🦄 Making great presentations more accessible. This project aims to enhances multilingual accessibility and discoverability while maintaining the integrity of original content. Detailed transcriptions and keyframes preserve the nuances and technical insights that make each session compelling.

Overview

📖 AWS re:Invent 2025 - Driving Intelligent Project Portfolio Visibility with AWS + Smartsheet (BIZ210)

In this video, Olen Ronning, Director of Product at Smartsheet, presents how AI transforms project and portfolio management through Intelligent Work Management. He demonstrates capabilities like generating project plans from documents in minutes using guided prompting and templates, AI-powered columns for automated analysis, and Smart Assist—an AI copilot that analyzes risks, creates executive summaries, and automates status updates. The platform features agentic AI that proactively monitors portfolios, alerts managers to resourcing risks via Microsoft Teams, and maintains RAID logs automatically. Built on AWS using Amazon Bedrock, Claude models, Neptune for knowledge graphs, and Agent Core Runtime, the system includes responsible AI governance with guardrails and full auditability. The solution enables continuous learning through a virtuous cycle where insights feed back into the knowledge graph, improving efficiency over time.

; This article is entirely auto-generated while preserving the original presentation content as much as possible. Please note that there may be typos or inaccuracies.

Main Part

Introduction: The Challenge of Portfolio Management and AI as a Solution

Welcome. I’m Olen Ronning, and I’m a Director of Product at Smartsheet. I’ll be talking about how we’re bringing AI to increase visibility into your project and portfolio management. Before we dive into all of that, just a quick intro. Who the heck is Olen? Kind of by the numbers. I’ve been at Smartsheet for six years, and I’ve spent a bunch of time in products. Before that, I was a designer, and I studied a combination of computer science and industrial design. I still scratch the itch on industrial design once a year for Halloween and build these displays of different movie themes.

Can anyone guess what this movie theme is for this year’s Halloween? I hear it. Louder. Tron. Yes, indeed, it’s Tron. So I use this as a note from before where I love this quote from Steve Jobs. He said many years ago that he saw computers as a bicycle for the mind. And I think about AI as a light cycle for the mind. It’s just going to go so much faster and get you even farther than ever before.

So let’s talk a little bit about how does that apply to portfolios. How many people here are managing projects or portfolios? Show of hands. Okay, cool. How many, I’m going to ask you, if you look across your portfolio, I’m using this frame of seeing into the void, right? There’s so much out there. When you think about managing those projects, managing those portfolios, how many of these things resonate with you? Anybody, does this resonate in terms of issues and things that you’re trying to do? Is there something that’s not on this list that should be on this list? Yeah, kind of matches. I’m getting kind of side head nods.

So what we’re thinking about is how can we leverage the power of work management tools coupled with AI to make it so much easier to get to those insights faster and ahead of time and more proactively rather than reactively. And so when we think about what are the specific challenges related to the ability to dig up those insights and identify risks, there’s a couple of different factors that come into play there. So one is obviously the lack of visibility. And that visibility is both in terms of what’s going on in the work, what’s the current status of it, who’s working on what, and what are the resourcing availability of the teams to do the work that we’re trying to do.

Having to constantly do all those things we just saw in the last slide where you’re trying to hunt down the different updates and statuses and things like that means that you’re working off of data that’s not up to date, or you’re constantly spending a bunch of time tracking it down. And then you’re missing risks and work gets delayed, and the actions to mitigate those risks come too late before they can actually have an impact, right?

The other thing that we see is that a lot of people who are, maybe this resonates with you, where you’re managing your portfolio, you have a lot of processes that map into that portfolio, right? So you have work that’s coming in. You need to triage or manage that process or go through an approval process. And that requires a lot of human oversight and manual processing and manual checks. And that’s just a lot of time that you’re spending to do that. And that’s time that you’re not spending actually doing the project delivery work, right? It is extra time that gets consumed, and you’re not being able to maximize your capacity.

And speaking of capacity, if you think about, I’ve talked to a lot of customers where, actually I was just on a call earlier today with a customer where they had three project managers. And how many projects do you think they’re managing at a time across three project managers? Two hundred projects across three project managers. So their life is just constantly looking at all these projects, right? And so we see this opportunity to use agentic AI to be able to scale your workforce.

