What I Learned Building a Seed-stage Startup Operations Stack

I’ve been working at Trent AI as Head of Operations for over a year now, and a recurring theme in community chats, blogs, newsletters, and conversations with other startup operators is:

Can you recommend a tool for XYZ? What do you use for XYZ?

Personally, I’ve asked these questions many times; sometimes a recommendation did it. Other times, I went through multiple demos, sales calls, free trials, and occasionally followed by another demo.

Some tools are good, some are great, and some are terrible. Some tools are amazing but sooooo expensive, while others are pretty cheap and pretty good.

Of course, an easy way out, you’ll think, is to go with the most popular and trendy tools in every category. We could buy the HRIS with the red logo everyone recommends, the travel platform with every possible feature that practically takes the trip for you, or the finance solution with reporting features our future CFO may appreciate one day, but the cost would make no sense for a company at our stage. For context, we’re a seed-stage startup ($13M raised) with a fully remote team of fewer than 20 people, working across two continents and multiple time zones.

So, I thought I’d share some of the tools on our day-to-day ops stack at Trent AI: what we use them for, why we chose them, and where the compromises are.

How we choose tools

I’d like to keep the suspense going; you can always scroll down to the juicy bit, but it’s important to give a little context on how we make these decisions, ’cause (and believe me on this one) winging it will give you more problems than solutions.

So, what do I ponder when choosing a tool?

• Does it solve the problem we have NOW and can people start using it right away?
• Does it integrate with our current stack?
• Is it secure, and does it support our compliance requirements?
• Does it have a responsive customer support team?
• And, honestly, is the price reasonable?

Rarely does a tool tick every box. More often, there’s a very good tool with a ridiculously high price tag (like the one that quoted $50K for a single three-minute tailored training video, no kidding), a cheaper option that gets 80% of the job done, and a spreadsheet quietly whispering, “You know, I could do this for free.”

A quick note before we get into it: this article isn’t sponsored. These are tools we actually use, and the opinions are entirely my own.

The tools we actually use

Where we communicate: Slack

I know; I know. After all that intro, I’m kicking off the list with Slack. But hey, can you blame me? Slack was an easy choice. It’s familiar to engineers and pretty much everyone working in tech, so there was no need to reinvent the wheel. It’s fairly priced and integrates with basically everything, so updates from the tools we use show up where people will actually see them. We can also build reminders and simple workflows into a place the team already uses every day. And, of course, it meets our compliance needs.

What don’t I like? Huddles. And yes, you inevitably end up with too many channels and have to remember to archive them from time to time, but all the good stuff compensates for that.

Where work and knowledge live: Notion and Linear

Notion is where pretty much everything lives. Planning, company policies, meeting notes, the engineering wiki, even CRM. I’ve used other similar tools in the past, including Coda, but Notion feels cleaner. It gives you fewer options to play with colors and customize every little thing, which I’ve come to see as a good thing. Those things don’t add much, and they can easily become a distraction.

I’m less convinced by its tables. Sometimes you just want an upgraded spreadsheet, but Notion’s tables aren’t great. So you create a database instead, but then every entry opens as another page, and before you know it, things get messy. Thankfully, Notion AI has made that much easier to navigate. It works better than the regular search, especially when you know the information exists somewhere but have no idea where (oops!).

I also like Notion’s agents. They’re very easy to vibe code and can handle simple tasks, such as duplicating a meeting agenda on schedule or sending a ping on Slack for whatever purpose you choose.

We use Linear for issue tracking and proper project management. It’s simply clearer than Notion when you need to see what’s being worked on, who owns it, and where it stands.

Notion is affordable for what it does. The AI credits could change that calculation, particularly as we add more agents, but for now the credits included in our plan are enough.

Where we manage people: CharlieHR

We use CharlieHR for employee records, time off, onboarding, and our org chart. I chose it because it’s simple. There are trendier HR tools that also handle payroll, benefits, expenses, and even device management. That sounds great, but for us it’s unnecessary complexity at a much, much higher price tag. CharlieHR is very affordable, has a good UI, and it’s easy to learn and navigate. It also offers add-ons we may find useful later, so I believe it can grow with us for the next few years.

It has its limits, like the fact that there’s no option to download reports at all (really? yeah). The onboarding process could also be more customizable… some things are set by default and cannot be changed, which I’ve learned to live with, but I still yell every time I remember I can’t change them.

We also need to be careful not to outgrow it without noticing. Migrating 30 or 40 people manually would be manageable (I hope!). Migrating more than 100 when you can’t export the reports you need? Not a great idea.

How we handle finance and accounting: Novabook

We use Novabook for bookkeeping, taxes, payroll, and managing our pension scheme. They also manage Xero for us. Their solution was built for startups, and they are a startup themselves, which I really like.

The app lets us upload receipts, manage payroll, see government filings, and reply to queries directly (I added a Zapier automation that works wonders for receipt management). At the same time, we have a real accountant assigned to us who can clarify whatever questions come up.

