Danny Tomsett

Husband. Dad. Founder & CEO of @UneeQAI. Investor. Building AI-driven digital humans and immersive roleplay for the future of training.

[Danny Tomsett @TomsettDanny]

Sama started as an ambitious idea: create an AI-powered digital cabin crew member that could bring @qatarairways' hospitality into the digital experience. Today, she’s trained across 1,000 Qatar Airways topics and data points, helping people discover destinations, answer travel questions, and even book flights through natural conversation.

Projects like this have also helped shape @UneeQAI. Check out the case study here: https://t.co/yTvP5LaqsW

That’s a big part of why we created XD Studios. It gives us the ability to work alongside brands on those bigger ideas (outside of training), from figuring out what’s possible through to creating and deploying a digital human that genuinely represents their brand. And it’s been a huge success.

Qatar Airways, you guys are awesome!

[Danny Tomsett @TomsettDanny]

For a remote-first company, I'm convinced this model only works well with regular in-person meetups, planning offsites, and company-wide gatherings. This is non-negotiable for how we operate as a leadership team at @UneeQAI and it pays off big time.

Something I reflected on from a recent leadership team offsite was the unspoken transition that took place after months of no in-person contact; it was a little prickly, tense, and, if I had to use one word, disconnected. Then we got together, and what I watched unfold was something like this:

- Relational connection restored, which led to;
- Safety in the room, which led to;
- Permission to be more vulnerable with each other, which led to;
- A deeper level of understanding, which led to;
- Better decision-making and navigating trade-off tensions

In that environment, you have honest debates where you challenge each other and say what you think. Ultimately, the strategies and decisions are more focused and aligned because of it.

I’m incredibly grateful for the quality and passion of our leadership team at @UneeQAI. Humble, hungry, and smart. I'm lucky to work alongside them.

[Danny Tomsett @TomsettDanny]

I’m grateful for @JensenHuang's position and voice on AI safety, especially given that his incentive is to fuel a competitive ecosystem of AI companies.

His analogy of water is excellent.

Just as we have running water for showers and flushing toilets, that doesn’t discount the need for high-quality bottled water. But a world where we have to use bottled water for everything isn’t better.

I believe the same is true for AI.

We’re seeing this play out at @UneeQAI. We use the latest and greatest frontier models for high-value, complex tasks, but we also use several open models to better support latency, security, and cost requirements.

Different needs require different approaches. Charlie Munger said it best:

“Show me the incentive, and I’ll show you the outcome.”

There are many effective ways to approach AI safety, but we should also pay attention to the incentives behind them. Some approaches can easily lead to shared monopolistic effects.

Safety is incredibly important. So is making sure the way we pursue it doesn’t limit the competitive ecosystem that will ultimately make AI better.

[Danny Tomsett @TomsettDanny]

I have to admit, I felt for Tilly Norwood here 😂 (in case you haven't seen it yet: https://t.co/fFFS9PEzFi). Real-time AI interaction is hard. @ElineHQ and the team at @particle6p are getting a very public introduction to some of the problems we’ve spent years obsessing over at @UneeQAI.

And the language switch might be the funniest part of the clip, but it’s far from the only challenge:

• Conversation is more than lip-sync; they lacked any contextual body language, personality, or even emotion. Just creepy.
• Latency/delays in speaking are uncanny, unnatural, and annoying.
• Speech recognition needs to detect accents and optimize; otherwise, people are constantly being misunderstood.
• Conversational design requires so much more than just prompt engineering. It requires in-depth guardrails, personality, memory, goals, and context considerations and instructions.
• Speaking a different language is a great AI feature, but only useful if those you are interacting with actually understand that language 😂

Really, though, I do love seeing more teams experiment with real-time AI characters because moments like this show how much engineering and design goes into making an AI interaction actually feel human.

Eline, welcome to the fun part. 😊

If anyone wants to compare notes with Tilly, come have a conversation with one of ours: https://t.co/Cru9D1RYsJ

[Danny Tomsett @TomsettDanny]

I've been reflecting on the future of work, and it's clear to me that AI is elevating the need for valuable human skills. The harder question is how students actually get them. That first job out of university is where you used to learn to handle a difficult conversation, maybe badly at first, in front of a real person, with someone to debrief you afterward.

