After three days loaded with plenty of AI and blockchain debates at Connected Enterprise 2017, it was time for a sanity break in semi-comfy chairs with Jay Goldman, Co-Founder and Managing Director at Sensei Labs.
Sensei Labs' "digital workforce solutions" veer heavily into automation and collaboration, so I wanted Goldman's gut check on the AI debates. I also wanted his best pitch on how their new offering, Conductor, can lift project management tools from the curse of the inbox and the spreadsheet.
Can project management become a living/breathing thing that actually - wait for it - makes people's lives better, rather than a stale, top-down plan that derails more often than not?
The great AI debate - ethics are lagging behind
Goldman's reaction the AI debates? We're venturing into a brave/uncharted new world, with ethics lagging behind:
We've been trying for centuries to come up with a universal set of ethical principles. The greatest philosophers in history have failed to do it, but now we find ourselves at a point where we have no choice but to do it - if we're going to build computer systems that behave ethically.
It's a really interesting dilemma, how do you encapsulate that into artificial intelligence, when we're not able to even articulate it for ourselves?
So how does Sensei Labs' obsession with "digital workplaces" connect to this tech? Goldman:
It's definitely a lot of focus on data and analytics and how to use it to get to better outcomes, so from the data side, the AI and machine learning side also fits into the picture of what we're doing. I think we're at the early days of understanding how machine learning can build better work places and drive better engagement for people. We sit at that point.
Why project management needs to change
The latest addition to the Sensei Labs platform, Conductor, has the ambitious goal of transforming project management. So how is project management broken?
The Project Management Institute has some numbers (PDF link) around what it costs the economy every year to do bad project management. It's into the tens of millions, closer to the hundreds of millions of dollars. If you take a lean approach to it and you look at the activities within the typical project management cycle, most of them would qualify as waste.
The stale project plan locks everything in:
So you have project managers who create very detailed project plans. Then they spend a lot of time trying to get people to look at the project plans. But those people are busy doing - they don't have time to look at the plans. Then they spend a ton of time trying to get status updates from people, so they can go back and have the plan reflect reality. In the end, they're stuck in this endless cycle of trying to keep plans updated that nobody looked at in the first place.
In other words: the real value-add is in the creation of the original project plan - but over time, that value deteriorates. So what's a better option? Goldman believes we need:
- "living breathing" project plans
- with early warning, alert-enabled detection systems
- those alerts should pull from collaboration data, not typically captured by project management tools
That's why they built Conductor:
We're trying to move a lot of the rest of that cycle [after the initial plan is built] into a tool in which the project work happens inside of the same tool that the project plan exists in. It makes the plan into a living, breathing representation of what's actually happening inside of the project itself.
So what about that missing data piece?
We also think there's a ton of data that people are missing out on. People have really good gut instincts about the projects that they're working on, and we tend not to capture that data. We often capture the quantitative data: "Are we on budget, are we on track?" But what we don't look at is the gut instinct of people. "How do you feel about how this project is going?"
But hold on - isn't there a danger in building predictive actions from gut reactions?
Yes, the individual gut instinct of one person may not be predictive, but when you get into a large project - and some of our clients who are early Conductor adopters are using this to launch billion dollar pharma brands and things like that - so you're talking about hundreds of people on a project.
That's the data that powers the "early warning" alerts:
In that case, you can very quickly, and without even actual machine learning, just through analysis, identify trends and patterns in peoples' gut instinct. Where is this project starting to go wrong? And if you can catch it at that point, you can often prevent something really bad from happening.
How does this impact real-world projects?
So how would this work in a customer setting? It starts by prompting users to contribute:
The reviews in Conductor actually happen from the project team. Once a week, everybody is asked to do their weekly review. The tool will adapt to the set of things you need to review based on business rules. It will say, "Hey, Jon, this project is due within two weeks. You need to provide a review. You haven't looked at his one in a month. How's it going?"
They can see their input alongside their co-workers:
The business rules dictate the frequency of the reviews. Then you'll put in your review alongside of everybody else, so you can see everybody else's review. Then we aggregate upwards from there, so the project lead can see the project team's reviews, and the work stream lead can see the project lead reviews, etc.
One problem: most users are not used to providing that kind of input. Goldman says Sensei Labs solved that by shipping a solution configured for transparency:
The Sensei Labs platform - even outside of the project management work - defaults to as much openness and transparency as we can. You can turn on business rules and permissions that restrict that, but by default, the project really tries to push people towards more openness and transparency. We've seen that happen in a number of our customers, where they have really adopted it because they know it drives more accountability, or it drives more transparency.
My take - the impact of culture, and conversational interfaces
Next time, I'd like to talk to a Sensei Labs customer, but in general, I'm an advocate of the "living document" style of projects where course corrections are easier. If you can automate alerts for warning signs in an intelligent - and not distracting - manner, so much the better.
I see some issues with culture fit - not every organization is ready for transparent work styles. Goldman challenged me a bit here, arguing that people aren't happy with their current tools, and they don't want to live out of the email inboxes (no doubt!). He told me about an early Conductor customer moving on from spreadsheet purgatory:
We have a few examples of a company that had a really complex spreadsheet. Take one of our pharma clients for example, a top 20 pharma company. They were managing the launch of a $750 million a year drug in a series of Excel spreadsheets... I think it was 27 countries, and they had 27 spreadsheets, one for each country. Not yet a cloud-based organization, so for them, that meant 27 copies of the spreadsheet in some version that you hope is the latest version, and you really did get versions that were version 13_final_updated sort of thing. The spreadsheet itself, you had to color code the rows based on the current status of that task.
Still, many people are reluctant to part with their home grown tools, however inefficient. Goldman acknowledged that the biggest challenge is adoption: The key? Demonstrating a quick user win: "If we're able to demonstrate value to people very quickly, then we tend to get very strong adoption."
After the show, the impact of voice-enabled interfaces heated up with the announcement of Alexa at Work. I wanted to know how disruptive Goldman thought voice was to workforce apps and Sensei Labs in particular. Goldman thinks voice is well-suited for certain tasks, but it won't dominate the workflow of the future:
It won’t become the predominant interface for all the reasons we increasingly find ourselves texting instead of speaking: it’s hard in loud environments, it’s not nearly as private, and people tend to think you’re crazy when you seem to be talking to yourself.
He sees voice as a huge UI asset in hands-free/focused situations like driving, or surgery. The key is that devices must be "multi-modal" to adapt throughout our workdays. That means SenseiOS and Conductor will likely adopt "interactive dialogs" for questions on projects, but voice will be one facet of future design.
It's a heady mix of interfaces - just one more challenge, but I'd put it far below user adoption. If Sensei Labs cracks that nut, they'll be a lot better off than most collaboration vendors are today.