If you can do all of those things you just saw in the previous slide, if those things are happening in the background, then you can scale out. You can actually manage such a large portfolio with a smaller workforce, right? And you can then begin to take on more work and scale out your capacity. So that’s what we see as the opportunity that you can go from getting those insights, and those insights then drive specific actions that you can perform. And those actions are not just necessarily things that you are doing, but you can actually then have the AI perform those actions on your behalf.

And then the second part is by being able to apply AI to those automations, to those workflows, you can scale your productivity. So you’re spending less time doing those manual checks, manual approvals, manual triage, or having a human at different stages of a process. You can automate the entire end-to-end workflow and save time. And as I said, scaling out your workforce by having these agents that are performing a lot of these duties on your behalf, and you basically

have doubled or tripled or quadrupled each project manager’s capacity within your organization. So in the end, what this really means is with all of this data, if you can’t see what’s going on, you’re not going to be able to manage it. And so our mission here is to give you that insight so that through a project management tool powered by AI, you have that broader horizon view out across everything that’s going on, not only current state, but past and predictive future.

Intelligent Work Management: Building on Data and the Knowledge Graph

So let’s talk a little bit about what that looks like. We are calling this Intelligent Work Management. As an evolution, we think about collaborative work management. It is intelligent and powered by AI at the very core. And to support that, at the foundation is data. You have to have really good data. And a collaborative work management tool like Smartsheet is a place where you can have that shared source of truth, whether you’re managing projects directly in it, or you’re integrating with other tools or bringing in tickets from Jira or whatever other tools you’re using for managing day-to-day work, so that you have this source of truth where you’re getting a sense of everything that’s going on with the portfolio.

And we are building, we have built a knowledge graph on top of that data that you have in Smartsheet. And what that knowledge graph does is it analyzes the connections between the people and the content and the work, and it uses those relationships to provide powerful context for the AI. So today in Smartsheet, there are a number of AI capabilities we have already launched. And so if you are currently a customer, any show of hands, is anyone currently a customer of Smartsheet? Okay, a couple. You don’t count.

So you have these abilities to use natural language to generate really powerful formulas that can calculate different things like budgets and risks. You can generate charts by asking questions, and you can convert those charts into dashboards. Where we’re going, though, is rather than thinking about individual AI features, we are thinking about, like I said, this Intelligent Work Management platform, and that works across the entire lifecycle. It’s a foundational piece across every part of the app that you’re using. So whether it’s from setting a strategy all the way through to execution and delivery, you’re using AI processing to improve planning and prioritization, you’re intelligently optimizing plans, you’re continuously monitoring execution, and then reporting out on that delivery.

So what does that look like? Here’s a few examples. I’ll walk through it, and you’ll see along the bottom how we’re going through that journey, that end-to-end lifecycle, right? You start with just being able to define what is it we’re going to do. I was talking to a customer where they talked about how they were spending weeks or even months sometimes just developing a plan before they even know that they have the capacity to take on that plan, right? But they have to figure out what are all the work items you have to do, what’s the resource capacity to do that. That’s a ton of time, and then that goes to planning, and then it might not even get approved or it may not meet the bar for work that gets actually done.

We see an opportunity to leverage guided prompting, templates, and document extraction and processing. So you can say, here’s a project brief I got from a client, or here’s a statement of work that I am working on, and here’s some instructions. And also, can you pull in best guidelines for this type of project and use this template that we have designed as the way that we work in Smartsheet. Take all those things and build out a fully functional project plan, dependencies, timelines, everything in minutes, as opposed to doing all of that legwork manually over several weeks.

I forgot to mention that that capability, there’s an early version of that capability already live in our trial experience. So if you go sign up for a Smartsheet trial, you’ll get a preview, not the document piece, but the other parts. And we’re bringing the document part of that to all our customers in the near future. So the other piece we talked about is once you get all those projects in, you have these proposals, you’re trying to triage that. How do you prioritize that? Again, that’s a lot of manual processing, but sometimes there’s ways to look at, can we just do an analysis of the budgets and the costs and is it aligned to an OKR or a program that we need to deliver on.

And we are bringing AI-powered columns that are processing data directly in your sheet using an LLM. And that allows you to do nuanced analysis. So you can do a sentiment analysis, summarization, extraction, et cetera. Imagine performing any kind of LLM prompt-based capability and automatically running that on every row in your portfolio intake. The really exciting thing is once you have all of these prioritized items, you can start to think about, okay, what does the future of that look like? What does that roadmap look like? Do we have the capacity to take on that work?

