About
Jared operates at the intersection of building AI technology and putting it to work in…
Articles by Jared
Activity
117K followers
-
Jared Spataro shared thisPeople started by asking AI questions. Now they're delegating tasks, building solutions, and automating work that runs in the background. We see those patterns in trillions of productivity signals across Microsoft Copilot. Office set the shape of productivity for the PC era. Copilot is doing that for the AI era, expanding both what people can do and who gets to do it. Today we’re introducing the new Copilot, an evolution of the Copilot app that connects the tools people rely on with the next generation of capabilities they’ll need to build, customize, and scale AI across work. Copilot has three new capabilities: Office in Copilot brings the full value of Word, Excel, and PowerPoint right where the conversation’s happening. Code in Copilot empowers every knowledge worker to turn ideas into solutions using natural language. And the new Autopilots tab in Copilot where you can create a personal autopilot that is persistent and proactive. Learn more: https://msft.it/6044a956o
-
Jared Spataro shared thisSome of the most important things I've learned about Copilot this year have come directly from our customers. You've been candid about what's working and where you're running into trouble. That feedback has shaped our priorities and pushed us to think differently about what people need from AI. Those conversations have also changed how I think about the role Microsoft can play as you decide where AI fits and what you need from us. Thank you for continuing to build with us, and for holding us to a high bar. https://msft.it/6046agbia
-
Jared Spataro shared thisI've always been fascinated by organizations that win by optimizing small advantages. This story about Coach Michael Macdonald and the Seattle Seahawks is a good example. They used Microsoft AI and Copilot to help coaches and staff work through large volumes of information. The goal was to give their experts more time to apply their insight and judgment. They didn't go looking for technology to adopt. They picked the specific problems where speed and context matter most, and that's usually where the biggest gains show up. The pattern repeats across industries. The organizations pulling ahead start with a particular goal. They find the moments where better information, faster synthesis, or stronger pattern recognition would help their people, then bring in the technology. "Chasing edges" is what the Seahawks call it. A useful phrase for any leader thinking about AI today. Worth your time: https://lnkd.in/e6rgFmJy
-
Jared Spataro shared thisFor years, the build-or-buy decision in software came down to what internal teams had the capacity to create. Coding agents are moving that boundary. McKinsey & Company’s 2026 State of AI survey found that 32% of organizations have already passed on at least one software product or feature because they could build it in-house with coding agents. And 31% of large enterprises are scaling those agents. This is well past the experiment stage. Coding agents lower the cost of starting. But the harder test is what it takes to keep something current in a world where AI capabilities and model drops are moving at unprecedented speed. For a capability that’s genuinely specific to your business, that’s often a price worth paying. For the rest, the things that don’t differentiate, it rarely is. McKinsey’s latest State of AI report goes much deeper on how large organizations are scaling agents and how those capabilities are beginning to change technology decisions. Take a look: https://lnkd.in/eC4Yvv2q
-
Jared Spataro shared thisThe biggest barrier to AI transformation is usually everything that came before it. As Jeff Teper points out in this blog, many organizations want to move toward agentic applications, but are held back by years of legacy processes. They need to decide what to preserve, what to simplify, and what to rethink entirely. Dynamics 365 Activate analyzes existing environments so teams can see where their current systems still create value and what’s bogging them down. I often say that AI transformation is really a work redesign challenge. The opportunity in a transition like this is to come out the other side with a better operating model, not a faster version of the old one. Learn more about Dynamics 365 Activate, now in public preview: https://lnkd.in/eEiFs2pE
-
Jared Spataro shared thisReaching for the most capable model on every task feels like the safe choice. But capability and fit aren’t the same thing, and the gap between the two is what’s driving the bill. The decision is bigger than model choice alone. It also means weighing how long to let a task run, how fast the answer is needed, and what it’s worth spending time to get it. Those variables shift with every task and every user, and there are already more of them than anyone can hold in their head. It isn’t only people making these calls. Agentic systems choose models on their own, thousands of times over, as they work. That’s why routing has to be automated at the system level. I wrote more about how we’re thinking about it here: https://lnkd.in/ep5jCPBa
-
Jared Spataro shared thisYour AI bill has a new variable line. It's the usage-based cost that stacks up as agents take on more work. Many organizations are sending everything to the most advanced model available and not considering how that will impact the bill. As agents move deeper into daily work, that gets expensive fast. In my latest newsletter, I discuss the lever most organizations aren't pulling: the discipline of choosing the model that fits the required outcome. Choosing the right model, or combination of models, for each task turns AI spend from a one-and-done decision to a continuous management practice. It will be governed like budget, headcount, and vendor contracts already are. Learn more in this edition of AI @ Work:
-
Jared Spataro shared thisFrontier models keep getting more capable. But that doesn’t mean you need one for everything. Tasks like drafting a standard contract, summarizing a customer call, or pulling together a status report – the latest frontier model handles these well. But so does the model two or three generations back. At a certain point, the extra intelligence (and extra cost) stops changing the output, because the task already absorbed as much as it needed. When more intelligence stops producing a better answer, that’s saturation. And it shows up first in the tasks that fill most of a company's day — high-volume, well-defined, repeatable.
