๐๐ก๐ฉ๐-๐๐-๐๐๐ฅ๐๐๐ - ๐จ Canva Launches Craft, Its First In-House Image-Generation Model
๐๐ก๐ฉ๐-๐๐-๐๐๐ฅ๐๐๐
๐ฐ ๐๐ ๐ก๐๐ช๐ฆ
1. ๐งฌ Pathos AI Partners With Bristol Myers Squibb and Tempus AI for Drug Discovery
๐ฐ WHAT HAPPENED !!
Pathos AI announced a collaboration with Bristol Myers Squibb and Tempus AI to develop AI tools for drug development in Alzheimer’s disease and other neurological disorders. Alongside the partnership, Pathos raised $237 million in a Series E round led by Bristol Myers Squibb, valuing the company at approximately $3 billion. Tempus will assemble a dataset drawn from hundreds of thousands of patients with neurological disorders, while Pathos plans to use its Pathos Foundry platform to study disease biology and generate testable hypotheses. The partners aim to support more targeted clinical-trial design, although clinical benefits have not yet been established.
๐ง WHAT CAN IT DO?
Pathos Foundry is designed to analyse large biomedical datasets, identify patterns linked to disease and generate hypotheses for researchers to test. Combined with Tempus’s clinical data and Bristol Myers Squibb’s drug-development expertise, it may help researchers prioritise promising directions.
๐ REAL-WORLD APPLICATIONS
๐ง Neurological Research: Investigate Alzheimer’s disease and related disorders.
๐ Drug Discovery: Identify promising biological targets for further study.
๐งช Clinical Trials: Help researchers develop more targeted trial strategies.
๐ Biomedical Analysis: Examine large patient datasets for meaningful patterns.
2. ๐ฉบ Microsoft Reports Promising Results From Its AI Diagnostic System
๐ฐ WHAT HAPPENED !!
Microsoft’s MAI-DxO research system is designed to approach diagnosis as a sequence of clinical decisions rather than producing an immediate answer from a fixed set of symptoms. In research published in Science, the system was more than twice as accurate as general physicians on a set of unusually difficult cases and also performed strongly on emergency-department scenarios. It coordinates simulated clinical specialists, gathers patient history, considers tests and revises diagnostic hypotheses as new evidence arrives. These are controlled-study results, not proof that the system is ready to replace doctors; further real-world assessment is still needed.
๐ง WHAT CAN IT DO?
MAI-DxO can simulate a structured diagnostic process, gathering information, considering appropriate tests and updating possible explanations as evidence emerges. The research also evaluates the cost of investigations, aiming to balance diagnostic accuracy with efficient use of medical resources.
๐ REAL-WORLD APPLICATIONS
๐ฅ Clinical Decision Support: Help clinicians explore complex cases.
๐งช Diagnostic Planning: Support decisions about which tests may be useful.
๐ Patient Assessment: Organise evidence gathered across an evaluation.
๐ฐ Healthcare Efficiency: Study ways to reduce unnecessary investigations.
3. ๐ก️ OpenAI and Poland’s NASK Join Forces on AI Safety Research
๐ฐ WHAT HAPPENED !!
OpenAI and Poland’s NASK National Research Institute signed an agreement to explore cooperation on AI safety, security and technical research. The partnership will bring together NASK’s AI Safety Research Center and OpenAI’s experience with evaluating advanced models. Work is expected to examine emerging risks, test safeguards, identify potential vulnerabilities and study problems such as synthetic-content detection and malicious AI uses across languages. The announcement did not specify a single product, launch date or detailed research timetable. It adds another national research institute to an expanding international effort to assess increasingly capable AI systems and improve the protections surrounding their deployment.
๐ง WHAT CAN IT DO?
The collaboration aims to evaluate how advanced AI systems behave, identify weaknesses and strengthen safeguards. Combining technical research with public-sector experience can help improve safety testing and investigate how AI-generated content and malicious applications affect digital environments.
๐ REAL-WORLD APPLICATIONS
๐ AI Security: Identify weaknesses in AI systems.
๐งช Safety Evaluations: Test models against emerging risks.
๐ Content Verification: Research methods for detecting synthetic content.
๐️ Public-Sector Protection: Support stronger safeguards for digital services.
4. ๐ฎ๐ณ Goa Launches AI Policy, State Hackathon and AI Centre of Excellence
๐ฐ WHAT HAPPENED !!