Stanford's Digital Economy Lab found that employment for 22–25-year-olds in highly AI-exposed occupations now sits roughly 19% below where it would be had it kept pace with their less-exposed peers. Fewer of those first jobs. And NACE's 2026 data shows employers rate communication as near-universally important, while only about half say new graduates are proficient.

So: how do you give a student 200 reps of a hard conversation when the workplace no longer offers the first 50?

@maysbusiness has an answer worth reading. Their Sales Lab leverages our Immersive training platform (powered by @Dell Hardware), which lets students practice interviews and customer conversations on demand inside the LMS, with feedback from their instructor and AI coaches.

If practice is how people get better at something, technology can finally make high-quality practice available more often than ever.

That’s an exciting place for learning to go next.

Check out the case study: https://t.co/LA0QDVw8yz

[Danny Tomsett @TomsettDanny]

Eight years ago, in 2018, we closed @UneeQAI's seed round. That was the last milestone round we've done.

I remember how good it felt; everything was going our way - next stop: accelerated growth! What followed was a long grind for product-market fit, uncertain markets through COVID, and a genuine test of my leadership as I navigated survival strategies and then brought back energy to thrive again.

I'd scaled a telecommunications company before, but boy, did I underestimate what it takes to find product-market fit for an emerging technology pioneering a category that didn't exist yet. Those are different muscles. Many of the players that started when we did no longer exist.

It's a hard feeling to describe: navigating hard times for more years than you expected, then, within 12 months, a few things 'click,' and the hope you held for customers who love you, high growth multiples, and a culture that believes is now a reality.

It's finally time for Series A. Weirdly, eight years in, and it feels like the beginning. I'm genuinely excited for this season.

Some honest reflections on fundraising:

What I enjoy: it forces you to sharpen how you articulate what you do well, why it's defensible, and where you're going. It's real validation of the value you're creating - when you find aligned investors, that's uplifting. You build real relationships, and I've found that even investors who pass are often genuinely willing to help and make introductions. Doors open well beyond the investment itself: partnerships, customers. And it makes you know your numbers and get your house in order, so anyone looking in has confidence this is a well-run ship.

What I find hard: the sheer amount of time you can't give to operating the business and the growth you're raising for. The rejection - being told repeatedly why this won't work and navigating those emotions. And the conflicting advice: you want to be humble and take it on board, but when it pulls in ten directions, it's hard to judge which counsel is right while you're still in uncertain waters.

[Danny Tomsett @TomsettDanny]

Can OKRs hurt early-stage companies?

Two years ago, our chairman @fgthiel challenged our use of OKRs. Not the measurement or quality of data - he challenged the distraction cost. Objectives, key results, cascading measurables. It all sounds sensible. But in a startup, it generates conversation about things that aren't the most important thing. And there is usually only one (maybe up to three at a stretch! 😊) thing that actually matters.

I admit, I was slow to get what he was talking about, but got there eventually. This year, we tried a new, much simpler system. Three goals max, and we just bring them into every conversation. That's it. Every department reports weekly in Slack with a red, yellow, or green status against those goals. Nothing else. No KPI reporting.

The effect: the goals now show up in every conversation, because there's so little else to talk about. Who's struggling, who's winning, what someone needs - it's all at the surface instead of buried inside a key result nobody reads.

We still need data. So we pulled it out entirely: operational metrics get their own monthly session, where we evaluate the numbers for data-led decisions, but we don't let it distract from week-to-week conversations.

Fewer things to talk about. More of them get talked about.

[Danny Tomsett @TomsettDanny]

We spend a huge part of our lives working. Why did we decide it had to be so serious?

During a meeting I had with @richardbranson 15 years ago, it showed how much he valued having fun in business. I’ve thought about that a lot lately, especially as we build technology designed to help people learn and practice.

Fun isn’t a distraction from learning. Sometimes it’s the reason people want to keep going.

Something I want us to keep getting better at, both in what we build at @UneeQAI and how we work together.

[Danny Tomsett @TomsettDanny]

I feel like this year we're really dialing in our feedback strategy. Thanks to AI, the speed at which we can now develop features requires more decisions around feature prioritization each week (versus each quarter). This is what our current feedback strategy looks like for the Immersive training product & CS team:

• Don't be shy: We seek feedback right after a key action or purchase, rather than days later.

• Product feedback: short, embedded prompts (like a 1-to-5 star scale or a thumbs-up icon).