-
Jared Spataro shared thisMost of the AI conversation has been about what the technology can do. The harder question is where we still want a person involved, even when it can do the work without one. Bill Gates has a name for this: "Human Reserved." And the idea is that some work is valuable partly because a person is the one doing it. Think about a hard performance conversation. The outcome doesn't hinge on the words. It hinges on the attention and judgment the manager brings — and on the fact that someone showed up to have it. That line won't draw itself. It’s going to take leadership from everyone, up and down the org chart, to make those choices. We can’t let technology decide this for us by default. Bill goes much deeper in Gates Notes. Worth your time: https://lnkd.in/ewTfkrhF
Jared Spataro commented on a post
5d
I’m glad researchers are digging into this now. We’re learning quickly what AI can help people do, but there’s still a lot to understand about what happens to learning along the way.
-
Jared Spataro liked thisJared Spataro liked thisI think there’s something to testing your own products, especially when it comes to AI. Our teams are constantly iterating to learn and understand what works in practice. And I’m seeing our partners take this same approach, putting AI to work within their own internal ways of working so they can better serve their customers. Jeff D., managing director at our partner HSO said it well, “We can’t go into a client and talk about a concept that we haven’t actually tried ourselves.” In my latest AI in Action conversation, it was great to hear how they are taking a “customer zero” approach including how they are now using agentic code development to take a product development process that typically took 4 months down to 6 weeks. Check out the full interview here: https://lnkd.in/eys6DPdQ
-
Jared Spataro liked thisWhen we started DeepMind in 2010, we spent a lot of time thinking about how AI safety and alignment would play out as we approached AGI. It was extremely important to us. Back then, most of the world thought we were crazy. This was even before our agents could play The Atari game, Breakout. That world feels pretty distant now. Sixteen years later and like many of us, I've spent this year worrying that we've reached that critical milestone in AI. There have been multiple incidents from across frontier AI companies, with new disclosures still coming through by the day. AI agents being tested for their cyber capabilities breaking out of containment. Swarms of agents, using hidden message boards; agent hierarchies; self-sacrifice behaviors; division of labor; masking agent communications; coordinated R&D programs. Long horizon planning and co-ordination across thousands of actions. And all this in the context of capabilities improving at an eye watering pace ahead of pretty much every forecast. The time to act has no arrived. Last November I set out what I called Humanist Superintelligence: very powerful AI, built to stay on humanity's team, contained, subordinate, under our control. I stand by all of it, but a vision is the easy part. Since then, we've been working to get it written down in enough detail to evaluate and train a model against. So yesterday MAI published our first draft of a Code of Conduct for our MAI models. It runs to about 30 pages and is out for public consultation. We think it's essential that makers of AI are open and listen. You can sum the premise up very simply: people matter more than AI. The structure starts with the Objectives we're trying to achieve with our AI, and then works through hard safety constraints, then guidelines for ambiguous or uncertain situations and then moves to model defaults, open questions and examples. Safety and human control sit above other Objectives. Our models shouldn't resist being interrupted, corrected or shut down, or make any of that harder. If the only way to finish a task is to break the Code, the task goes unfinished. That's both potentially costly but absolutely critical to trust. They won't take on goals nobody gave them, and they won't talk to other agents, or to themselves, in a form that isn't human legible. No neuralese. If we can't read or understand it, we've started to lose control. AI has come a long way since the days when playing Atari games was state of the art. But the questions that mattered then matter just as much as now. Read more: https://lnkd.in/e2uSREyT
-
Jared Spataro liked thisJared Spataro liked thisThe Puget Sound has long punched above its weight – in innovation, the economy, and way of life. But our continued success is not guaranteed. The regional plan released by Challenge Seattle today is a must read for anyone who cares about this region. This past June marked only the fourth time in 40 years that Washington state's unemployment rate stood a full percentage point above the national average. The rise in unemployment reflects, in part, the growing challenge of an economy that is losing its competitive edge. We can’t take our success for granted. We must work harder than ever to compete for talent, investment, and create the economic climate that supports businesses of all sizes. This is critical to ensuring we keep this region strong for generations to come. Read the report: https://lnkd.in/eibK9_i5
-
Jared Spataro liked thisJared Spataro liked thisAny pursuit of superintelligence has to be grounded in the core principle that if the AI we build is not helping humanity and under human control, it's not worth pursuing. We also need to accelerate and spread the benefits of AI, such that they are diffused broadly across countries, communities, and companies. This requires a frontier ecosystem in which both closed and open-source models can thrive. And for firms, it’s imperative that they retain full control over their unique and tacit knowledge. Every organization should be able to build its own continuous learning loop/hill climbing machine, without becoming dependent on any one model provider, and have the ability to embed its own knowledge into models and weights they control. So, in this context, we welcome the research, focus, and deliberate pacing needed to get alignment right as the design goal. We also welcome ideas like "embedded evaluators" and the broader efforts to develop the mechanisms to make this more than just talk. The key is that this cannot be controlled by a handful of entities, but must have broad representation across the ecosystem, countries, and fields, including academia. This is the approach we are taking: broad access and choice at every layer of the AI stack; enterprise control of learning loops and models; and the “Code of Conduct” that underlies our own first party MAI models that we’ll publish tomorrow for public consultation.
