (JS/WASM).
Observability
- Crash & ANR tracking: captures unhandled exceptions, ANRs, and startup failures, grouped with session and architecture context to surface root causes fast
- Session timeline: real-time graphical view of user sessions, showing events, component interactions, and failures leading up to an issue, with no manual log correlation
- Dependency graph visualization: automatically maps the Koin graph and component lifecycles, making it immediately clear which modules or injected dependencies are implicated
- Startup performance monitoring: tracks boot time, slow module loading, and initialization order regressions
Fix production issues in minutes instead of hours
- Kotzilla MCP connects production runtime data directly to the IDE or terminal via the Model Context Protocol, enabling AI-assisted debugging with full application context, not just a stack trace
- Koin Knowledge Base provides expert remediation guidance specific to Koin-based architectures
- IDE plugin surfaces performance reports and issue context inside the development environment, reducing context switching
Deployment
Developers integrate Kotzilla by adding the lightweight SDK to their existing Koin setup. Instrumentation is automatic, no manual trace configuration or annotation required, with a holistic view of the application out of the box. The Kotzilla Console offers a web dashboard for monitoring app health, visualizing session timelines, and sharing diagnostic reports across teams.
Designed for mobile and KMP teams looking to reduce monitoring overhead and accelerate issue resolution, Kotzilla reports a 50% reduction in monitoring costs through focused runtime data capture, and a 90% reduction in time to fix when using the Kotzilla MCP integration.
Netgain is a strategic IT business partner to organizations in the healthcare, financial and legal verticals that require stringent regulatory-compliance adherence.
Camptra Technologies offers comprehensive Oracle Cloud services designed to facilitate seamless digital transformation for organizations. Their end-to-end solutions encompass implementation, optimization, and management of Oracle Cloud applications, ensuring businesses maximize their investment through tailored strategies and expert guidance.
Key Features and Functionality:
- Cloud Implementations: Agile deployment of Oracle Cloud applications, including Human Capital Management (HCM), Enterprise Resource Planning (ERP), and Enterprise Performance Management (EPM), tailored to align with unique business processes.
- User Adoption and Training: Utilization of tools like Camptra Cloud Wiki to provide real-time guidance, on-screen assistance, and embedded tutorials, enhancing user adoption and productivity.
- Change Management: Structured methodologies to assess organizational readiness, engage stakeholders, and develop communication and training strategies for smooth transitions.
- Managed Services: 24/7 monitoring, issue resolution, security management, and performance optimization to maintain Oracle Cloud applications efficiently.
- Strategic Advisory Services: Assessment and planning to align technology strategies with business goals, including cloud strategy assessments and governance planning.
- Training and Enablement: Customized training programs, available on-site or virtually, to empower teams across HR, finance, and IT departments.
Primary Value and Solutions Provided:
Camptra Technologies addresses the complexities of migrating to and managing Oracle Cloud applications by offering a holistic suite of services. Their collaborative and customer-centric approach ensures that implementations are efficient, scalable, and aligned with business objectives. By providing proactive change management and comprehensive training, they facilitate smooth transitions and high user adoption rates. Their managed services offer continuous support, allowing organizations to focus on strategic initiatives while ensuring their Oracle Cloud environment operates optimally. Ultimately, Camptra Technologies empowers businesses to achieve long-term success through expert-led Oracle Cloud solutions.
Karma Advisory helps mid-market organizations reclaim wasted technology return on investment so every dollar advances their mission. The firm bridges strategy, operations and technology, offering advisory and hands-on implementation support across IT portfolio management, project delivery and operations.
Labii Inc. is addressing research data documentation and management difficulties within biotech companies and academic labs. Labii integrates LIMS and ELN.
Using Microsoft SQL Express as the database engine, Sample Master is Windows compatible, and has components such as an intuitive user interface, integrated calculations, automated email alerts, Master Query functionality, scheduling, and automated reporting.