Ryan Alshak

CEO / Returning Time.

[Ryan Alshak @ryanalshak]

You cannot manage (or automate) what you don’t measure.

Managing your time is the single most powerful use case for AI.

Time is the ultimate platform.

[delian @zebulgar]

Damn I am extremely Claude CoWork pilled

I think I can probably cut 5-6h of my work per week with this

and the @rabois-style calendar audit can now happen automatically

List out my top 3-4 priorities, and Claude can check if my time allocation matches to those

So cool

[Ryan Alshak @ryanalshak]

Perfectly said @Alfred_Lin

Time will always be the rate limiter so invest accordingly.

[Alfred Lin @Alfred_Lin]

https://t.co/fnQEM0WuFJ

[Ryan Alshak @ryanalshak]

The obsession with AGI is a distraction from the AI we already have and are already failing to deploy.

The most sophisticated minds in technology are debating when machines will become generally intelligent, what values to instill in them, and how to govern their emergence. This is an interesting philosophical question. It is not where the leverage is.

We already have AI that can accelerate drug discovery. We are not deploying it at scale because the FDA approval process was designed in 1962. We have AI that could transform K-12 education. We are not deploying it because teachers unions and school boards have structural incentives to resist it. We have AI that could rebuild how governments process information. We are not deploying it because procurement cycles run on 7-year timelines.

In professional services, we have AI that can pass the bar and the CPA exam. We have not disrupted the leverage model because pricing and packaging have not meaningfully updated since the 1960s.

The bottleneck is not model capability. It is change management, regulatory permission, and incentive structure that was created for a world that no longer exists.

AGI discourse is, in part, a way for the industry to avoid accountability for the much more tractable and much more relevant question: why haven't we done more with what we already have?

[Ryan Alshak @ryanalshak]

Thank you @SFBusinessTimes for the recent feature.

The headline will change imminently 👀

But we’ve never built Laurel for valuation.

What keeps us Day 0 is our mission, our vision, and the opportunity to improve our customers’ lives.

To our team and our customers: this is yours.

None of it happens without the people who showed up every day before it was obvious.

[Ryan Alshak @ryanalshak]

The AI consensus is backward. Everyone is asking: “Which model wins?”

The real question is: which enterprises best leverage their proprietary context to get ahead of their competition?

This is why the Kirkland and Freshfields news is interesting. Because they recognize their edge is in the context, not the models or infrastructure.

Asking which models win today is like asking in 1999 which browser would win.

Google was not valuable because information became abundant. Google was valuable because they built the world’s greatest web graph.

Facebook was not valuable because people came online. Facebook was valuable because they built the world’s greatest social graph.

The narrative violation is this:

The winners in AI won’t differentiate on models or infra, and definitely won’t differentiate on the app layer.

They’ll be the ones that build the world’s greatest work graph — the structured map of how enterprises actually operate: who knows what, who decides what, where work flows, where it stalls, what patterns precede outcomes.

This is what Kirkland understood with its $500M bet. They're encoding 30 years of how the most consequential deals in the world actually get done — every negotiation pattern, every decision node, every risk inflection — into a proprietary graph that they don’t want any model company getting their hands on so no competitor can buy it.

The model companies are already commoditizing. The context layer — the work graph — is where the compounding value accrues.

[Ryan Alshak @ryanalshak]

This is the game on the field.

Look out for what we're working on in our Act 2.

Built on years of proprietary data, client-side real estate, and data labeling flywheel from our Act 1

[Garry Tan @garrytan]

This is the actual bottleneck. The models are smart enough already. What is missing is the company-specific context locked in senior people heads. Whoever cracks knowledge extraction at the company level unlocks the rest.

As you work on this, please consider using GBrain as your OSS retrieval layer

https://t.co/0F5uDQzPHu

[Ryan Alshak @ryanalshak]

We wanted to know how our Power Users were leveraging Laurel. The top 20 billers across our platform averaged 794 billable hours in Q1 2026.

Annualized: that’s over 3,000 hours. Industry target is 1,700 to 2,300.

That gap isn't because these professionals are working more. It's because they are getting the credit for the work they were already doing.

[Ryan Alshak @ryanalshak]

Thank you to @SFBusinessTimes for the profile

[Ryan Alshak @ryanalshak]

We in the AI industry are measuring the wrong thing.

