I have used every version of Classic Outlook since the mid-1990s. Over the years I have tried the browser version many times, and each time I have gone back to the original. So I understand why many people are not looking forward to this change.

But change is coming, and it pays to be ready.

From March 2027, your business will no longer be able to opt out of New Outlook. Microsoft has pushed the deadline back from April 2026, but that leaves around five months to get ready.

Classic Outlook will be supported until 2029, yet Microsoft 365 Enterprise users are already being switched to New Outlook automatically. Users can switch back for now. After March 2027, that choice disappears (Microsoft Message Center MC949965).

For AEC firms, email is where project decisions, instructions and approvals live. A rushed switch is not just an IT inconvenience; it is a project risk.

Our advice: prepare now, but stay on Classic Outlook for as long as you can. New Outlook does not yet offer all the features of Classic, so there is no benefit in switching early. Use the time to get ready, and move when you have to.

Why it matters

New Outlook is built on the web version of Outlook, so it is simpler but less capable. Two issues will affect most practices.

Features your team relies on are missing. These include Quick Parts and AutoText, rules based on categories, custom calendar views, ribbon customisation and full Public Folder browsing. Search and filtering are also less precise.

Many add-ins will stop working. New Outlook supports only modern "web add-ins". Older COM add-ins and VBA macros will not run, and Microsoft is not expected to change this. If you file project emails, manage correspondence or connect Outlook to other systems through an older add-in, that workflow is at risk.

There is a real upside. Traditional add-ins load onto every copy of Outlook across the business, slowing it down for everyone. Many of us also know the sinking feeling of wondering whether an email actually filed, and whether we will find it when we need it. Modern web add-ins run in the cloud, keep Outlook lighter and make filing and searching more reliable.

Five steps to prepare

1. Audit what you use. List every add-in, macro, integration and custom workflow connected to Outlook, and note which will not work in New Outlook.
2. Check your contracts. Review renewal dates for third-party add-ins. Some require several months' notice to cancel, so act early.
3. Pilot with a small group. Start with IT and a few willing users. Test business-critical workflows such as filing project correspondence.
4. Train and communicate. Short guides and clear dates prevent surprises and reduce support requests.
5. Roll out in stages. Use Microsoft 365 admin policies to move one team at a time, starting with low-risk departments.

Your timeline

How Setout can help

The switch to New Outlook is a good moment to ask a bigger question: is your email properly connected to the rest of your business?

We start with the problem, not the software. We will review how your team handles email and project correspondence, identify what will break, and recommend the right replacement. For example, Cloudfiler is a modern web add-in for filing and searching email that works with both Classic and New Outlook.

AI is the latest technology wave to reach architecture, engineering and construction. It will succeed or fail for exactly the same reasons Professional Services Automation (PSA) did.

I have spent more than 30 years helping AEC practices adopt new technology. I watched firms move from spreadsheets and timesheet books to integrated PSA platforms that connect projects, resources and finance. Some transformed their margins. Others spent heavily and ended up with expensive software nobody trusted.

Today the same conversations are happening about AI. The vocabulary has changed, but the patterns have not. Here are six lessons from PSA that every AEC leader should apply before investing in AI.

1. Start with the business problem, not the software

The most successful PSA projects I worked on never began with a product demo. They began with a frank question: where are we losing money, time or control?

For one practice, the answer was unbilled time leaking between timesheets and invoices. For another, it was project managers discovering overruns weeks after they happened. The software was chosen to fix those problems, and success was measured against them.

AI is no different. "We need an AI strategy" is not a business problem. "Our bid team spends three days assembling every tender response" is. Define the pain, put a number on it, then ask whether AI is the right answer. Sometimes it is not, and that is a valuable thing to learn early.

2. Your data foundations decide the outcome

Every PSA implementation lives or dies on data. Inconsistent project codes, duplicate client records and fee structures held in someone's head will undermine the best platform on the market. I have seen go-live dates slip by months because nobody had agreed what a "work stage" actually meant.

