Our Co-Founder & CEO Tarun Thummala reflects on the first half of 2026, while staying laser-focused on what's ahead for the PressW team.
You know, we've been awfully quiet the last couple months. A friend asked me the other day how things were going, I said fantastic, and he almost seemed surprised.
The nerve! (kidding) Apparently going quiet reads as either crushing it or quietly falling apart.
Thankfully, I'm happy to report it's the first one. And as we wrap up H1, which came faster than I'd like to admit, it feels like a good moment to share some wins and what's on our mind heading into the back half of 2026.
I had a feeling last year that 2026 was going to be the year for AI. It was the first year enterprises leaned in, in earnest, the year it stopped being a wait-and-see. That's played out. Enterprises have roared, and nearly every business we talk to is at least dabbling now.
On the technology, I'll say it plainly: the models have finally crossed an intelligence threshold that makes nearly every knowledge-based task automatable.
As far as we see it:
There's less than 5% of digital tasks that agents can't handle.
Sit with that for a second.
The gap now isn't the models. It's adoption, training, and the actual work of getting the systems in place to run that automation. (Hint: professional services.) The models work. We need to empower smart humans to deploy and manage them properly.
Our numbers reflect that fervor. Revenue is up 129% year over year, with bookings up nearly 350% against the same stretch of 2025. While I do believe part of that is because of the great work the team has been putting out over the last couple of years and the investments we've made, a portion of that is definitely just because of our market is waking up in a big way.
The growth is awesome, but if I'm being real with you, the growth matters to me mostly because it's let us keep bringing in great people to problem-solve & build. I am a fundamental believer that a business is all about the people. And when we set out to build PressW, we thought a lot about what we wanted our business to look like. We didn't want to fall into the traditional services consulting based model. We believed that AI could allow us to innovate and rewrite what it means to be a services business.
The tax nobody warns you about
Three and a half years ago we made a bet: that you could scale a services business on a different model than what came before.
Here's what usually happens when a services company grows. You land more clients, so you hire people to serve them. Then those people need managing, so you hire managers. Pretty soon you've got layers, and the layers have layers, and the thing that made you sharp in the first place is buried three orgs deep under process. The revenue chart goes up and to the right and the company gets slower every quarter it does. It's a tax. Most people pay it because they think it's the cost of scale.
We decided we weren't paying it.
From the inception of PressW, we've been building our firm as AI natively as possible. For the first 3 years or so we really saw the benefits on the coding agent side of things. Thanks to coding tools available and the work we put in to develop the framework and engine around them, our developers are consistently outputting 2-5x benchmarks we were seeing from other orgs. A project that used to take a team of 5, now could be done by 1. Faster. Better too.
But still it wasn't until this H1 that we really saw the AI operating system elevate to become firm wide. With the introduction of managed agents, we now have an incredibly robust agent driven GTM engine. Our customer requests, project management, and finance is largely agent driven and automated as well. More on this from me later (hint: future post), but for now - I see it all compounding extremely quickly and it makes us more excited than ever.
“Wait, but you're still hiring humans?
In H1 of this year, we nearly doubled our team size.
Huh? Didn't you just go on and on about how you're so efficient and are attempting to resist the typical drag that businesses like yours go through?
Yes - it's actually an interesting dynamic and a bit counterintuitive. As we've enabled more and more of our organization, our possibilities and opportunity have exploded. Of course there is an undeniable need for more bodies as we've grown the services side of our business. But more interestingly we've gained the separation needed to properly invest in productizing, platforming, and building teams of 1 that handle entire departments.
A quote from our founding AE, who just joined a few weeks ago and has been working on enriching our TAM using our agents:
…manual enrichment for our entire TAM would take one person over a year this will take a week. This is not just a game changer, this is literally a different game altogether.
We've emulated the structure we've seen from leading organizations like Anthropic, Ramp, and Shopify. The meta now is to give someone a quest, give them supplies and weapons (agents), and let them run headfirst at the problem. No more building big teams, planning, waiting around. See the problem. Launch a missile at it in the form of tokens and human agency.
The bet we actually made: Managed AI Ops for the future of Agentic Business Ops
The pattern's not complicated. When you stop making people the glue between your tools, they stop spending their day being glue. They spend it on the big vision, the judgment calls, and high ROI activities, which is what clients should be paying for. The future of every business will eventually center around agentic operations sooner than later.
To that point - PressW has always been laser focused on the why rather than the how. Meaning, we've always been fundamental believers that AI and technology are just a means to an end. We must be laser focused on the business outcome that we're driving towards, and then fixed on that, figure out the best deployment to get there.
In H1 we noticed a gap in our existing offering. What we kept running into was simple: most companies don't actually need a bespoke application.
