Touch Grass: Why the Best AI Strategy Might Be Knowing When to Switch Off

Lessons from Pascall + Watson's thoughtful approach to AI adoption in architecture

If you could put one message on a billboard seen by millions, what would it say? For Pavan Birdi, Senior Associate at Pascall + Watson and their resident AI champion, the answer came down to two simple words: "Touch grass."

It's a beautifully ironic message from someone who's spent two years implementing AI across one of the UK's leading architectural practices. Here's a professional who can reduce fee proposal preparation from four days to an afternoon using ChatGPT, who's rolling out multi-model AI platforms across his firm, and who's using artificial intelligence to justify design decisions with hard data. Yet his billboard message? Disconnect. Step away. Go outside.

In our latest Project Flux episode, Pavan shared insights that challenge the rush-to-adopt mentality sweeping through construction. You can listen to the full conversation here, but what struck me most was how his approach reflects a deeper wisdom about technology adoption that many firms are missing.

The paradox: Sometimes the best AI strategy includes knowing when to switch it off.

The Strategic Patience Advantage

While other firms dove headfirst into every new AI tool, Pascall Watson took a different approach. When your primary sectors are aviation and defence, you don't have the luxury of treating AI as an experiment. The risk tolerance is fundamentally different.

This patience paid off. When DeepSeek released their language model at a fraction of existing costs, NVIDIA's stock briefly plummeted. Pavan saw validation of their approach—proof that waiting and watching can be more valuable than being first to market.

Key insight: The firms that succeed with AI aren't necessarily those that adopted it first, but those that adopted it most thoughtfully.

Their conservative approach forced harder questions from the start:

• What are the environmental implications? (They're carbon neutral)

• What are the real costs beyond licensing?

• How does this align with our values and risk tolerance?

The lesson: In an industry that rewards the loudest voices, there's real value in taking time to think things through.

Beyond the AI Hype

"I get worried when they're purely focusing on producing these dreamy, amazing imagery which a lot of them come up with flaws."

If you've spent time on architecture Twitter, you've seen them: stunning AI-generated images of impossible buildings that look more like concept art than actual design proposals. They're impressive, often beautiful, and completely missing the point.

While architects get seduced by AI's ability to generate spectacular visuals, they often create more problems than they solve. Those impossible cantilevers might look amazing in a rendering, but they create headaches for engineers who have to figure out how to actually build them.

The contrast is stark: While architects use AI to generate pretty pictures, aerospace engineers use it to solve fundamental challenges at the molecular level. Zoologists use AI to map animal behaviour and decipher communication patterns.

Pavan's approach: Use AI for early-stage brainstorming and concept exploration? Absolutely. Use it as a crutch to avoid making design decisions? That's where things go wrong.

Real value example: When Pascall Watson designed a feature stair for an investment bank, they didn't just propose it because it looked good. They used simulation to demonstrate how it would improve circulation, increase sightlines, and ultimately enhance business performance through better communication between departments.

"When you increase sightlines to people in the workplace, you increase communication. If you increase communication, you improve business performance."

The takeaway: AI's real value isn't in generating spectacular imagery—it's in providing the analytical backbone that helps architects make better decisions.

The Conversation Revolution

Here's a question that reveals everything about how someone uses AI: Do they treat it like Google, or like a colleague?

"If you use it in that way, it's a lot more beneficial. It responds a lot better... thinking of it as your colleague, you know you're working, you've got a back and forth with it to tackle a task, rather than a Q&A session."

Most people fire off quick queries, expect instant answers, and get frustrated when results aren't quite right. But Pavan discovered something different through his fee proposal work.

The transformation: What used to take four days of back-and-forth with directors and clients now takes an afternoon of productive AI collaboration. But this wasn't magic—it was expertise plus iteration.

The key: Pavan could have productive conversations with AI precisely because he brought experience to the table. When AI suggested unrealistic timelines, he could push back and refine. The result? A fine-tuned system that produces realistic programmes and fees.

Scaling the approach: Pascall Watson now uses Omnichat, a platform providing access to multiple AI models (Claude, Gemini, ChatGPT-4), taking the best of each depending on the task.

The insight: Instead of expecting AI to read your mind, invest time in building a productive working relationship. Provide context, iterate on responses, and gradually train the AI to understand your specific needs.

