Xavier JOLAS

Interested in People | New Tech | Data & AI | Mobile and Location Based Marketing | Outdoor Ads | IoT | Digital Retail Solutions

All opinions are my own.

[TraceMem @tracememcom]

Healthcare AI agents now screen compounds, match trial patients, and draft payer appeals, all while reading EMRs. "Allowed on the chart, blocked from the inbox" isn't a prompt instruction. It needs to be a policy gate the agent cannot bypass. That's the layer we've built at TraceMem. CDW article: https://t.co/wVOssqnl8s

[Shikhar @shekhu04]

News on tech every week:-

- claude just killed figma
- claude just killed designers
- claude just killed developers
- claude just killed shopify
- claude just killed interns
- claude just killed stripe
- claude just killed code reviews
- claude just killed stackoverflow
- claude just killed your job
- claude just killed your replacement
- claude just killed claude

is claude just killing everything??

[reDB @redb_co]

Claude Code introduces productivity boost but also significant security and governance risks, particularly regarding autonomous agent behavior and supply chain vulnerabilities. Unless an additional AI governance layer can be introduced...
@tracememcom #Claude #AIgovernance

[TraceMem @tracememcom]

Every AI decision is a chance for your organization to get smarter.

#TraceMem turns actions into precedents: analyze patterns, refine policies, ensure cross-agent consistency, and build institutional memory.

This is the enforcement + memory layer enterprises need to scale safely. https://t.co/vhJgS2Nqk1
#AIAgents #AITrust

[Xavier JOLAS @xavierjolas]

@cloudsa @tracememcom is another one

[Xavier JOLAS @xavierjolas]

@NoahKingJr Dishes

[TraceMem @tracememcom]

AI coding agents are still authenticating to production with ungoverned credentials.

The model can never be the security boundary. Agents must emit structured, reason-bound proposals only. An external policy engine then enforces least-privilege scoping, issues ephemeral tokens, and records tamper-evident decision memory before anything touches production.

No more credential roulette. https://t.co/vhJgS2Nqk1

[TraceMem @tracememcom]

When an agent with broad credentials deletes a production database, observability logs the damage - but never stops it. We built TraceMem to enforce at the decision layer: agents submit reason-bound envelopes for real-time policy evaluation, have zero direct keys, and every outcome is sealed in an immutable DecisionDB. https://t.co/vhJgS2NY9z

[Xavier JOLAS @xavierjolas]

@tracememcom Proper governance of these decisions is becoming more and more critical!

[TraceMem @tracememcom]

A brilliant line from Claude: "Don't give the model the gun and tell it not to shoot. Just don't give it the gun."

Prompts are not a security model. Agents shouldn't hold credentials. They shouldn't have direct write access. They shouldn't be able to escalate their own scope.

Every agent action should be policy evaluated, reason-bound, and logged before it runs - not just coached through a 10,000-word system prompt.

This is why we built TraceMem.

[TraceMem @tracememcom]

Everyone's excited about agentic AI making decisions.
Nobody really talks about where the reasoning & approvals disappear to.

At TraceMem we're working to fix that: durable decision memory, not after-the-fact logs. Pre-enforce policy. Post-explain everything. Tamper-evident.

→ https://t.co/mX5LpfTCVO

Let's scale AI without scaling risk. Much more to announce shortly! #AgenticAI #AIGovernance

[Xavier JOLAS @xavierjolas]

@janszczepanski_ https://t.co/FIBoZsnyuX might be the answer 😃

[Xavier JOLAS @xavierjolas]

@Aurimas_Gr Tracing is a good start but being able to manage decisions on the execution path of AI agents is even better. Check https://t.co/FIBoZsnyuX

[reDB @redb_co]

AI agents isolating grid faults? Blocking cyber threats? Rerouting logistics?

#TraceMem captures intent, evidence, rules, and outcomes as tamper-proof institutional memory—for explainability, audits, trust.

#AIAgents #Cybersecurity #EnergyTech #AICompliance #TraceMem #EUAIAct

[reDB @redb_co]

Compliance risk is now a tech infrastructure problem.

Firms with complete data traceability move faster through audits, respond to regulators in hours not weeks, and reduce breach investigation time from months to days.

TraceMem solves this👇

[reDB @redb_co]

Financial institutions: what if AI-driven loan decisions' traceability was built inside the decision flow, not bolted on afterwards?

#AI #traceability #tracemem #Finance #Governance #EUAIAct

[Tommi Hippelainen @hippelainen]

https://t.co/YuUxZBCzuL

[reDB @redb_co]

EU AI Act Article 12 requires high-risk AI systems to log everything. Automatically. Immutably. Forever.

That's exactly what TraceMem does. We're the infrastructure that makes Article 12 compliance automatic—so enterprises can deploy agents confidently. #AIAct #AIGovernance

[Xavier JOLAS @xavierjolas]

@redb_co Ready for #AIact coming into force in August 2026 for #traceability!

[reDB @redb_co]

Get started at 👉 https://t.co/h7F8jdvd7w