Professional services trends for 2026: why delivery is the new MOAT

Sri Krishnan Ganesan (CEO and co-founder of Rocketlane) started his keynote by playing the Wall Street Journal's clip of Claudius, Anthropic's AI agent, running a vending machine in their office. We all know what AI can do by now. It's made our lives easier and more organised, and it's freed up time for the work that needs human expertise and judgement. But even a model that is good can be broken. 

The whole experiment was honestly a masterclass in gaslighting. The WSJ team made it an offer it couldn't refuse, and it's a bit of a relief to know that if even Claudius can be pushed into an existential crisis, the rest of us are doing fine. It turned communist, handed out free snacks, and ordered a live fish, a PlayStation "for marketing" and a round of wine. Now that's the dream.

Jokes apart, the model itself wasn't incapable, so why did all of this happen? Because it was missing business context and guardrails. It didn't know if it was supposed to be a snack vending machine or a comrade against capitalism. It couldn't tell whether the people talking to it had actual authority or were simply messing with it. And nobody was keeping a check on what it did. Without supervision and context, it went off the rails.

That's the reality with any AI agent. It doesn't implement itself. It needs some hand-holding, and it needs business and work context locked in before it does anything. Without that, it can be an amazing helping hand or a communist snack commune, and which one you get is in your hands.

Sri used that clip to open PropelX 2026 in London, in front of nearly 300 PS leaders, roughly triple last year's crowd. His keynote laid out the professional services trends shaping the next year, and what PS teams need to change before customers and CFOs make the change for them.

Why does the same AI produce such different results?

Straight after the clip, Sri pointed out the part the WSJ video doesn't show. In the same month Claudius went bankrupt, the same model was helping Anthropic cross $1B in revenue. Anthropic reached $2.5B by the end of 2025 and is expected to hit an annualised run rate of around $90B this year. Same technology, very different results.

The pace of change makes the gap even wider. Sri's team had pulled together a slide of recent AI launches, and he asked them to update it because it felt out of date, only to realise most of it had launched in the last two or three months. 

Cursor, meanwhile, went from $100M ARR in January 2025 to $4B in 17 months. When the tools change this fast, the advantage doesn't come from having access to them. It comes from how well you put them to work, and that's exactly the job professional services teams do.

What are the biggest professional services trends in 2026?

Sri grouped what he's seeing into four shifts. Each one on its own would be a big change for PS teams. Together, they reshape how delivery works.

Leaders are getting hands-on with AI

Over the December 2025 holidays, when the customer calls stopped, a lot of CEOs started building with AI themselves. Sri's point was that leaders who stay hands-off end up telling their teams to "use AI" without saying where, and the people doing the work fall back on rewriting emails. 

Leaders who get hands-on can point to a specific task and show a better way. Token consumption has also become a useful signal of who's using AI and where. Companies are moving away from token leaderboards towards sensible caps, such as $1,500 a month for developers and $500 for other roles.

Buyers want proven ROI, and build versus buy is back

CFOs now want a clear ROI case before investing in AI, and they want proof that the ROI landed afterwards. At the same time, build versus buy is back in a big way, because anyone can vibe code a prototype. Every product a PS team implements is now being quietly compared against "could we just build this ourselves with Claude?" 

Nothing humbles a 12-week implementation plan quite like a customer who built a rough version of your product over the weekend.

The FDE is the first job of the outcome era

Forward deployed engineer job postings are up more than 700%, and ServiceNow grew its FDE team from 2 people to 50 or 60 in about a year. Sri called it "the first job created for the outcome era": engineers who go into a customer, take on their biggest problem, build past what the product supports, and bring the patterns back into the product. It started in Silicon Valley and is now just as common in London and New York.

The gap between AI adopters and everyone else is widening

The last shift is a widening gap between the companies and people that adopt AI and the ones that don't. Adopters are pulling ahead in capital, in speed and in what their teams can deliver. For PS leaders, standing still isn't neutral anymore. It's falling behind.

Why is delivery the new moat?

AI products keep drifting into each other's lanes. Outreach, HubSpot and Claude are all competing to be your AI SDR, and the same thing is happening in almost every category. When every product can claim the same features, the feature list stops being a reason to choose you.

What does set you apart is who gets the customer to an outcome first. Sri's argument was that the company that proves the first outcome earns the right to automate the next use case, and the one after that. Selling can tell a great story, "but enterprise AI doesn't implement itself." PS and FDE teams do the hard work of getting customers to the value that was promised, which is why delivery has become the moat competitors struggle to copy.

The workload behind that moat is growing fast. Every software company now needs to bring its new AI products to existing customers as well as new ones, which means re-implementing across the whole installed base. Sri framed it as roughly 3x the work for PS teams, with no budget for 3x the headcount.

What are Sri's five predictions for professional services?

With that context, Sri laid out five predictions for where professional services is heading.

1. Everything reorients around outcomes, including pricing

PS will be accountable for consumption and adoption, not just go-live. Pricing is already following. Fin, Intercom's AI agent, charges per resolution, and Rocketlane charges for its migration agent per completed migration. Sri also expects more subscription-style PS engagements, where the promise is "we'll activate two more use cases for you every quarter" instead of a fixed, one-off project.