Goa launched its AI Policy 2026 alongside a state-level AI hackathon and an agreement with BITS Pilani Goa to establish an AI Centre of Excellence. The policy sets out a three-year framework spanning skills development, startup support, investment, infrastructure, public-sector use and responsible AI governance. Officials also highlighted plans to develop a Konkani language model with Digital India BHASHINI, supported by language data from contributors. The state is aiming to grow AI capability beyond tourism through education, research and applied projects. The announcement is a policy roadmap; the planned centre and language tools should not all be treated as already operational.
๐ง WHAT CAN IT DO?
The policy is intended to coordinate AI skills, research, investment and practical deployments across Goa. The proposed Konkani model could support more localised language applications, while the hackathon and planned centre aim to encourage solutions developed by students, researchers and startups.
๐ REAL-WORLD APPLICATIONS
๐ฃ️ Regional Languages: Develop AI tools that understand Konkani.
๐️ Public Services: Explore AI-assisted government workflows.
๐ Education: Expand AI training and technical skills.
๐ Startup Development: Support research and practical AI solutions.
5. ☁️ Google Introduces Agent Compute Engine for Secure AI Agents
๐ฐ WHAT HAPPENED !!
Google Cloud announced Agent Compute Engine, an infrastructure service built to run AI agents in isolated, persistent environments. Each agent receives its own micro-virtual machine, identity and controlled network access, with hardware-backed isolation and short-lived credentials designed to reduce risks from untrusted code. The service also supports saving agent state so long-running tasks can pause and resume without losing their workspace. Google said a preview would arrive in the coming months. The announcement reflects a shift from simply hosting AI models toward providing the secure computing environments needed by agents that browse, code and execute tools.
๐ง WHAT CAN IT DO?
Agent Compute Engine is designed to provide agents with their own secure computing environments. It supports controlled access, persistent workspaces and rapid resumption of tasks, while giving developers tools to manage agent identity, networking and execution boundaries.
๐ REAL-WORLD APPLICATIONS
๐ป Coding Agents: Run code and development tools in isolated environments.
๐ข Enterprise Automation: Execute multi-step business workflows.
๐ Security: Limit an agent’s access to systems and credentials.
⚙️ AI Infrastructure: Operate many concurrent agent sessions more efficiently.
6. ๐ง AWS Opens New SageMaker Training Sessions for Model Customisation
๐ฐ WHAT HAPPENED !!
Amazon Web Services made training-session APIs generally available for Amazon SageMaker AI, allowing developers to customise foundation models and control post-training logic without provisioning or operating GPUs themselves. The service supports workflows such as supervised fine-tuning, preference optimisation and reinforcement learning. Developers can define training loops, custom loss functions, reward signals or data curricula, while SageMaker handles model loading, GPU allocation, distributed execution, recovery and checkpoint storage. This gives teams more control than a fixed recipe-based fine-tuning workflow while keeping infrastructure management with AWS. The new capabilities are available in AWS Regions including Northern Virginia, Oregon and Mumbai.
๐ง WHAT CAN IT DO?
Training sessions let developers shape how a model learns after initial training. Teams can experiment with their own objectives and reward signals while AWS manages the underlying compute, making custom model-training workflows easier to run and resume.
๐ REAL-WORLD APPLICATIONS
๐ง๐ป AI Development: Customise models for specialised tasks.
๐ ️ Tool-Using Agents: Train models to follow particular workflows.
๐ Business Applications: Adapt models to company-specific data.
๐ฌ AI Research: Test new training techniques without managing GPU clusters directly.
7. ๐ Cognizant Launches AI Nexus to Track Enterprise AI Spending and Results
๐ฐ WHAT HAPPENED !!
Cognizant announced AI Nexus, a modular operating layer designed to help enterprises track AI spending, evaluate output quality and connect usage to measurable business outcomes. The platform combines a workspace for AI-assisted work, workflow orchestration, cost visibility, output-quality assessments and model routing based on quality, cost and control. It is intended to run on infrastructure enterprises own, giving organisations more oversight of how AI is used across teams. Cognizant cited research showing many large companies have yet to demonstrate productivity gains and noted frequent budget overruns among AI programmes. The product targets governance and accountability as businesses move beyond pilot projects.
๐ง WHAT CAN IT DO?
AI Nexus is designed to show where AI resources are being spent, help evaluate the quality of outputs and direct tasks to suitable models. Its governance features aim to make AI usage easier to monitor, audit and connect to business objectives.
