Attri

Most AI conversations focus on what the technology can do. But after spending the last few months working at the intersection of legal, operations, and product, I’ve found that the harder question is something else entirely: What should AI actually own inside a workflow? Because that’s where things start to break. Not in demos. Not in capability. But in the moment when a team has to decide whether they can rely on what the system produces. I wrote about this in my first newsletter piece — AI × Workflows — where I unpack: → why good AI outputs can still slow teams down → how unclear ownership creates friction → and what actually needs to change for AI to work in real environments Would love to hear how others are thinking about this—especially if you’re trying to move from experimentation to real usage. #AI #AIInOperations #AIInLegal #AIAgents #Attri

Most AI tools today follow the same pattern. You open a tab. You describe what you need. You copy the output. You paste it somewhere else. You adjust it. You explain it to someone. You do it again. The tool is powerful. The experience is exhausting. This month something changed. Anthropic shipped Claude Design and Claude for Creative Work. OpenAI has Codex. Two of the most watched AI companies in the world made the same bet in the same month, stop competing on benchmarks and start competing on workflow. A PM went from rough idea to working prototype before anyone left the room. An engineer delegated a complex task and came back to it done. A musician found samples without leaving their recording software. That's not a feature launch. That's a shift. Are you reading these launches as product signals or just feature updates? Curious what you're seeing. #AIProduct #ProductManagement #Anthropic #OpenAI #WorkflowAI #AIPM #StartupLife #BuildingWithAI #Attri

Watched a founder I know lose a Fortune 500 healthcare deal last month. Not on the model. On a 47-page audit questionnaire he couldn't answer. "Who has access to the customer data?" → Couldn't say. "Show me the evidentiary trail for any agent decision." → Silence. The CIO bought from a slower, less impressive competitor with a real audit trail. Speed was the 2024 metric. Audit-readiness is the 2026 one. #EnterpriseAI #AICompliance #AIGovernance #ResponsibleAI #AIStartups #AIInProduction #RegulatedIndustries #AuditReady #AIInfrastructure #FounderPerspective

A PM on my team filed a bug last Tuesday. An agent read it, fixed it, wrote the tests, shipped it, and closed the ticket. Total human engineering time: zero minutes. Every LinkedIn post I've seen this year would tell you that story ends with "so we fired the engineer." It doesn't. And the wrong lesson is going to cost a lot of companies a lot of money. I wrote about why in this week's newsletter. Quick context: I've been sending a weekly piece like this to Attri customers and a small group of contacts since November. A few of them kept forwarding it, kept asking to add their friends, kept telling me to put it on LinkedIn. So I'm putting it on LinkedIn. One pragmatic take every week, from a CTO building AI for regulated industries. No hype. No "everything is changing." Just what's actually shipping in enterprise AI, what's breaking, and what to build next. This week's piece: why the companies announcing "we're replacing roles with AI" are about to regret it, and why the real play in 2026 is human leverage, not human replacement. Feel free to subscribe if you want to get notified every time I post. #Attri #AIInProduction #FutureOfWork #EnterpriseAI #AIAgents #AgenticAI #CTO #HumanInTheLoop #TechLeadership

Something clicked for me after shipping multiple AI agentic projects. The bottleneck was never the AI. It was the tools we were forcing the AI to use. CRMs that expect a human to log in. APIs that mirror dashboard clicks. Permission models built for people, not programs. Our agents were spending more time pretending to be human than doing their actual job. That realization led us to build something different. Wrote about the whole thing — the frustration, the build, and the three lessons I didn't see coming. ↓ #ArtificialIntelligence #AIAgents #AgenticAI #OpenSource #CRM #EnterpriseAI #LLM #AIStrategy #Opentooling

The best product lesson I learned this year didn't come from a book, a course, or a conference. It came from a conversation with a colleague. We were talking about discovery, how to actually understand what a customer needs before you build anything. And somewhere in that conversation, one line stopped me completely. "You know you've done discovery wrong when you're doing more talking than the customer." That hit differently. New article is up, everything I took from that conversation, through a product manager's lens. How listening is the most underrated skill in product. And why the best AI products start with silence, not specs. Thank you Miguel Bravo for sharing your experience and insights so openly. I learn something new every time we talk. #ProductManagement #CustomerDiscovery #AIProduct #AIPM #StartupLife #BuildInPublic #Attri