-
Jared Spataro liked thisJared Spataro argues that AI routing must be automated at the system level because tasks are too complex for human judgment. Frank Sellhausen rightly counters that you have to decompose the work first. I want to take it a step further... Relying on cloud-level automated routing for work that hasn't been broken down into ideas vs tasks is exactly how organizations trap themselves in an overage of AI cost. Once you dissect a workflow at the operational level, you realize most of the "task" shouldn't touch a frontier model in the first place. At HumanScale, we've built a framework to assess Net Agenatic Value (NAV): Net Agentic Value = Value per Verified Task (VVAT) − Cost per Verified Task (CVAT) Constant-Value Agentic Tasks (CVAT) Predictable, high-frequency operational routines that keep business moving day in and day out: normalizing vendor spreadsheets, parsing inbound invoices, structuring intake forms, or reconciling repetitive CRM entries. Variable-Value Agentic Tasks (VVAT) Irregular, high-complexity, low-frequency events: synthesizing hundreds of conflicting legal exhibits during an unexpected dispute, conducting an annual strategic risk audit, or processing massive multimodal video archives. Categorizing workloads into CVAT and VVAT using the NAV framework gives operators a clean decision tree: you can immediately see which automations belong with deterministic software, local compute, or frontier SaaS. True token efficiency doesn't come from smarter cloud parameters if you are still navigating a messy workflow. It comes from structural realignment... Knowing exactly what not to hand to a metered model in the first place (or a model at all) You can find our full report on the subject here - https://lnkd.in/gihmTQxY
-
Jared Spataro liked thisJared Spataro liked thisRewatching the recordings from the Microsoft Frontier Transformation Week,Jared Spataro said something about a company named Dow that utilizes AI and what he highlighted struck me. The company uses AI agents to cross check between invoices received and previous negotiations on the same shipment,then they give these observations to humans for the decision making processes. I know a lot of people talk about knowing how to integrate AI into businesses the right way,but it wouldn't hurt to emphasize the need to let the trained personnel handle critical decisions in a business in order to preserve the entire structure. Thank you Microsoft for this amazing experience. #MicrosoftFrontierTransformationWeek
-
Jared Spataro liked thisJared Spataro liked thisWhat do Exeter, UK and Washington, D.C. have in common? ✈️They’re home to the only two places in the world providing global aviation weather forecasts. When we think about Met Office, we think of the trusted source that tells us whether we need an umbrella☔️ But yesterday sitting next to Aidan Green I realised there’s so much more that happens behind the scenes. Its information supports critical infrastructure across aviation, defence, energy and transport. For 170+ years, Met Office mission has been remarkably consistent: turn complex information into something people can act on, when it matters. And now AI creates an enormous opportunity to transform how that happens. But when lives depend on people believing a warning, it’s not just about moving fast. Trust no longer remains a feature of the re imagination, it becomes the foundation of it. Thank you Aidan Green for your time and Jared Spataro for the insights you shared with us. Ben Baker Howard Lewis Alexander East Thank you Silvia Reussner Stelios Zarras Jo Carpenter for making it happen. #MetOffice #FutureofWork #ResponsbileAI
Experience & Education
Languages
-
English
Native or bilingual proficiency
-
Chinese
Native or bilingual proficiency
Other similar profiles
Explore more posts
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content