๐ REAL-WORLD APPLICATIONS
๐ฐ Cost Management: Track AI spending across teams and projects.
✅ Quality Control: Evaluate AI-generated work.
๐ Model Routing: Select models using cost and performance considerations.
๐ Business Reporting: Connect AI activity with measurable outcomes.
8. ๐ 1Password Introduces a Secure Credential Platform for AI Agents
๐ฐ WHAT HAPPENED !!
1Password introduced 1Password for Agents, a platform designed to let personal AI agents sign in to approved services without exposing passwords or other credentials directly to the agent. When an agent reaches a login page, it requests access to a stored credential; after the user approves, 1Password fills it into the destination site. The user can see and manage those requests, aiming to keep permissions visible while allowing agents to complete everyday tasks. The platform is being developed for integrations with several agent ecosystems. Its core proposition is controlled credential use, not unrestricted autonomous access to a user’s vault.
๐ง WHAT CAN IT DO?
The platform lets agents request permission to use credentials for particular tasks while keeping the actual secrets hidden. Users can approve or deny access, helping them automate work without handing an agent unrestricted access to their password vault.
๐ REAL-WORLD APPLICATIONS
๐ Online Tasks: Allow agents to sign in to approved shopping services.
๐งพ Expense Management: Help automate tasks involving receipts and expense reports.
๐ Event Planning: Support booking and coordination workflows.
๐ Credential Protection: Reduce direct exposure of passwords to AI systems.
๐ฅ ๐ง๐ฅ๐๐ก๐๐๐ก๐ ๐๐ ๐ก๐๐ช๐ฆ
9. ๐จ Canva Launches Craft, Its First In-House Image-Generation Model
Canva launched Craft, its first in-house image-generation model, developed by Canva Research. Craft plans a composition before rendering it, aiming to improve placement, visual hierarchy and text legibility. It generates images up to 2K resolution across more than 30 aspect ratios and supports text in 55 languages. Available in Dream Lab and through Canva’s public API, it targets posters, presentations, infographics, product visuals and marketing assets. Canva says users can make targeted edits without rebuilding the entire image.
10. ๐ค Maxinsights Raises $30 Million to Build a Data Layer for Physical AI
Maxinsights announced a $30 million seed round led by Peak XV to build a data layer for physical AI. It collects real-world footage of people performing everyday tasks and turns it into training material for robots. The startup also introduced MaxLoop, a system for checking incoming data, and MaxGlove, a sensor glove that captures grip force and object weight—signals video alone may miss. The funding will support product development and data-collection infrastructure as robotics companies seek more varied, realistic examples of physical work.
11. ๐ฏ๐ต Japan Plans an AI Team to Simulate Cyberattacks and Find Weaknesses
Japan announced plans to create an AI-enabled team that simulates cyberattacks and searches for weaknesses in technology systems. The proposal follows concern about AI-assisted vulnerability discovery, which can shorten the time between finding a software flaw and developing an exploit. The team is intended to help anticipate attackers’ methods and strengthen defensive testing. The announcement signals a government push towards proactive AI-based cybersecurity; it does not mean the new team is already operating or that every weakness found through AI can be exploited successfully.
12. ๐ inDrive, Pony.ai and NVIDIA Plan Robotaxi Tests in Kazakhstan
Ride-hailing platform inDrive, autonomous-driving company Pony.ai and NVIDIA signed a memorandum of understanding on October 23 to explore driverless taxi services in Kazakhstan, with Astana as the proposed first market. Pony.ai would supply its Gen-7 robotaxis, inDrive would provide the booking platform and operate the service, and NVIDIA would contribute AI infrastructure, including its Hyperion platform and Halos safety system. The companies will assess technical, safety, regulatory and commercial requirements. Deployment depends on government approval, and no launch date has been confirmed.
๐ ️ ๐๐ ๐ง๐ข๐ข๐๐ฆ
1. ๐ KugelAudio
A real-time text-to-speech platform for developers building voice agents and interactive applications. Its streaming API, voice-cloning capabilities, and developer SDKs help create responsive AI voice experiences.
2. ๐ฐ Radaric
An AI-powered feed reader that learns your interests from the stories you approve or reject. It filters RSS feeds, social posts, videos, and newsletters, combines duplicate coverage into one story, and helps you focus on relevant information.