AI raises the stakes further. A PSA system with poor data produces wrong reports. An AI tool with poor data produces confident, plausible answers that are wrong, and that is far more dangerous.

The good news is that the work overlaps. Firms that have already cleaned and structured their project, resource and financial data through PSA are years ahead in AI readiness. If you have not done that work yet, it is the best AI investment you can make today.

3. People and process come before technology

The hardest part of PSA was never the software. It was persuading a senior engineer to fill in a timesheet daily rather than on Friday afternoon, or convincing project leads that resource planning was a shared responsibility.

PSA forced practices to standardise how they set up projects, record time and manage change. The firms that redesigned their processes first, then configured the system to match, got lasting value. Those that bolted software onto broken habits simply automated the mess.

AI adoption carries the same human challenge, often with added anxiety about jobs and professional judgement. Be clear about what AI will and will not do. Involve the people whose work it touches, train them properly, and set simple rules on how AI outputs are checked before they reach a client or a drawing.

4. Connected beats standalone

The real breakthrough with PSA came when it stopped being a standalone timesheet system and started connecting to everything else: CRM, accounts, document management and design tools. Once a won opportunity flowed automatically into a set-up project, a resource plan and a budget, practices finally had one version of the truth.

AI follows the same rule. A chatbot sitting on its own adds novelty. AI connected to your project history, fee data and documents adds insight, such as spotting which project types consistently overrun or drafting a fee proposal from comparable past jobs.

Before buying any AI tool, ask how it connects to the systems you already rely on. If the answer is copy and paste, keep looking.

5. Start small, prove value, then scale

The "big bang" PSA rollouts I witnessed rarely went smoothly. Switching on timesheets, resourcing, billing and reporting across every office on the same Monday overwhelmed staff and buried early wins.

The better approach was phased: start with time and expenses in one team, prove the benefit, then add project financials, resourcing and forecasting. Each phase built confidence and funded the next.

AI rewards the same discipline. Pick one well-defined use case, such as summarising meeting notes, drafting bid content or reviewing contract risk. Run a short pilot with a clear success measure, learn from it, and only then expand. A few quick wins will do more for adoption than any strategy document.

6. Leadership and measurement keep it on track

Every successful PSA project I have been involved with had a director who owned it. Not an IT manager or an enthusiastic administrator, but a partner or MD who used the dashboards in management meetings and made it clear the system mattered.

They also measured results: utilisation, lock-up, write-offs and project margin before and after. Without a baseline, nobody could prove the investment had paid off, and support quietly faded.

AI needs the same sponsorship and the same rigour. Decide who owns it at board level, agree how you will measure success (hours saved, win rate, risk reduced) and review progress every quarter. Managing Directors will look for growth, Operations Directors for efficiency, and Finance Directors for control. A good AI plan answers all three.

Where AI is different

The similarities are striking, but AI does bring new considerations. It moves much faster: PSA platforms evolved over decades, while AI capabilities change month by month. That makes flexible, connected architecture more important than ever, so you are not locked into today's tool.

AI also introduces new risks around data privacy, intellectual property and professional liability. Who owns an AI-assisted design note, and who checks it? These questions need clear policies, not assumptions.

Finally, AI is often already in use informally, with staff trying free tools on their own. With PSA, nothing happened until the firm chose a system. With AI, the first job may be understanding what is already happening and bringing it under sensible governance.

Five questions to ask before your next technology investment

Whether you are considering PSA, AI or both, these questions will save you time and money:

1. What specific business problem are we solving, and what is it costing us today?
2. Is our project, resource and financial data clean and consistent enough to rely on?
3. Who in the business will own this, and how will we bring our people with us?
4. How will it connect to the systems we already use?
5. What does success look like in six months, and how will we measure it?

After 30 years, my conclusion is simple: technology rarely fails on its own. Implementations fail when they start with the software rather than the business. Get the foundations right, and PSA becomes the launchpad for AI rather than a separate project.