AI builder. Founder. Operator log.
Carlos KiK
RSS FeedI write from inside the build: AI, persistent memory, founder reality, and the future of human-machine connection. I am currently building KAi at Digital Human Corporation in Seoul.
This site is the public notebook for the work around it: what is changing in AI, what I am learning while building, and what it costs to keep going when the obvious path is not available.
Current Bet
DHC, KAi, persistent memory, and the long fight to build useful AI without turning people into engagement inventory.
The working thesis is simple: the next serious AI companion category is not about hotter chatbots. It is about memory, trust, privacy, continuity, and whether the system helps people return to real life with more clarity.
ckik.dev tracks that thesis in public through daily AI analysis, founder notes, and the story behind the company. Some pieces are fast dispatches. Some are long arcs. The center of gravity is always the same: what is real, what matters, and what I would bet my own time on.
For the company-side version, read The Architect .
Follow The Lab
AI Intel
The daily signal: models, agents, robotics, memory, compute, security, and the product moves that matter.
Startup Life
Founder reality from Seoul: public story arcs, technical scars, hard lessons, and the cost of building for real.
DHC And KAi
The company and companion I am building around persistent memory, privacy, and adult AI support.
- Field notes from the AI frontier
- Decision ledgers from the build
- Founder stories with the dangerous parts filtered correctly
- Unsung heroes behind the systems everyone uses
Recent Posts
-
Amazon Bee Shows the Ambient AI Consent Problem
Amazon's Bee wearable points toward proactive ambient AI, but always-available memory has a hard social problem: other people did not opt in.
-
Anthropic Just Said the Quiet Part About Agent Security
Anthropic's Claude containment post makes the real agent-security lesson obvious: permission prompts are not enough. The boundary has to be deterministic.
-
Claude Code May Be Making Developers More Technically Adventurous
A new arXiv paper studies 5,838 GitHub developers and finds Claude Code adoption coincides with more commits, more repos, and broader language use.
-
DeepSeek Is Turning Cheap Cached Inference Into an Agent Strategy
DeepSeek says V4-Pro pricing will move to one quarter of its original price after May 31, while Reasonix shows what cache-first coding agents can look like.
-
Memory Is Becoming the AI Chip Tax
Epoch AI estimates high-bandwidth memory rose to 63% of AI chip component spending by late 2025. The AI infrastructure story is no longer just about logic dies.
-
Constraint Decay Is Why Coding Agents Break in Real Repos
A new arXiv paper found coding agents lose about 30 points as structural backend constraints accumulate. The lesson is simple: demos reward output; production rewards constraint discipline.
-
Project Glasswing Moved the Bottleneck From Finding Bugs to Fixing Them
Anthropic says Claude Mythos Preview and roughly 50 partners found more than ten thousand high- or critical-severity vulnerabilities. The scary part is not discovery anymore. It is disclosure, triage, and patch throughput.
-
Microsoft's Security Copilot Agent Is the Boring AI Win
A new Microsoft Security Copilot paper says its Dynamic Threat Detection Agent runs across tens of thousands of Defender customers with 80.1% precision. This is what production agents are starting to look like: narrow, audited, always-on, and embedded inside existing workflows.