Right now, we have in beta a capability we call Scenario Planning, and it is a portfolio-level view. It is a view of all your resources, all your projects, side by side, and

you can see in real time how shifting work out is going to impact the capacity you have. We’re going to layer on to that AI capabilities that will then build up different scenarios. So it could be if I optimize for balancing my capacity versus optimizing for cost versus optimizing for timely delivery, I can map out different scenarios and compare those and have a really data-informed decision about what we’re going to work on across our roadmap and portfolio.

Smart Assist and AI Agents: From Monitoring to Continuous Improvement

So once you have that work going, you’ve decided, you’ve prioritized the work, you’ve got a bunch of projects in flight, how do you know things are going well, right? That slide I put at the beginning of the word cloud, right, all that work to just track things and get things up to date, get people to update things and connect to things and understand the state of things. We are bringing what we call a Smart Assist experience. And this is an AI copilot experience that is tuned for project portfolio management, and it allows you to do things like analyze for risk within your project, get summaries of recent activity, create executive summaries of progress, like a progress report that you can share out with your executives, and be able to do all those things just in a few seconds and across multiple projects.

So here we’re seeing it in the context of a project. We also plan to bring this to a portfolio level view. So you can imagine, instead of a singular project, I’m asking these types of questions across my entire portfolio or a subset of my portfolio. The other part, like I said, is keeping track of those updates, right? So Smart Assist actually has a lot of really powerful capabilities to remind people to update things, to pull in updates from different tools. And this copilot can automate a lot of that process. So for example, here we’re seeing, you know, build me a daily standup where it’s basically aggregating people’s updates directly from whether it’s Slack or Teams. It’s summarizing those, highlighting risks, pulling those out, sending you a summary update, and you’re able to not have to go do that drudgery work, right, of collecting all the information. It’s doing that for you automatically in the background.

And when we think about combining those things, right, automated actions, plus the kind of project expertise and risk analysis and all that, we’re bringing those together into agents, and agents then can perform a lot of that work for you in the background automatically. So rather than you going to log in to Smartsheet and saying, hey, what’s the risk in this project, this portfolio, it is continually monitoring that in the background, looking for changes to the risk profile of different projects, doing heuristic-based analysis of how urgent is that risk, is it really a high risk or a low risk, and then sending you alerts and warnings here. So this little demonstration is showing a project manager getting an alert in Microsoft Teams that there’s a resourcing risk within their project. And so they don’t even have to be logging in and monitoring. It’s monitoring it for them. It brings them in, shows them some options, gives them the opportunity to reassign work, address that risk, take action, and it can actually perform the action on their behalf.

And this is my favorite part here, where at the end of that, it’s also maintaining a RAID log. How many people know what a RAID log is? A couple of people, okay. So it’s like risks, actions, issues, and decisions. This is a great way, it’s a good project management practice, but it takes a lot of diligence and to remember to do that every time, like, oh, this changed, or I’m going to log that. So the agent can track project changes, like when dates shifted, when risks occurred, actions were taken, and it’s maintaining this RAID log for you. And that is going to help inform continuous learning and improvement over time.

And so when we think about how that portfolio lifecycle comes to a close, if we’ve successfully delivered on the project, you know, projects never run perfectly, right? So you have those learnings, you have those RAID logs, you have the data that’s been collecting and it’s feeding back into the data, into that foundation of what’s happening across your portfolio. We can start to run analysis on patterns and trends that are happening across your work. So we can use, for this example is saying, hey, we’ve noticed this common pattern within your portfolio where you have this issue, and we are recommending based on best practices and other customers who are in similar domain as you, that they have implemented this practice within their project management solution. And because we have the ability to deploy changes to your solution design for your portfolio, you can immediately deploy those changes out, even to projects in flight and make those changes. So you’re continuously improving the work that’s going on.

All right, so bringing it all back together, we’d look at the project life cycle from setting the strategy through execution all the way to delivery, and looking at that across the portfolio. This kind of a little bit of a mess of a diagram here is showing how we walked through each of those stages, and you could see how agents and AI are powering each stage of that process. But at the end of it, you’re getting these postmortem learnings. So yes, you’re getting value on day one, but the idea is that all of those learnings are feeding back into that data and it’s a virtuous cycle. So every time you go through this loop, you are increasing the knowledge that the knowledge graph has, and that is improving and increasing the level of sophistication and insights tailored to your business so that you are continually improving and increasing the efficiency and throughput of your portfolio of work.