The Human Factor

"People are incredibly complex, they're irrational... Projects change completely right through the lifecycle of the project."

If you've ever managed a construction project, you'll recognise this: You start with a clear brief and aligned stakeholders. Six months later, half the stakeholders have changed, the brief has evolved beyond recognition, and everyone's working toward slightly different goals while insisting they're on the same page.

This is where AI hits its limits. AI excels at logic and systematic approaches, but projects aren't logical systems—they're human endeavours, complete with irrationality, ego, and unpredictability.

The evolution: Project managers aren't being replaced by AI—they're becoming orchestrators of both human and artificial intelligence. They're managing not just schedules, budgets, and stakeholder relationships, but also AI agents with specific capabilities and limitations.

"I can imagine AI agents being implemented into the workflow and project managers now not only managing people and risk and stuff like that, but they're also managing different agents."

The future: Some project managers will become AI-native, developing expertise in orchestrating human-AI teams. Others will specialise in purely human elements—stakeholder management and complex interpersonal dynamics that no AI can navigate.

Both approaches have value, representing different career paths and value propositions in an AI-enabled industry.

Data as Your Competitive Edge

Here's the question keeping architecture firm leaders awake: If everyone has access to the same AI tools, how do we differentiate ourselves?

Pavan's answer: It's not about the tools—it's about the data.

"Different practices have their niches and they have their expertise and they have their experiences on projects... At Pascall's, you know, biggest sector is aviation, right, they've been doing it since the 90s."

The competitive moat: While anyone can subscribe to AI platforms, not everyone has Pascall Watson's three decades of aviation expertise, their specific approaches to passenger flow analysis, or their accumulated knowledge of what works in airport terminal design.

The shift: "LLMs being more important than your website. You know a website is your identification right... But the quality of your service, the quality of what you will do, will be all based on your data that you're putting into your LLM."

The challenge: Data quality becomes everyone's responsibility. When project managers document lessons learned, when architects capture design rationale, when cost consultants record value engineering decisions—all of this becomes potential training data for AI systems.

The opportunity: Firms that invest in systematic knowledge capture and curation will develop AI capabilities that truly differentiate their services.

The Balanced Future

"While [AI] is there to assist us and aid us, we should be able to escape from it as well. Turn off your device."

Coming from someone implementing AI across an architectural practice, this might seem odd. But Pavan's "touch grass" philosophy represents something deeper—recognition that the most effective AI strategies include knowing when not to use AI.

The healthcare insight: A recent study found AI was more empathetic than human healthcare workers in patient calls. But this says more about stress and workload pressures on human workers than about AI's capabilities.

The real value: When AI handles routine tasks that consume professional time and energy, it frees people to focus on work requiring human insight, creativity, and relationship-building skills.

"When I think about the way I approach project management or architecture... it's about dealing with people more than anything."

The philosophy: The goal isn't to become more machine-like—it's to become more human. Use AI to handle systematic, repetitive work while focusing human capabilities on creativity, judgment, and interpersonal skills.

Key Takeaways: The Pascall Watson Model

Start with strategy, not tools: Understand your risk tolerance, values, and objectives before adopting any AI platform.

Treat AI as a conversation partner: The most productive interactions are iterative dialogues building on expertise and context.

Focus on data quality: Your competitive advantage comes from the quality of data and expertise you bring to AI tools.

Embrace experimentation with human focus: Give people space to discover valuable applications while remembering the goal is enhancing, not replacing, human capabilities.

Know when to disconnect: The most effective AI strategies include stepping away from algorithmic assistance to engage directly with human elements.

This conversation with Pavan was one of our most thought-provoking episodes yet. The full discussion covers even more ground, including his thoughts on AGI and why his theme song right now is "Staying Alive" by the Bee Gees. Listen to the complete episode here.

The future belongs not to those who use AI most extensively, but to those who use it most thoughtfully. And sometimes, the most thoughtful thing you can do is turn off your device and touch grass.

If you found these insights valuable, you'll love our weekly newsletter where we dive deeper into practical AI applications in construction and project delivery. We cut through the hype to focus on what actually works, sharing real-world case studies and actionable strategies from industry leaders like Pavan. Subscribe here to stay ahead of the curve with insights that matter, not just headlines that grab attention.