NOW · BUILDING IN PUBLIC

What Bloss0m is tracking and building now

Daily AI signals, active research, and side-project progress — published as a living engineering log.

Last updated Jul 28, 2026 · Important signals are published as they emerge.

latest signals shown
5
active directions
2
active side projects
2

01 · DAILY AI RADAR

Signals worth following now

Models, agents, MCP, developer tools, and enterprise adoption. Tracked daily; published when a signal deserves technical context.

  1. Industry Pulse Generative AIAWS

    Bundesliga Partners with AWS: Generative AI Creates a Global Fan Experience

    How the Bundesliga is leveraging AWS and Generative AI to bring an innovative, closer-to-the-action experience and automated content production to its 1 billion fans worldwide.

    Read signal →

  2. Industry Pulse AIMeta

    Meta Launches Muse Spark: Architecture and Practical Judgment of the Next-Generation AI Model Towards 'Personal Super Intelligence'

    Meta Superintelligence Labs launches its first model: Muse Spark. A comprehensive breakdown of its natively multimodal reasoning mechanism, the test-time computing architecture behind the highly discussed 'Contemplating Mode', and its RLHF practices in the health and medical domains.

    Read signal →

  3. Industry Pulse MetaAI Image Generation

    What Is Meta Muse Image? Agentic Generation, Availability, and Limits

    A source-backed analysis of Muse Image's search, coding tools, self-refinement, and test-time compute, with its product availability, evidence, and adoption limits.

    Read signal →

  4. Industry Pulse GeminiGoogle

    Nano Banana 2 Lite and Gemini Omni Flash: Architecture and Impact on Image and Video Generation

    An in-depth analysis of Google's newly released Nano Banana 2 Lite image model and Gemini Omni Flash video generation and editing model, exploring how they bring new possibilities to developers with ultimate speed, cost-effectiveness, and multimodal integration.

    Read signal →

  5. Industry Pulse AnthropicAI

    The Impact of AI on the Labor Market: A New Measure from 'Theoretical Capability' to 'Observed Exposure'

    A summary based on Anthropic's 'Labor market impacts of AI: A new measure and early evidence': introduces the 'observed exposure' metric, explains which occupations are most exposed to AI, its relationship with employment growth and unemployment rates, and implications for policy, businesses, and individual careers.

    Read signal →

Browse all frontier signals

02 · CURRENT DIRECTIONS

Long-term workstreams in motion

Each direction has a current stage, concrete next step, and public evidence.

  1. Continuous tracking and validation

    Enterprise Agent Technology Radar

    Track models, agent harnesses, MCP, memory, evaluation, observability, security, and enterprise adoption, translating frontier signals into actionable engineering decisions.

    This week：Established a Signal → Prototype → Pattern → Evidence loop; next is organizing Agent research, papers, and builds around enterprise adoption questions.

    Related AI Agent engineering map → Agentic AI Platform →

  2. Development center expansion

    Clubhouse Games: One-shot / Few-shot Development Center

    Test whether AI can complete game specifications, scaffolding, interaction logic, testing, and publishing from one or a few instructions.

    This week：Using 22 existing game specifications and multiple web implementations to design a record for prompts, shot count, revisions, elapsed time, and playability.

    Related Clubhouse Games project → GitHub Repository ↗

03 · SIDE PROJECTS

Projects moving forward

Stage and next milestone are shown instead of arbitrary completion percentages.

IN PROGRESS

  1. In progress

    Clubhouse Games

    Expanding the existing game collection into a One-shot / Few-shot AI game-development test bed.

    Next milestone: Define the first comparable experiment record: prompt, shot count, revisions, elapsed time, and playability.

    Open project →

  2. Reference system evolution

    LINE Chatbot · n8n Workflow Platform

    Continuing to turn Agent platform experiments into reusable enterprise architecture patterns.

    Next milestone: Connect orchestration, MCP, evaluation, observability, and governance into one inspectable delivery path.

    Open project →

RECENTLY DOCUMENTED

  1. Titanic — Machine Learning from Disaster Three pipelines tie; Step 5 feature recipe alone adds +3% LB
  2. pdf2ppt: Convert NotebookLM PDFs to Editable PPTs NotebookLM / slide-style PDF → editable PPTX
  3. Portfolio Opening Bell Agent Combine real holdings, realtime quotes, and news context into a structured opening report and decision dashboard so investors can gauge risk and opportunity before the open.

04 · RESEARCH QUEUE

Questions waiting to be tested

  1. Validating

    Enterprise Agent evaluation and observability

    How should teams connect task success, traces, cost, latency, and human review?

  2. Next

    Agent memory and context engineering

    Compare persistent memory, retrieval updates, forgetting risk, and long-running context strategies.

  3. Watching

    Self-improving and computer-use agents

    Track reliability, security boundaries, and where autonomous interaction becomes useful in enterprise workflows.

05 · FOLLOW THE WORK

Choose the signal you want to follow

Use RSS for new writing, GitHub for source changes, or email for the small digest pilot.

Follow RSS GitHub ↗ Email pilot