We track tokens, parameters, benchmarks, model rankings. We argue about AGI timelines. We almost never measure the only number that actually matters: the hours of human life on the other side of the screen.

A 40-year career is roughly 80,000 working hours. If 30% of that is meetings that shouldn't exist, documents nobody reads, and status updates engineered to demonstrate effort — that's three full years of a human life surrendered to coordination overhead.

AI is the first technology in history capable of giving those hours back. Not "saved" the way email saved time by inventing new channels to monitor. Actually returned.

The companies that win this decade won't be the ones with the best models. Everyone will have those. They'll be the ones with the clearest answer to: what is an hour of a human's life worth, and are we wasting any of them?

[Ryan Alshak @ryanalshak]

AI hasn’t (yet) delivered on its promise to enable humans to work less. Talk to anybody working at an AI-Native company right now and the ground reality is the exact opposite. William Stanley Jevons would be proud.

But what it is doing is revealing how much of what we called "work" was never work to begin with.

If a workflow is repetitive, requires no judgment or creativity, and exists only as an artifact of human coordination — it was never designed for people. It was designed for systems that happened to be staffed by people.

AI produces these artifacts in seconds. And we finally get an honest answer to a question the knowledge economy has been avoiding for decades: was any of this ever worth doing?

For most of it, no.

The AI-native company isn't the one that uses AI to do more of what it was already doing — faster decks, faster memos, faster meetings about meetings. That's just the same theater at higher frame rate.

The AI-native company is the one with the courage to stop doing the things AI just exposed as theater, and to redirect those hours into the few things that actually compound.

The bottleneck was never capability. It is the courage to admit what was always optional.

[Ryan Alshak @ryanalshak]

The best ROI a company can earn isn't revenue. It's when the customer who championed you gets promoted because of your platform.

One of our customers used Laurel as her case study in her promotion interviews.

She was a Controller. She championed Laurel. Both purchasing and deployment. The firmwide rollout was a resounding success — both the numbers and the qualitative feedback.

That gave her the confidence to go after a COO role. She walked into the interviews and used Laurel as her case study.

She got the job.

Her first move at the new firm? Bringing Laurel with her.

Laurel the product exists so knowledge workers can spend their hours on the things only humans do well. Laurel the company exists so knowledge workers can operate in their purpose.

This is what that looks like.

[Ryan Alshak @ryanalshak]

The easiest thing to build in AI right now: Agents.

The rate limiter in the enterprise: Knowing what the agent should actually do.

A model can write the email. But should it?
A model can summarize the meeting. But who is the audience?
A model can draft the invoice. But how do you describe the work in a way that increases the likelihood it gets paid on time?

This is the gap between demos and enterprise production.

Agents have intelligence without context.
Automation without observability.
Workflows without work data.

Everyone wants to automate the enterprise.
Almost no one understands the enterprise.

Whoever bridges that gap will win.

[Ryan Alshak @ryanalshak]

A 100+ year-old law firm in New England had sworn off third-party time-entry tools. Bad experiences. Broken promises. They were done.

Then their CIO found us through the ILTA community forum. Word of mouth. No outbound. No ad. Just one firm leader hearing about us from other firm leaders who were already using the product.

During evaluation, one of their technical leads said this:
"You guys are doing a lot more in the time entry space than the actual major time and billing system vendors are."

They signed firm-wide.

The ROI model showed over $3.7 million in potential revenue lift at 28 recovered minutes per person per day. Even at the most conservative 10-minute scenario, the math cleared a million dollars.

But what actually closed the deal wasn't the model. It was the fact that their CIO heard about us from someone who wasn't trying to sell him anything.

[Ryan Alshak @ryanalshak]

One of our timekeepers:

"Well, I will tell you this, in March, I had 130 hours billable, and in April, I had 153."

Worked exactly the same. The only change variable? Laurel

Yes, that’s an 18% lift in billable time. But much more importantly, "We literally gave someone a day back."

That’s what it looks like to deliver on our Mission to Return Time.

[Ryan Alshak @ryanalshak]

Everyone is worried AI is making us worse at our jobs.

The concern is legitimate. The conclusion is wrong.

Aviation faced the same panic when autopilot arrived. Pilots would lose their edge. Skills would atrophy. What actually happened? Flying became the safest form of transportation in human history. Not because pilots got worse. Because their job got better.

The best pilots today aren't the ones who hand-fly every minute. They're the ones who know when to trust the system, when to take over, and how to move between the two without thinking.

