# Decision Points for Embedding AI Agents in Enterprise Systems # Employee-facing vs Customer-facing 🎯 The Hook Are you using the same design for your internal employee agent and your customer-facing agent? The trust model, data boundaries, guardrail strength, and blast radius of failures are fundamentally different depending on who is using the agent. Get this fork wrong early and you will either expose customer data through a lax design or crush employee productivity with excessive restrictions 🔑 📋 Overview Whether the agent's users are employees bound by employment contracts and NDAs or external customers who may submit adversarial inputs is the very first architectural fork. For employee-facing agents, you can assume a baseline of trust — authentication is unified through corporate IdP (Okta / Entra ID), data stays within internal knowledge bases (Notion / Confluence / Box) and business systems (Salesforce / ServiceNow / Workday), and the failure cost is typically "rework" rather than "lawsuit." Customer-facing agents are a different world: you must design for jailbreak attempts and indirect prompt injection, outputs directly affect brand reputation and legal liability, strict per-tenant data isolation is mandatory, and the authentication layer must handle CIAM (Auth0, etc.) including anonymous access. Scale is also orders of magnitude apart — thousands of employees versus millions of customers with 24/365 spike resilience 📊 🔍 Decision Points The two driving variables are user trust level and failure blast radius. Users are employees only → Employee-facing design is sufficient External customers are included → Customer-facing design is mandatory Both are included → Design as separate planes with isolated data paths The critical insight is distinguishing what can be shared from what must be isolated. The orchestration layer and model gateway (AI Gateway) can be shared across both planes. Data access paths, guardrails, and audit logs must be separated. For customer-facing output pipelines, integrate DLP (Data Loss Prevention) to prevent internal data leakage ⚡ 💡 Key Details Employee-facing characteristics: - Input trust assumption holds (employment contracts and NDAs provide accountability) - Data scope is limited to internal knowledge bases and business systems - Failure cost is "inefficiency" or "rework" — mostly reversible - Concurrent users in the thousands to tens of thousands, concentrated during business hours Customer-facing characteristics are fundamentally different: - Design must assume adversarial inputs (jailbreak, indirect injection) - Incorrect answers or data leaks escalate to legal liability and brand damage - Strict per-tenant data isolation is non-negotiable - Concurrent users in the tens of thousands to millions, 24/365 spike resilience required In hybrid configurations, a typical pattern is an e-commerce platform on Shopify where the customer-facing inquiry agent and the internal order-operations agent share the same orchestration backbone but run on separate planes. On Zendesk, the customer-facing agent is given access only to a read model projection of publicly shareable knowledge, with direct access to internal Notion and Box data completely blocked 🔒 ⚖️ Trade-offs Reusing an employee-facing design for customers is the most common and most dangerous mistake. Data access scopes that are acceptable internally get applied to customer-facing agents, and suddenly one tenant's data leaks to another. Going the other direction — applying strict customer-facing guardrails company-wide — tanks employee productivity and kills the ROI of the entire agent initiative 😩 Deferring plane separation to "later" is equally risky. Starting with a single stack and planning to separate at customer launch is a recipe for ballooning migration costs. Separation must be decided at initial design time. Audit log co-mingling is another overlooked trap — when customer PII and internal data share the same log store, data retention policies and access controls become unmanageable ⚠️ 🛠️ Use Cases E-commerce operations + customer support: A Shopify-integrated retailer runs a customer-facing inquiry agent (strict guardrails, tenant isolation, DLP enforcement) alongside an internal order-management agent (broad internal data access, wider operation permissions) on the same orchestration backbone. The model gateway is shared; only the frontend and data paths diverge 🛒 IT helpdesk: An employee-facing IT support agent has broad access to Active Directory, Jira, and Confluence, automating password resets and software provisioning. When the same platform is extended to customer-facing support, access is restricted to the public knowledge base, guardrails are significantly hardened, and topic restrictions, tone control, and refusal policies are layered on 📚 Practical tip: If there is even a possibility that you will need both employee-facing and customer-facing agents, design for plane separation from day one. Retrofitting separation is among the most expensive categories of technical debt 💪 #AIAgents# #EnterpriseArchitecture#