When we build something fully custom (that's our transformation engineering line), it can run six figures, sometimes well past it, and it's usually pointed at one exact use case. If you're a mid-market or larger enterprise operating at scale, that math works, and we do a lot of it. But for a lot of organizations, it just doesn't. They don't need us to build them a shiny new thing. They're just deploying Claude to their organization and they need help with where they already are:
- Training and enablement: teaching their people how to actually make AI work for them
- Security: making sure the AI doesn't leak their data or nuke their database
- Connectors: wiring the system into the other tools and data they already run
- Observability: seeing how the org is actually using it, so token waste and blind spots don't pile up
- Encoding and automating the workflows that eat their team's time
What they really need is a team of experts to empower and accelerate their team. A true technology partner that knows how to work with these foundation platforms like Claude and other tools and take all the burden of implementation off their team.
That's it. So that's what we built. As an official Anthropic partner, we become your embedded AI team, the group on the hook for taking you from where you are today to something a lot more AI-native, deployed on top of the tools you already have instead of something expensive built from scratch. (For us that platform is Claude. Honestly, most of the major ones would do.)
That's the whole idea: strip out everything an organization doesn't need, and just give them the team.
The scorecard six months in:
- 22 MSP clients onboarded in five months, from a standing start in February
- Zero churn to date
- 7 figure annualized run rate achieved, up from $0 in January
- MSP bookings up 217% quarter over quarter
This has also led to a big shift in our thinking. I went in thinking of our AI Operations division as going to be one offering among several, but six months of conversations and work have changed my mind. We're putting a stake in the ground - this is becoming the thing the rest of the company is built to support.
The PressW numbers (because I know all of you are nosy and I don't mind sharing)
All versus 2025:
- Clients / engagements: we added 26 net-new clients in H1 (31 year-to-date), plus five existing clients who expanded into new 2026 engagements.
- Headcount: 13 → 25 (21 billable), and each person is doing what used to take a small pod.
- Bookings: up 349% YoY. Our first half out-booked all of 2025 by 60%, with two full quarters still left to sell.
- Full-year growth: 91% already locked in. FY26 contracted revenue is nearly double all of FY25, before we've booked a dollar of H2.
- Momentum: bookings up 78% quarter over quarter, and monthly recognized revenue up 58% from January to June.
- Q1 revenue up 151% YoY, and June alone came in +150% over last June.
You'll notice I'm not leading with a revenue dollar. That's on purpose. The dollar isn't what I care about here. What matters is the shape of it: we roughly 4x'd the output and nowhere near 4x'd the overhead to get there. Once you've seen a company run that way it's hard to unsee!
The part I'm actually proudest of isn't the software
If I let this whole update be about the tech, I'd be telling it wrong.
The reason any of this works is the people. Specifically, it's that the folks here at PressW will happily throw out a workflow they built last month the second they find a better one, and then write the better one down so everybody gets the upgrade for free. That habit compounded harder than any model we plugged in. One of our newest AI strategists leapt into the company and was part of 8 different clients within a month. The onboarding was steep. But he drank from the fire hose and built out an auto-ticketing and triage tool that helped him manage the requests effectively. Within a month of joining. Now everyone on the team is using the same system to improve their own client management.
What still isn't figured out
I don't want to make this sound cleaner than it was. It's been exciting no doubt, but of course it's been messy to get here. We've tried dozens of ideas at this point that have never seen the light of day. We've created full products that we used and then had to throw away as the market shifted. I imagine this messiness will continue to occur as the market we play in is moving so rapidly and we learn to scale.
This last quarter we as a company had to realize the complexity that arises when you functionally are building two businesses at once (our PressW Labs side vs. our Managed AI Ops side). Trying to balance delicately how to use our stable business to help fuel the new business has been complicated to navigate from partly a pure business economics perspective as well as an org structure perspective.
But the important part is that it's working. It's not the cleanest, but this has by far been the most excited we've ever felt as a business. The demand is like an ocean all around us and we have the parts required to go out there and capture it. These are the good kind of problems. You only get them when the core thing is working.
Where this goes next
First half was proving a leaner company could actually grow faster. Second half is time to accelerate. I mean, it's in our name after all: press W means to go forward.
I think we're early in something a lot bigger than us. The whole playbook for building a services business is quietly getting rewritten right now and most of the industry hasn't looked up yet. We're going to keep operating like we've looked up. We're diving into an aggressive hiring sprint to keep up with demand as we continue to document and automate 80–90% of workflows by the end of Q3.
To the clients who took the bet this half, and the team that keeps getting out ahead of me: thank you. I mean that.
And if you're running a company and trying to work out what this shift means for you (the actual version, not the LinkedIn version), come find me. tarun@pressw.ai.
More soon.