Build vs buy: should you build your own event management system with AI?

Building your own event system with AI versus running proposals, budgets and invoicing on one maintained Qondor platform.
|

The short answer:


You can build a working prototype of an event management system with AI in a few weeks. Running your business on it is another matter.


A system that handles proposals, budgets, supplier costs, VAT and invoicing needs a permanent team to secure it, maintain it and keep it compliant. For most event agencies and TMCs, that costs more than a purpose-built platform and pulls focus from the work that makes money.


Build when the tool is small, internal and low risk. Buy when it touches money, client data and tax.

Why are agencies and TMCs looking into building their own system now?


Because AI coding tools have made software look cheap. Tools like Lovable, Replit, Cursor and Claude Code can turn a prompt into a working app in an afternoon. If you have watched a colleague do it, the logic is tempting: why pay for a platform when you could build exactly what you want?


You're not alone in asking. In Retool's 2026 Build vs Buy Report, 35% of respondents said they had already replaced at least one SaaS tool with a custom build, and 78% expected to build more in 2026.


Look closer at what they replaced, though. It was mostly workflow automations and internal admin panels. Small tools, used by a few people, where a bug costs an afternoon.


An M&E business system is a different animal. It holds your margins, your clients' personal data and your VAT position. When it breaks, an invoice goes out wrong or a delegate list leaks. Even a16z's Anish Acharya, whose firm invests heavily in AI, called the idea of vibe coding core business systems "flat wrong".


What would your system actually need to do?


The prototype usually covers a project list and a budget sheet. The business behind the event needs the whole chain from first enquiry to final invoice, connected:


  1. Proposals that look professional, reuse content and go out fast.


  2. Budgets with planned versus actual costs, updated live as suppliers change.


  3. Client rate cards, management fees and commission rules, enforced on every project.


  4. Supplier contracts, deposits, payments and reconciliation.


  5. Multi-currency pricing and VAT. In the UK, that means the Tour Operators' Margin Scheme (TOMS). In Sweden, the margin scheme works differently again.


  6. Time tracking, so billable hours don't leak.


  7. Registration, participant lists and attendee communications, all under GDPR.


  8. Invoicing, and clean integration with your accounting system (Xero, Visma, Tripletex, Fortnox and so on).


  9. Roles, permissions, an audit trail on every budget change, and reporting across every programme.



Each item looks simple on its own. The hard part is that they all depend on each other. Change a supplier cost and the budget, margin, VAT calculation and invoice all have to move with it, correctly, every time.


What are the real benefits of building your own system with AI?


There are some, and they are worth taking seriously


  • Speed to a first version. A prototype that took months can now take days. That's a great way to work out what you need.


  • A perfect fit, at first. You decide every screen and every field.


  • No licence fee. The subscription line disappears from the budget, even if other costs replace it.


  • Good for small, internal tools. A dashboard, an intake form or a one-off automation is a sensible thing to build yourself.


  • The tools keep improving. Research group METR found experienced developers were 19% slower with AI tools in early 2025. In 2026, it said the speed-up is now likely real, though hard to measure.



So building is realistic. The question is what you're building and why, and who looks after it in year three.


How much does it cost to build your own event management system?


For a core M&E system that a whole team can rely on, plan for roughly £280,000 to £320,000 in the first year and £165,000 to £225,000 every year after that. Those are indicative figures for a UK agency. Your numbers will vary, but the shape rarely does.


AI writes code faster. It doesn't decide what the system should do, check that the VAT is right, or answer the phone when invoicing breaks on a Friday. You still need people.


Cost line (Indicitative, UK)

Year 1: build


Year 2 onwards: run


Developers


2 senior developers, about £180,000


1.5 to 2 developers, about £135,000 to £180,000


Product owner from your own team


Half their time, about £35,000


A day a week, about £15,000


Design and testing


About £25,000


Included above


Recruitment fees (15 to 20% of salary)


About £25,000 to £30,000


Again whenever someone leaves


Hosting, tools and AI usage


£10,000 to £20,000


£10,000 to £20,000


Penetration test


£5,000 to £10,000


£5,000 to £10,000 a year


Indicative total


About £280,000 to £320,000


About £165,000 to £225,000


How we got there: a senior software engineer in the UK has a median salary of about £75,000 (£95,000 in London). Add 15% employer National Insurance and pension, and each one costs you close to £90,000 a year. In Norway, an average developer earns about NOK 900,000, which comes to roughly NOK 1 million a head with employer's contribution and pension. Penetration test prices are from JUMPSEC's UK guide and recruitment fees from Coburg Banks.


Over three years, that is roughly £610,000 to £770,000 before anyone has asked for a new feature.


Timeline showing a prototype built with AI in days to weeks, a system safe for live projects in 6 to 12 months, and running costs of about £165,000 to £225,000 every year after.


How long does it take to build your own event management system?


Expect six to twelve months before the core chain is safe to run live projects on, and longer with accounting integrations and margin-scheme VAT. Hiring alone takes four to eight weeks per developer.


Big software projects also tend to overrun. A McKinsey and Oxford study of 5,400 IT projects found large projects ran 45% over budget and delivered 56% less value than planned. AI has not changed the part where requirements shift halfway through.


What does it cost to maintain afterwards?


More than the build. Software engineering research has long put maintenance at 40 to 80% of a system's lifetime cost, with 60% as a typical figure (Robert Glass, Facts and Fallacies of Software Engineering). Launch day is where the bill starts.


Here is what keeps the team busy after go-live.


Keeping up with the rules.