Autopilot Update: I have reviewed the positions and there's nothing I feel like should be changed. We were well diversified enough that we handled recent drag with minimal downside, and are now positioned for what may be a decent leg up. Repricing is done from yesterday. The hike was priced in, not Warsh's tone or pressure, so that down and bounce repricing is c o m p l e t e. Oil price is falling. This means, despite the dot plot on the SEP, the next likely rate hike date (Dec) might lose traction if pass-through inflation dies out. Granted, we have yet to see much of the sustained oil prices it hit economic data in my opinion, but it would be lagging. So the market will be guessing. AI trades are running. Quad witching tomorrow. So: hopeful that this sustained period of stagnation and consolidation in prices is being broken out of, but we will have to see if it's a relief rally or a run with sustainable legs. Time will tell, as always. Adam

I think Muse is what AI is supposed to look like. Feels like every lab has been shipping the same thing recently with minor incramental improvements. Text boxes in front of a model = shiny new toy syndrome! No one cares if Muse Spark is the best model in the world rn because it literally doesn't matter to 99.9% of people. Only the top 1% of coders/engineers care about what's changed from one Claude release to the next ChatGPT release. Personally, I have found little improvement since Fable 5 was released and I think that says a lot. I'd probably say it's regressed recently for higher end research tasks. Muse makes me more convinced now that Anthropic/OAI have absolutely zero moat at the consumer level. Enterprises are a different story and that's where both frontier labs need to (already are) focus rather than doomerism and frontier pacing bs. Agents are where AI endgame probably goes - most users just care about usefulness so having one run laborious tasks is way more important than a tiny push on the frontier by Fable 5.1 / Astra to most people. Checking in with your agent will become force-of-habit, same way you check your WhatsApp or emails every morning. Not everyone logs onto Claude or ChatGPT in the same way. (Already a known thesis on Agents, but yeah, worth resurfacing). For the average Joe, this is all the matters. Same way most people don't change phones from one generation to the next because their current phone is good enough. Same way laptops were so important for the PC industry because they were just more convenient than desktops. In that sense, I guess you could say that Muse has pushed the frontier from a usefulness POV. I think we think that most people care about AI like we do...they don't lol. If AI is gonna become mass adopted, it'll be via agents doing stuff for people just how Muse is rn. Since most people already know ChatGPT (noun for AI e.g. "ChatGPT it" similar to "Google it"), it seems like OAI doing a solid enough agent could be enough to oust Anthropic longer term at the consumer layer to have all your context with one lab in an agent + frontier model combo. Sorry for the ramblings and no idea if this logic holds lol - I'm walking to get lunch and wanted to write after a stressful morning!

BRASILIA E AÍ 🧡♾️ Tonight’s the last show in South America for The Lifetimes Tour and I’m so full of joy and many varieties of empanadas. You know when I say I love you for life, I mean it. Thank you for showing up like you did Cats (& Rats too… I guess) ♥️

AI is still so early, but so bullish. We're witnessing a huge unlock of economic value with the adoption of AI models like Claude, and more recently, agents like Grok Bot. This then filters down into the most bullish of bull points for AI. Enterprise AI adoption increases productivity -> increases enterprise earnings -> increases token demand as a flywheel effect -> increases ROI for the AI labs -> increases demand for AI infra e.g. compute. I recently heard from an ex-colleague that a large consulting firm is increasing their AI spend by 100x at their London office for their back office teams. Just incredible growth really. We can kinda see the enterprise flywheel manifest in the $NVDA earnings too. AI Clouds, Industrial & Enterprise revenue was up 25% sequentially compared to 13% for the hyperscalers. I think that this is also why $NBIS and $CRWV keep signing enterprise and lab capacity faster than capacity comes online too. 2027 capex will be huge because the labs / hyperscalers still can't get enough compute to serve enterprise demand right now...and that's before AI adoption has scaled properly across the wider economy. Fun times ahead with AI.

Naomi Watts is the voice of Peter's new AI assistant E.V. in ‘SPIDER-MAN: BRAND NEW DAY’. They previously starred together in Tom Holland's first major movie ‘THE IMPOSSIBLE’.