• Behavioral Data: We pair what customers say with session recordings and heatmaps to see how they use our product.

• Patterns: Customer Success and sales-recorded conversations that are then analyzed by AI to identify trends and patterns.

I love innovating and pioneering on conviction, but I must admit that I'm getting massively energized by the speed of optimization and value we gain from our customers and partners. Grateful!

[Danny Tomsett @TomsettDanny]

After hundreds of conversations with many corporations, the consistent theme I've picked up on is this:

In-person training and coaching is engaging, fun, and scores high for effectiveness and value.

Digital training, however, is often considered boring, unnecessary, low engagement, and not fun in any way.

My takeaway: Good coaches understand the correlation between fun and increased learning and engagement. If Digital Platforms are to become more effective, instead of investing in ways to ensure poor souls can't fast-forward the video, they need to find a way to make it more interesting, relevant, and... well, yes, FUN!

We recently ran a survey with our trial and pilot customers; over 97% of participants said they preferred the digital human interaction experience because it didn't just feel real and like what they would face in the real world, but they found it fun and engaging and said they would use it again.

We’re also seeing people spend half an hour, sometimes more than an hour, practicing with digital humans because they want to get better at their craft. I love that.

How many people do you know who voluntarily spend an hour watching corporate training videos and answering multiple-choice questions in their own time?

We know play works, we know interaction works, and we know people learn when they actually want to participate. We need to scale more of that.

And remember, not all AI roleplays are the same. Try the @UneeQAI free trial and see the difference for yourself.

https://t.co/Cru9D1RqDb

[UneeQ @UneeQAI]

How do we make AI roleplay so realistic? 👀

We use real performances to simulate the way people actually behave in real life. 💫

Our Creative Team works with professional performers to capture full-body animations through mo-cap sessions, like the one you see in this video.

We then use the data to constantly improve Synanim™ – our AI animation and behavior engine.

All that comes through in your roleplay session, from the big obvious gestures to the subtle facial expressions.

😠 A skeptical CFO you're trying to sell to crossing their arms.
🤔 A customer visibly getting more and more confused by technical jargon.
💆 An employee feeling calmer as you successfully talk them through a performance review.

An LLM isn't built to drive this behavior.

But Synanim™ is!

Some things are worth going the extra mile. For us, that's turning the dial up on roleplay realism. 💜

#roleplay #training #learning #learninganddevelopment #immersive

[Danny Tomsett @TomsettDanny]

I play the drums.

If I sat down at a kit in front of you, you'd accept I know enough to be in the band. And if you'd never heard a great drummer, you might even think I'm good. Then you hear a great drummer. And you know immediately - I'm average at best

That's where AI roleplay is right now.

Anyone can build a roleplay bot. That's why 30+ companies claim to have solved it with their "roleplay feature." Compare what they've shipped to what was possible only a couple of years ago, before reliable AI, and I understand why some are having success with a sprinkle of prompt engineering and a few AI tools stitched together. It plays. But honestly, it's average. And most buyers haven't heard great yet, so average will do.

Today we're changing that!

Sometimes you can't explain something - just let people experience for themselves the difference. So today we launched a new "try simulation" straight off our homepage - no sales call, no form maze. And we've opened a 30-day free trial so you can see what the Immersive Training Platform does when roleplay and coaching are built to really make a difference in preparation for the real world.

Don't take my word for it. Go experience the difference.

https://t.co/Cru9D1RqDb

[Danny Tomsett @TomsettDanny]

Following my theme of doing new things this year, I can now tick off Sockeye Salmon fishing in Alaska! Wow, what a place; it was gorgeous and raw. Loved seeing Grizzlies in the wild, catching fish, and just getting fat. What a life 😊

Some lessons I took away from this trip:

• Time away with the lads is good for your soul. Good conversations, plenty of laughs, and investing in lifelong friendships. This is a must.

• To catch the salmon, you must think like the salmon 😉 ...the real lesson is the small, minute details that separate a great fisherman from a wannabe like me! So much of life is like this - including my world of AI - getting to 90% is easy, making something really great and being truly valuable and effective takes an obsession and passion to grind out that 9.9999% (always seeking perfection).

• There is so much beauty in the world - I want to see more! Nature is so impressive and worth preserving.

Back into normal life again, already dreaming of the next new challenge or experience.