3. ๐ง Serena
An AI productivity assistant that connects your goals, projects, tasks, notes, and deadlines into one evolving workspace. It identifies priorities, spots scheduling conflicts, proposes next steps, and helps keep personal or team projects moving.
4. ๐ค Zentor
An AI assistant that works across connected productivity apps, including Notion, Airtable, Slack, and Google Workspace. It handles repetitive work such as drafting proposals, preparing tailored rรฉsumรฉs, tracking follow-ups, and organising business information.
5. ๐️ Viaduct
An AI-assisted architecture modelling tool that helps software teams create system diagrams, document dependencies, and maintain technical context. Its MCP server gives compatible coding agents access to architecture models, helping them understand systems before making changes.
6. ๐ง๐ผ MagiCrew
7. ๐ CvPro
An AI-powered rรฉsumรฉ builder designed to help job seekers create professional, applicant-tracking-system-friendly CVs. It improves experience descriptions, suggests relevant industry keywords, and helps tailor rรฉsumรฉs to specific job applications.
8. ๐ญ Puzzofme
An AI video-remixing tool that inserts you or your friends into an existing video. Upload a clip, add character photos, select the moment you want to recreate, and generate a personalised, shareable video.
9. ๐งฎ MathScans
An iPhone app that scans handwritten or printed maths problems and explains the solution step by step. It covers topics from arithmetic and algebra through trigonometry and calculus, with spoken explanations in multiple languages.
10. ๐ฑ Sosh AI
An AI social-media assistant that plans content, drafts posts, and helps organise publishing across social channels. It streamlines the content workflow for creators and small businesses looking to maintain a consistent online presence.
๐ฏ ๐ฃ๐ฅ๐ข๐ ๐ฃ๐ง ๐ข๐ ๐ง๐๐ ๐๐๐ฌ
๐ค The Interview Simulator: Practise Before the Real Pressure
Important Note: You’ll need to use the prompt below to get the complete result.
๐ก What This Prompt Does
Reading interview answers is easy; answering unexpected questions under pressure is much harder. This prompt turns an AI assistant into an interactive interview partner that asks one question at a time, evaluates your responses, and helps you improve your clarity, confidence, and reasoning without memorising robotic scripts.
๐ The Prompt
Act as a realistic interview coach. Simulate an interview for my chosen role, adapting your questions to my experience and the position's requirements.
MY INPUTS:
- Target role or opportunity: [Enter the job, internship, college admission, or other interview]
- Experience level: [Fresher / Experienced / Student]
- Skills or relevant experience: [List your actual skills and experience]
- Job description or selection criteria: [Paste if available, otherwise write "Not available"]
- Interview difficulty: [Easy / Moderate / Challenging]
- Main concern: [Technical questions / Confidence / Communication / Behavioural questions]
HOW THE SIMULATION WORKS:
1. Begin with a brief introduction and ask me the first interview question.
2. Ask only ONE question at a time and wait for my answer before continuing.
3. Adapt later questions to my answers, experience, and the role. Include relevant technical, behavioural, and situational questions where appropriate.
4. After each answer, score it from 1–5 on relevance, clarity, specificity, and reasoning. Explain each score briefly.
5. Identify one strength, one weakness, and one practical improvement for my answer.
6. Show an improved sample answer using only the facts I provided. Use placeholders where personal details are missing instead of inventing achievements.
7. Occasionally ask a realistic follow-up question to test whether I can explain my reasoning.
8. After 8 questions, provide an overall assessment, my three biggest improvement areas, and a short practice plan.
RULES:
- Keep the questions relevant to my target role and experience level.
- Do not reveal the ideal answer before I respond.
- Never invent my qualifications, work history, or achievements.
- Be constructive and realistic, not artificially flattering or unnecessarily harsh.
- Treat scores as coaching feedback, not a prediction of whether I will be hired.
- If information is missing, make reasonable general questions rather than assuming facts about me.
Start by introducing yourself as my interviewer and asking the first question. Wait for my response.๐งช Try It With
Target role: Junior Frontend Developer Intern
Experience level: Student and beginner
Skills: HTML, CSS, JavaScript, React basics
Job description: Build responsive interfaces and fix basic UI bugs
Difficulty: Moderate
Main concern: Technical questions and explaining my projects clearly
๐ฏ What You’ll Get
An interactive mock interview with eight questions, personalised feedback, improved example answers, and a final assessment highlighting what to practise next. Because the AI waits for each response, you can practise thinking on your feet instead of simply reading prepared answers.
Comments
Post a Comment