There are 10,000 AI agents. Nobody knows which ones they actually need. Build vs buy is just the first question. Run your use cases through this 3 step decision tree: 1. Is it generic or unique to your company? - Generic: a standard function every company does the same way. Support inbox, meeting notes, invoices. Buy a vertical agent. - Unique: it runs on your expertise and playbooks. The stuff no vendor has. Build, or configure a horizontal agent. 2. One-off or recurring? - One-off: analytics deep dive, some landing pages, a research report. Buy and configure a personal agent. No need to build from scratch with an agent framework or SDK. Grok Bot is getting crazy good. Integrations are easy. I spun up a half decent marketing analytics agent in an hour. The missing piece is evals. But if it's just for one project, and you're the human in the loop, you're fine. - Recurring: it runs every week, every deal, every client. Build it. 3. How much does a human touch it? The less human in the loop, and the more complex the work, the more engineering you need. Human babysits everything, internal, low stakes, handles basic tasks? Nontechnical person can vibe code it. E.x.: marketing dashboard, lead enrichment script, content repurposing workflow. Little or no human, your playbook and expertise at scale, customer facing, handling a large body of complex work? You need AI engineers. E.x.: healthcare customer support agent, background agents that work while you sleep, custom web apps non-technical teams use all day.

Sixth question on AI. Does AI need to be a zero-sum race between the US and China? Many influential voices in tech as well as the current administration argue that it does. (See, for example, and this from the White House, There are many aspects of these arguments, but I would like to focus on three that have received less attention than they deserve. The first aspect is that this framing typically assumes that there is a single direction of progress for AI, and the US and China are racing along this dimension toward superintelligence or at the very least some sort of powerful AI that will confer huge military and strategic advantages to the winner. If, on the other hand, the US and China are developing these technologies in different directions, there is much more room for coordination and even for each country learning from the other’s best practices. My reading of the evidence is that there are huge differences in the American and Chinese approaches. The US is focused on foundation models, such as frontier large language models. China is also investing – and arguably catching up – in these models, but there is much greater emphasis in China on applications of AI (with the government setting targets for the spread of AI across industries, This reflects several factors, in my assessment. First, Chinese AI investments started with government demands for surveillance technologies and government-supplied data to AI companies (e.g., Second, China is facing a huge demographic transformation, which will create scarcity of labor in the near future. This encourages companies to integrate robotics and AI much more actively into the production process (and China accounts for 59% of all robot installations in the world, Third, applications and integration of these technologies into the production process requires a lot of engineering. China has a large and very well-trained supply of engineers (whereas the US has far fewer engineers and a lot of engineering talent is currently going to Silicon Valley, which hasn’t focused on applications). Fourth, China was relatively late in catching the “artificial general intelligence,” AGI, bug, although now there is more emphasis on this in China as well. Whatever the exact causes, this difference in approach implies that there is plenty for the US to learn from China as well as the other way around. The second aspect is that the zero-sum framing ignores or downplays the crucial need for global coordination. This framing claims that the strategic and military advantage that powerful AI will create is all that matters. But the race to AGI also poses huge risks (and this is true regardless of whether this race is truly advancing toward superintelligence or just producing distorted intelligence, as I argued in my first post in this series). And these risks are global. If the US and China are racing toward AGI with no guardrails and coordination, even the follower (probably China) will have models capable of creating havoc. In this case, racing without any global coordination is a recipe for disaster. The third aspect of the problem is that this framing narrows the scope for any kind of coordination on other global issues between the US and China – including climate change, pandemics, nuclear nonproliferation and world peace. Tellingly, climate cooperation between the US and China has now largely stalled). Let me add that I’m saying all of this not as an admirer of the Chinese system. On the contrary, I have been a vocal critic of both the Chinese Communist Party and the Chinese development strategy since Why Nations Fail ( and I have argued that the Chinese success has some of its roots in severe labor repression ( Indeed, this argument does not ignore the fact that China has an authoritarian, repressive regime, and it is using AI for surveillance and to stamp out dissent. Global coordination and efforts to set up reasonable rules that apply to all parties do not require the counterparty to be fully democratic or ruled by angels. In fact, sometimes it is more important to set up such a global framework when dealing with dangerous adversaries (as the arms-control treaties that the US and the Soviets signed demonstrate). In sum, my assessment is that it is a colossal mistake to frame the AI race as a zero-sum one between the US and China. It poisons the much-needed dialogue and coordination between the US and China on broader issues; it makes the AI race more reckless; and it discourages the building of reasonable guardrails and regulations. The last point is particularly relevant, and makes me sometimes say: if China did not exist, Silicon Valley would have to invent it.

Anyone who ever owned a foil card knows the tilt. Now you can just describe one. We asked for a fruit fly playing with a Rubik's cube on a starry background. Back came "Fruit Fly AI" — ukiyo-e composition, coloured sumi-e linework, a two-state morphing front that flips. It opens in a browser. Drag to rotate, flip to the back, tilt it and the foil moves like cardboard does. $1 a card. Under an hour.