For me, that mix of software and access to a real person works well. We can handle most things independently without having to navigate the more complicated parts of corporate finance alone. They’ve also been responsive to feedback and have improved the product during the 15 months we’ve used it.

Eventually, we’ll need someone in-house, but Novabook covers what we need right now, and the pricing is very reasonable. I can’t say there’s anything I truly dislike. Some services aren’t included by default and are priced separately, but well, they’ve gotta make a living too.

How we handle security and compliance: Vanta, Okta, 1Password, and Jamf Now

Being a cybersecurity company means being extra careful about the tools we use. Although we can dogfood our own product, there are other aspects of security and compliance we need to cover. I won’t go through our entire security stack, but these are the four tools I use most as Head of Operations.

Okta handles single sign-on and gives us one place to manage access across our tools. It also makes onboarding and off-boarding staff much safer. It’s a market leader for a reason. We use 1Password to manage passwords, API keys, and other secrets safely. Jamf Now manages our company-issued Macs. Jamf Now, not Jamf Pro; that one would be overkill; trust me, I checked.

Then there’s Vanta, by far the most expensive of the group. I chose it because it gave us the best option to self-manage our SOC 2 compliance journey. Its automations and integrations got us about 60% of the way there. The remaining 40%? Painful manual work. Lots of it. Vanta doesn’t make compliance automatic, but it gave us enough structure to manage the process ourselves. It’s expensive, but I would choose it again.

My AI favorites: Granola, Claude Cowork, and Lovable

Everything is AI now, and to be fair, so are we. If I included every tool with an AI feature, I’d probably end up listing half our stack again. So I’m going with my personal favorites.

Granola: Its magic is that it’s non-invasive. It leaves a message in the call chat saying “I’m using Granola”, but it doesn’t join as another caller. Their transcript search AI is great. Now I can go back and ask, “What exactly did the CEO say about this?” and bam, there it is. Love it. I tried Fireflies and Fathom, and they were good, but the extra bot on the call was annoying, and I didn’t find the summaries as accurate.

Claude Cowork: This is where my digital coworkers live, and where the build-vs-buy dilemma gets interesting. For example, one of my agents works as a pipeline manager. It collects updates from a specific Slack channel and uses them to keep our deals database in Notion up to date. The team only needs to leave a message in Slack.

We could buy a dedicated tool to do this, and it would probably do much more. But another tool would also give the team another platform to update and give me another platform to manage. For now, I’d rather build the simple version we actually need. Fair, it’s a bit more complicated than I’m making it sound, but this isn’t a dissertation on how I built our CRM agent.

The downside is usage limits. These operational agents can be costly, so I have to juggle their schedule to make sure they run at different times and don’t get stuck.

Lovable: We use it for one-pagers, decks, and the occasional game at team meetings. I’ve tried AI tools built specifically for deck design, but they never quite met my expectations. Somehow, I get closer to what I want with Lovable, even though that’s not what it was primarily built for. Go figure.

What don’t I like? Credits disappear fast. The Pro plan also doesn’t include internal publishing only, which is a bit inconvenient when what you are building is, well, internal.

The tools we deliberately don’t have

After mentioning all these tools, I should also talk about the ones we chose not to buy.

First, we decided to skip a travel management platform. People travel for conferences, meetings, and our quarterly company off-sites, which are super fun, if I may brag. Everyone books their own travel, and we reimburse them. We have policies around reasonable spending, and we trust people to stay within those boundaries. They do. Our travel volume isn’t zero, but it’s not significant enough to create a real problem either.

I still sat through a few demos with well-known travel platforms, just to check whether I was missing out. The cost tracking is great, and I can see how the approval process becomes useful once managers are responsible for larger teams and budgets. But we’re not there yet. When our travel volume grows to the point where reimbursements and budget visibility become painful, I’ll happily reconsider. Until then, not adding another tool is also a stack decision.

Then there’s the dedicated applicant tracking system we don’t have. We use the free version of Airtable. Hiring managers use it to move candidates through screening, technical interviews, team fit, and eventually the offer stage. We hire mostly through our network, so we don’t need automated application processing, interview scheduling, or detailed recruitment analytics yet.

I’ve used dedicated applicant tracking systems before, and they make sense when you’re hiring at volume. At our stage, adding one would mean paying for features we don’t need and giving hiring managers another platform to learn. The day we publish a role and receive hundreds of CVs, Airtable may quietly hand in its resignation. Until then, free works.

Takeaways

• The most popular tool isn’t automatically the right one for your stage.
• “Starting at $X” rarely tells you what you’ll actually pay. You’ve been warned.
• Security and compliance need to be considered early, no matter what market you are in, but even more so for us since we’re safeguarding critical systems.
• Check the billing process before signing, or prepare for recurring headaches.
• If you can’t understand a tool during the free trial, more features won’t save it. Run.
• You don’t need to build the stack for the company you might become in three years. You need the lightest stack that solves today’s problems.

One thing is for sure: our stack will change. Some of these tools will grow with us, some will be replaced, and at least one will probably raise its price just after this is published.

If you have questions about these or other tools you’re considering, send me a message. I may be able to save you a few demos 🙂