[Danny Tomsett @TomsettDanny]

When I first met Ashley, she was working for an awesome company, @veritoneinc - I thought to myself that this is exactly the type of person I would love the opportunity to work with, lead, and empower to “unlock” their full potential.

I never imagined a few years later our paths would collide, and I’m so grateful that Ashley joined @UneeQAI to lead our marketing team. She is humble, hungry, and smart - and is just getting started! Look out, world!

Her latest piece in @HRFuturemag gets into something we both care a lot about: how AI can help people get better at their jobs.

Ashley talks about what this can look like across recruitment, onboarding, and ongoing development, and where digital humans fit into that future.

A great read.

[UneeQ @UneeQAI]

And speaking of sports teams*...

Could you train any athlete using only YouTube videos? ⛹️

Of course not. But it's a lesson many leaders learn the hard way, as @TomsettDanny shared from first-hand experience.

Luckily, better simulation training is available today — and not just for athletes.

#learning #training #leadership #development

🏀 *See yesterday's post where our CTO, Tyler, tries to get Dallas Mavericks season tickets in return for training all of Mark Cuban's staff. Well, it's worth a try. 😄

[Danny Tomsett @TomsettDanny]

One of the fun parts of the product-market fit journey is observing customer behavior and usage on this "thing" you built.

When it comes to practice, we worked really hard to simulate real-world conversations. Most in-person meetings or video calls are 30+ minutes, so if we are building an effective product, we should be able to observe interactions that are just as long.

My team sent me a report last week, showing that we have thousands of users now frequently going 20+ minutes in roleplay, with some even as long as 50 minutes. Very excited to see this, as we work really hard on capturing the nuances of a meeting, the way people talk to each other, and enabling the most realistic practice possible.

It just makes sense: if people stay in the conversation long enough for it to get messy, useful, and real, that’s where the learning happens.

[Danny Tomsett @TomsettDanny]

I recently finished Going Infinite, Michael Lewis's book about Sam Bankman-Fried (SBF). Great read! It is another example of a company or individual that, for a time, looks brilliant, only to crumble because of a fractured character... and arguably time! Integrity and relationships are critical to sustainable businesses - something he overlooked, and it cost him.

SBF treated everything as a probability calculation, including honesty. Truth wasn't a principle. It was a variable. If a lie had positive expected value, run the trade. Infinity only lasted 3 years in his story.

The most insane fact is that if SBF had not been forced to liquidate his early-stage tech and crypto investments, the portfolio would arguably be worth between $100 billion and $114 billion today. This value is nearly 10x the initial $8 billion hole in FTX's balance sheet.

Time is what he needed. Integrity and relationships would have made a difference...good lesson.

[Danny Tomsett @TomsettDanny]

Some companies are discovering AI. At @UneeQAI, we've been deep in it since 2016.

This may be more technical than my usual posts, but I'm so proud of what my engineering team is doing, I'm going to brag on them a little.

Digital humans that hold real-time conversations don't leave room for hand-waving. Every millisecond of latency is visible on a face. That forces a different kind of engineering culture, one where we take nothing at face value and benchmark everything ourselves.

A few things we've been up to lately:

• We're pushing local inference to its limits. On @NVIDIA's new DGX Spark hardware, we've tuned multi-token prediction to hit 75-80 tokens/sec on a 26B model and swept the speculative decoding budget to find the true throughput optimum, rather than trusting the marketing numbers.
•We also caught two bugs in a major open-source inference engine and a silent tool-calling failover trap and shared the fixes with the community.
•We're building AI-native workflows. Our engineers run agentic pipelines where a frontier model architects the plan and local models execute the build, full codebases with database layers, payments, and test coverage. We rigorously measure where autonomous execution holds up and where it still needs a human in the loop.

And we're going deep on the fundamentals: KV cache optimization, micro-scaling quantization, and the memory engineering that lets models inhabit long conversations without losing the thread.

Ten years in, the bar keeps rising. So do we.

[Danny Tomsett @TomsettDanny]

Our Digital Humans are now getting deployed by the biggest players in the AI game - this is a cool story with the City of Amarillo saving $1.8 million since deploying Emma!

https://t.co/ObIlPOZJ88

We may not be growing at the speed of some of these token-maxing outliers, but we are solving valuable problems that generate real ROI.

There is satisfaction in knowing you're forming part of the AI story that actually is making a difference.

@Dell, @nvidia