AWS Foundation and Responsible AI: Technical Architecture and Future Vision

All right, well, I can’t really talk about this unless I mention AWS because we’re here at the AWS conference, of course. So how is all this working? A lot of this that I showed you, some of it we have, that Smart Assist copilot experience is in private beta for some of our customers today and is coming out to our early adopter program later. Those other capabilities are in development, and all of that is built on AWS as a foundation. So we think about these, I’ve kind of grouped this more into conceptual layers to think about. At that foundation is that data. We have that knowledge graph, the knowledge graph is being powered by Amazon Neptune. We have, we’re building conversation history and long-term memory, and that is using a PostgreSQL database. We have, we’re building out a data lake, so we are aggregating and collecting all of that data and bringing that together, and that forms that data foundation again that you’re cycling through.

On top of that, we have agent tools. So these are the actions that the agents can perform, and these are both tools that are available inside the product so that the agent is running those tools within the project that you’re managing. But we are also going to be exposing those through an MCP server that allows you to access them through external AI services. And those agent tools, we actually started this process before Agent Core was a thing, or at least before it was available, so we’ve built a lot of this, some of this kind of from scratch. And right now we’re going through that evaluation of like, oh, maybe there’s a way that we can leverage a lot of the capabilities that we just got announced here today, or this week. So we’re currently evaluating Agent Core as part of that experience, specifically Agent Core Runtime for these tools.

Then you have the agent orchestration layer. So this is where we have different subagents that are specialized. You have a subagent who is tuned to understand resource management, and a stellar subagent is tuned to look at risk profiles and evaluating risk. And those subagents then are, we’re using Amazon Bedrock and the Claude models built into Amazon Bedrock, and using the Strands SDK to then build and define how those agents are operating. And those agents then deploy those tools as needed to take action. So we talk about this as insight to action. So the agent is monitoring, looking for risk, looking for challenges, identifying like stale data or missing data, and then deploying these tools to take the appropriate action to fill in the data, get updates, flag risk, change resourcing adjustments, et cetera, to take action on that.

And then on top of all of that is a responsible AI layer. So this is governance across all of this. You can’t just have the AI running wild. You need to be able to observe exactly everything it’s doing. So we have guardrails, which is managing moderation of inputs and outputs. You have OpenSearch to track logs and auditing, so there’s full traceability and auditability of all the actions that are being performed. And we have an administrative layer experience. So our Smartsheet admin experience has data access controls, user access controls, and feature enablement, disablement, and overall auditing across all the actions across, not just, you could actually have multiple portfolios operating within Smartsheet and you can observe that at the administrative layer.

And so what does it mean that we have tools? Talk a little bit about that for a second. We’re building all of these out today. You’ll see kind of a gradient here. The tools that are full opacity, as it were, on this slide are available as part of our Smart Assist experience, and we’re continuing to add to those. So these are, again, going to be available inside the product.

They enable a lot of capabilities inside the product, not just existing features, but new ones. We’re going to be enabling those through an external MCP server so that you could take advantage of, for example, being able to call a Smartsheet agent externally through the MCP to perform actions on your behalf.

And when we think about responsible AI, that layer on top really has four principles that we see. First is data privacy and security, so that’s control. Second is accountability, so that’s being able to see everything that it’s doing, that you can govern every action that’s being performed. Third is transparency and explainability, understanding each action, why it’s recommending actions, and why it performed that action. And then making sure that it is providing fair and reasonable responses to everything that you’re inputting.

So let’s wrap it up. It looks like I’m out of time anyway. The future of portfolio management looks like this, right? From hours spent building out project plans to minutes turning a doc straight into a working project. From stuck processes to intelligent end-to-end flows. From manually tracking down and hounding down a particular person, you have agents automatically updating, tracking, and responding to those stale data or updates that are required. And instead of just the status quo, you’re getting this continuous learning and virtuous cycle of new insights that are feeding back in, and you’re improving your process as you go.

And that’s it. Thank you very much. I really appreciate it. You can come visit us and learn more at our booth. I think it’s 586. Yes, perfect. And if you want to scan this QR code, this will take you to learn more about our AI features and our white paper. All right, thank you very much. Thank you.

---

; This article is entirely auto-generated using Amazon Bedrock.