That redefinition is coming for every knowledge profession. Fast.

The best engineers won't write the most code. They'll architect the systems, orchestrate the agents, and leverage their customer and business context to ensure the right things get built.

The best lawyers won't bill the most hours. They'll decide what work matters and deploy the right resources to deliver it.

The best operators won't execute the work. They'll design how it gets done — and finally have the time to think bigger.

The professionals who thrive won't be the ones who resist the shift. They'll be the ones who master orchestration — who learn to direct agents, own the output, and raise the bar on what's possible.

https://t.co/BCLZiWhQET

[Ryan Alshak @ryanalshak]

CEOs are asking the wrong AI question.

The question is not: “How do we get every employee to use AI?”

The question is: “What work should still exist?”

That sounds harsh until you realize most organizations are filled with invisible work:

Status updates. Coordination costs. Administrivia. Work about work.

AI will not transform the enterprise by making every person 20% faster at broken processes.

It will transform the enterprise by making the work visible enough to redesign.

You cannot automate what you cannot see.

You cannot redesign what you cannot measure.

You cannot improve what you only understand through meetings, surveys, and folklore.

The future of work starts with a complete picture of work.

[Ryan Alshak @ryanalshak]

Wait until the world sees what Laurel is cooking. Principles:

-- surfacing work builds the map. FDEs become informed explorers

-- change management (i.e., what agents to use when) is now the rate limiter. Understanding work in real-time solves this

-- understanding work only happens by owning the desktop

[Allie K. Miller @alliekmiller]

The most expensive mistake in enterprise AI right now: treating FDEs as your whole transformation plan.

Forward deployed engineers (FDEs) are important for custom deployments, but they won’t fix the change management issue most enterprises are facing.

It’s likely more the former that Anthropic and OpenAI will continue to prioritize (and hire into the thousands, who knows). Beyond performance and cost, it’s systems integration, ROI, and literal usefulness that drive revenue and stickiness.

*However*

External FDEs, in my opinion, will not make your company an AI-first company.

You can have the sleekest multi-agent orchestrations and still have the majority of your employee base hating AI, avoiding AI, and distrusting leadership decisions on AI.

And we already know this because we see this in traditional SaaS too: you can customize the heck out of your Salesforce deployment, but that doesn’t mean your sales team will improve their data hygiene or even attempt to change the way they track and grow with it.

Buying a fancier car doesn’t mean you magically learn to drive better overnight.

If you’re an enterprise exec and FDEs are sold as the immediate and sole solution to your company transformation woes, walk away.

It’s the combination of tech *and* people enablement *and* process reinvention that compounds into actual business outcomes.

Large complex enterprises will stall out if they only prioritize the first.

[Ryan Alshak @ryanalshak]

Last week, we were invited to speak to 150+ enterprise leaders.

Here’s how the host introduced Laurel:

“Laurel is one of the sharpest live examples of category creation in enterprise AI. Their central insight is simple: The average knowledge worker puts in nine hours a day, but creates real leverage for three. That means six hours per person, per day, are waiting to be returned.

Laurel is one of the few companies that can actually measure the ROI of AI in the enterprise by showing how work changes before and after AI is introduced.

Companies are spending real money on AI tools. Most still cannot prove what those tools are doing.

Laurel can.

And their Work Intelligence framing may be one of the cleanest provocations any leader will hear this year: 25% of enterprise work is invisible to the systems meant to track it.”

The future of enterprise AI will be won by the companies that can prove how work actually happens.

[YesPress @yespress_io]

A lawyer who billed $650/hr is now building AI to end the 6-minute timesheet

Our cover story on @ryanalshak

https://t.co/33XtmsjxDg via Yespress

[Ryan Alshak @ryanalshak]

The next great enterprise AI company will be a system of record for work.

For 30 years, companies have built their operating models on self-reported data:

CRM: what salespeople say happened
HCM: what managers say people do
ERP: what finance says moved

But AI needs ground truth, not human guesses.

What work actually happened. Where it happened. How long it took. Who was involved. What systems were touched. What judgment was required. What should never have existed in the first place.

The companies that can’t see work will keep buying agents and wondering why nothing changed.

58% of enterprises have no clear AI ownership. 75% lack governance. And most can't even inventory what AI tools are running in their org.

The companies that can see work will not only automate it, but also understand how it should be done.