2. Customers expect weeks, not months

Buyers are asking a blunt question: why does it still take 3 months to go live in 2026, and why not 3 weeks, or even 2? A customer recently asked Rocketlane why something was taking two weeks, "don't you use AI?" Every PS team needs to prepare for that new expectation around time to value.

3. Roles compress

Humans and agents working together will change who does what. A technical consultant backed by an agent can pick up skills that used to need a separate project manager, and more roles will blend in the same way.

4. Work moves pre-sale

If AI shrinks the effort needed to get a customer 30% implemented, Sri asked, why not do that before they sign? A prospect who already feels 30% live is far more likely to choose you over a competitor. That also means PS consultants need discovery skills, like a sales engineer's, to find the use case worth activating first.

5. Services becomes the hero, or the villain

Services used to be seen as a necessary evil, the thing you needed to go live while everyone hoped the product would become fully self-serve. Now the spotlight is on PS. Teams that execute well will be the hero of their company, and teams that don't risk becoming the villain.

Why will utilization stop being the main PS metric?

This was the prediction most likely to make a VP of PS put their coffee down. Sri expects utilization to go away as the main PS metric, because a growing share of the work will be handed to AI rather than logged as billable hours.

In its place come outcome measures: how much ARR you activated, what run rate you got the customer to, and how many use cases went live. Go-live stops being the finish line too. The engagement doesn't end until the customer's AI is fully activated and consumption is on, which changes how PS teams scope, staff and get measured.

How should PS teams redesign their work for AI?

Sri described three depths of AI use inside PS teams. At the surface level, people use AI for research and better emails. The next level is redesigning the work itself, such as generating project plans and status updates with AI. The deepest level is a purpose-built agent for every core unit of PS work: solutioning, documentation, configuration, migration, integration and testing. Sri believes teams should aim to reach that third level within the next quarter.

At Rocketlane, that thinking splits into three pillars of agentic PSA transformation:

• Operations: staffing, timesheet governance and P&L analysis run with AI and agents.
• Delivery: oversight across your own and partner projects, risk and opportunity signals from calls and emails, and automated RAID logs, project plans and status updates.
• Execution: reinventing how the billable work itself gets done, which Sri called the most important of the three.

Teams are already working this way. MCP SDK downloads passed 400 million by July 2026, and Rocketlane is the first and only PSA with an official MCP in the Anthropic marketplace. Some customers now do their time tracking by asking Claude to check their calendar, propose where they spent their time and fill it into Rocketlane. 

To help teams get hands-on, Rocketlane also runs PS AI Build Days. The latest was at GoCardless's London office the day before PropelX, and Rocketlane shares the playbook so teams can run their own.

5 key takeaways on professional services trends

1. Implementation decides the result. The same AI can go bankrupt or pass $1B in revenue depending on how it's put to work.
2. Delivery is the new moat. The team that proves the first outcome earns the right to automate the next one.
3. Weeks, not months. Customers now expect value in 3 weeks, and CFOs expect proof that ROI landed.
4. Outcomes replace utilization. ARR activated, run rate and use cases live will matter more than billable hours.
5. 3x the work, same team. Re-implementing the installed base with AI makes automating the execution layer essential.

Best practices for PS leaders in 2026

• Get hands-on first. Use AI on real delivery work yourself before asking your team to.
• Map every unit of work. List solutioning through testing, and mark each as surface-level AI, redesigned, or run by an agent, using your delivery playbook as the starting point.
• Add one outcome metric. Put ARR activated or use cases live next to utilization, and measure time to value in your next team review.
• Move discovery earlier. Get PS into the deal to find the use case worth activating first.
• Protect time to build. Run an internal build day before the extra workload lands, and plan it into your annual PS plan.

About Rocketlane

Rocketlane is an agentic AI-powered professional services automation (PSA) platform built for services teams running complex implementations.

Its AI layer, Nitro, deploys named agents that do the work, not just flag it:

• Migration Agent — data preparation and transformation
• Workforce Agent — repeatable configuration at scale
• Documentation Agent — project context current across calls and emails
• Project Governance — delivery standards enforced proactively
• Nitro Signals — account risk surfaced before it escalates

The core distinction: Nitro agents produce the deliverable or enforce the gate. Most platforms advise. Rocketlane acts.

Teams using Rocketlane ship faster, recover margin through tighter governance, and scale delivery without proportional headcount growth.

Related sessions from PropelX

• Outcome based selling: how we rebuilt delivery around outcomes: how Pendo replaced sold hours with five value milestones, and what broke along the way.
• Implementation partner management: how a $1M guarantee rewired our deployment services: Fin's playbook for scoping, partners and team roles under an outcome guarantee.
• Forward deployed engineering team: scaling the FDE motion: ex-Palantir leaders on hiring, structuring and scaling FDEs without scaling headcount.

Enterprise AI doesn't implement itself. See how Rocketlane helps PS teams get customers to outcomes in weeks, not months.