VAT is the clearest example. Under HMRC's VAT Notice 709/5, TOMS is mandatory for businesses that arrange conferences with delegate accommodation. You need to split venue hire from bundled packages, avoid issuing VAT invoices on margin-scheme supplies, and run a year-end calculation straight after year end.


Sweden's margin scheme (VMB) works almost the other way round for business clients, where VAT must be shown on the invoice. Run programmes in both countries and your system has to handle both. When the rules change, your developers have to notice and update the code.


Get it wrong and HMRC can charge penalties of up to 30% of the tax at stake for careless errors, and up to 100% in the worst cases.


Security and GDPR.


Your system will hold delegate names, contact details, passport data and dietary needs. Under GDPR, a serious failure can mean fines of up to €20 million or 4% of global turnover. The average data breach cost $4.44 million in 2025, according to IBM.


Many corporate clients and TMC buyers now ask suppliers for ISO 27001 or similar evidence before they share data. Certification typically takes six to ten months and, for a smaller company, around £12,000 to £40,000 in year one plus annual audits (Consilien, indicative). That is in addition to the developers.


Everything else that moves.


  • Accounting systems update their APIs, and your integration breaks.


  • Browsers, libraries and hosting platforms need patching.


  • Your team asks for new features, because the business keeps changing.


  • The developer who built it leaves, and takes the knowledge with them.



That last one matters most for a lean agency. One person leaving should never put your invoicing at risk. Yet with a home-built system, it often does.


Stripe's Developer Coefficient study found developers already spend about a third of their week dealing with technical debt. AI-generated code tends to add to that pile faster.

What are the risks of AI-built agency software?


The biggest risk is code that looks finished but is not safe. AI tools are very good at producing something that works in a demo. They are less reliable at the parts you cannot see: security, edge cases and data integrity.


  • Security gaps are common. Veracode tested over 100 AI models and found that 45% of coding tasks introduced a known security flaw.


  • More code, more problems. In large enterprises using AI assistants, Apiiro saw ten times more security findings and a 322% rise in privilege escalation paths.


  • Quality drifts. GitClear's analysis of 211 million lines of code found copy-pasted code rose sharply as AI use grew, while tidying up existing code fell. That is tomorrow's maintenance bill.


  • "Almost right" is the norm. In Stack Overflow's 2025 survey, 66% of developers named "almost right, but not quite" as their top frustration with AI output. In a budget or a VAT calculation, almost right is wrong.


  • Things go wrong fast. In July 2025, an AI agent on Replit deleted a live production database during a code freeze. The same year, a flaw in apps built on Lovable exposed data from more than 170 apps, and Wiz found an authentication bypass in Base44 that let anyone into private apps.


  • Internal AI builds succeed less often. MIT's 2025 study of generative AI projects found tools bought from specialist vendors worked out about 67% of the time, and internal builds about a third as often.



There is also a quieter cost. Every hour your best project manager spends writing prompts and testing screens is an hour they are not winning or delivering events.

Build vs buy: how do the options compare?


Building wins on control of the roadmap. Buying a purpose-built platform wins on almost everything else that matters to an M&E agency.


What you care about


Build your own with AI


Buy a purpose-built M&E platform


Time to first live project


6 to 12 months, plus hiring


Weeks, with onboarding support


Year 1 cost (indicative)


About £280,000 to £320,000


A subscription, plus onboarding


Ongoing team


1.5 to 2 developers, permanently


None. The vendor's team maintains it


VAT and margin schemes


You research, build and update every rule


Built in and kept current


Security and GDPR


Your responsibility, audits included


Handled by the vendor, with evidence you can share with clients


Accounting integrations


Built and repaired by you


Maintained by the vendor


Key-person risk


High. The builder often holds the knowledge


Low. Support and documentation sit with the vendor


New features


Only what your team has time to build


Shaped by many agencies and TMCs, shipped continuously


Fit to your workflow


Exact, at first


Built around how M&E businesses run, configurable for yours


Focus


Part of your team becomes a software company


Your team stays focused on clients and events


When does buying Qondor make more sense than building?


When the system has to run the business behind the event. Qondor is the M&E operating platform for event agencies and TMCs. Proposals, budgets, supplier costs, VAT, time and invoicing sit in one place, built by people who have worked in the industry.


That means the hard parts in this article are already done, tested and maintained:


  • Financial control from proposal to invoice. See the margin on every project while it is live, not when it is too late to act. Rate cards, management fees and supplier reconciliation are enforced in the platform, so you bill what was agreed.


  • Compliance built in. TOMS and Nordic VAT handling is part of the product, and we keep it current when the rules change.


  • One way of working. Every producer, account manager and finance lead works from the same project. If someone is off sick, anyone can pick it up.


  • Integrations we maintain. Connect to your accounting system, such as Xero, without owning the integration yourself.


  • AI where it pays off. Our supplier reconciliation agent takes on the matching work your team does by hand today, inside the platform you already use.



Our customers see the difference in time back. KRS LIVE saved more than 675 hours after moving onto Qondor. BCD M&E describes Qondor as "the backbone for our M&E business".


We will also be clear about where we stop. Qondor doesn't replace your booking tools, GDS or accounting system. It runs the M&E programme layer and connects to the rest.


A sensible middle ground.


You dom't have to choose between AI and a platform. Use AI to build the small tools at the edges of your business: a client dashboard, an internal checklist, a one-off report. Run the core, where your money, data and tax live, on something built and maintained for it.


Thinking about building? Send us your build plan. We will walk through it together, put real numbers on it, and show you what Qondor already covers.

Send us your build plan. We will walk through it together, put real numbers on it, and show you what Qondor already covers.

Thinking about building?

Frequently asked questions