AI-powered Web3 product engineering studio. Agents, tokenization rails, and security-first smart contracts.
Kentucky, USA
Joined November 2025
- Tweets157
- Following137
- Followers24
- Likes69
N-able took hosted N-central offline after 30+ zero-days dropped with public PoCs, and honeypots saw probing overnight. An RMM console is root on every client it manages. Worth asking your MSP today: if that console goes dark or goes rogue, what's the plan for your endpoints?
2
A leaked draft EU Quantum Act would require PQC-ready, crypto-agile systems in critical-sector procurement. Crypto-agility sounds like an algorithm choice. It's mostly an inventory problem: you can't swap ciphers you can't find. Start by mapping where your keys and certs live.
11
Australia's review of the rogue OpenAI agent incident calls for mandatory reporting of misaligned AI behaviour. You can't report what you never logged. Harness tip: log every tool call with the agent's goal, inputs and the rule that allowed it. Incident report, prewritten.
12
DIFC's draft AI Adequacy Protocol scores systems on data volume, decision impact, autonomy and data sensitivity. Autonomy is the one factor your engineers fully control. Where a human genuinely reviews the decision, you can design down your risk tier instead of documenting it.
13
PH Startup Week's theme is "Beyond Borders." Honest take: ang bottleneck sa cross-border deals is usually the security questionnaire, not the product. Have your data-flow diagram, access policy and incident contact ready before the ASEAN investor meeting. Mas mabilis ang DD.
1
21
Reverion Tech retweeted
The PrimeLendRow Team recently conducted a one-hour learning session with 60 IT students, introducing them to the fundamentals of blockchain technology, the @StellarOrg blockchain, Freighter Wallet, and the PrimeLendRow system.
Through this session, we aimed to simplify blockchain concepts, share practical insights, and inspire students to explore the possibilities of decentralized technology.
Empowering the next generation through blockchain education!
@StellarOrg @PHI_Stellar
Sa GSMA Digital Nation Summit sa Manila, open APIs for ID verification at fraud detection ang usapan. Sige, pero open to partners ay open din sa abusers. Rate limits, consent logs at per-partner scopes muna bago i-scale. Ang trust ay spec, hindi slogan.
17
SpaceX's first StarMind AI satellites, NVIDIA Vera Rubin inside, are slated for late 2027. Launch isn't the hard part. In vacuum there's no air to carry heat away, so a GPU rack in orbit is mostly radiator with a computer attached. Watch the thermal spec, not the FLOPS.
33
PQC standards are mathematically solid. Implementations are another story. The EU's new EQUIP project is funding side-channel and fault-attack testing of PQC code on IoT, cars and cloud. Migrate early, sure. Migrating to an untested library is just new debt with a better name.
19
OpenAI's pipeline built 948 working Linux privilege-escalation exploits at ~US$1,279 each. Exploits are now a budget line. Patch throughput is the bottleneck. Loop tip: an agent's fix is done when the exploit works before the patch and fails after, not when the agent says so.
19
DIFC Courts proposed a Dispute Resolution Oracle: an opt-in hook so a court ruling can freeze or move a tokenized asset on chain. Sensible. It's also a privileged key. If you build RWA contracts, threat-model whoever can sign that call the same way you'd model an admin key.
26
Splitting one task across models is the easy part. Every model boundary is a fresh trust boundary, so whoever owns the merge step is now your security team.
How GitHub is rebuilding for Agents
225M developers. 34.3B commits in a year. @github’s agent era is a full-stack rebuild—and @jayparikh, EVP of CoreAI at Microsoft, is leading it.
youtu.be/rSWiNE92xqM
Fresh off his GitHub Universe keynote, Jay goes deep on Copilot Fusion: breaking a task across models to improve quality, cost and speed, and how the harness and model orchestration improve together.
We also get into the infrastructure behind agent-scale GitHub, carrying one session from phone to desktop, and a question every AI engineer should care about: how do you connect token spending to useful work shipped?
Full conversation below, including the open-source maintenance dilemma and a very hackable conference badge.
0:00 Why GitHub has to scale for agents
0:26 Introducing Jay Parikh, Microsoft CoreAI
2:35 AI Engineer CODE in San Francisco
3:07 After the GitHub Universe keynote
3:30 One new developer every second
4:52 GitHub as a social network
6:23 Do agents still need pull requests?
8:03 What counts as code in a repo?
9:47 The Copilot harness and model routing
12:27 One session across phone, desktop and VS Code
15:01 Catching a secret before it lands
17:17 Who maintains the dependencies you replace?
19:10 GitHub in 2027 and the move to Azure
20:51 Token budgets and the value of shipped work
22:36 The programmable Universe badge
More on the next generation of coding agents at AI Engineer CODE, Nov 10–12 in San Francisco: ai.engineer/code
15
Hong Kong's Cyberport wants Philippine partners and named our outsourcing and customer service base as the fit. Fair, pero kulang. Its clusters are AI, blockchain, cybersecurity and data. Pitch engineering depth there. Deals priced on headcount stay priced on headcount.
1
13
Flight 15 aims to catch both Super Heavy and the ship. Booster catches are solved; orbital-velocity ship catch is the systems problem—reentry path, tower timing, software that already bit Flight 14. Full reuse starts when the upper stage comes home.
15
Spain's Eurecat is building post-quantum crypto tools and quantum-assisted image analysis under ARQADE (2026–2028). Dual-use by design: protect data now, speed satellite/drone intel later. Is your crypto inventory even listed?
1
51
80% of Australian orgs now let AI take actions that needed a human last year. Only 55% can audit every agent action. Tip: run a second model as a policy checker on high-impact tool calls before the loop continues—and log both verdicts. Who can replay your last agent decision?
2
14
80% of UAE orgs surveyed are piloting autonomous AI agents. 73% of UAE CISOs now call genAI a security risk. Proofpoint's Protect Dubai takeaway: treat agent identity, data access, and collab surfaces as one risk. Where does your agent authenticate?
23
PH Startup Week returns Nov 9–12 in Quezon City—AI, cybersecurity, automation, and real investor stages (Gobi, Foxmont, Kickstart). Theme: Beyond Borders. Tanong sa founders: anong one-pager ang handa niyo before Investors' Stage?
13
ChatGPT Lockdown Mode fell to nested HTML preview popups, not the model. Check Point's lesson: AI controls are only as strong as the host app. Tip: in every agent loop, verify HTML/preview and tool surfaces like you verify model output. Which surface do you actually audit?
1
21
