Last week a guy named Raoul reached out to me via WhatsApp. He runs a company called Land First Investments. He buys cheap farmland, spends 3–7 years getting planning permission, then sells the land to housebuilders for 40–250 times what he paid. His entire operation runs on Excel, PDFs, and a secretary organizing folders. He told me: “Instead of hiring someone, why don’t I just try AI?”
I went to his office. I sat next to him at his computer. I watched him show me his file system, his Excel “CRM,” his stack of PDFs. And then I started looking into his industry. What I found is that there are about 67 companies doing exactly what Raoul does, they all run the same way, and nobody has built tools for them. Not a CRM. Not a dashboard. Not AI. Nothing.
This document is everything I’ve learned. I need the team aligned on whether this is the vertical we go after.
Land promotion is one of the most profitable businesses in the UK that most people have never heard of. Here’s how it works in plain English:
A farmer has 50 acres of land. As agricultural land, it’s worth about £10,000 per acre — so roughly £500,000 total. But if that land got planning permission for houses, it would be worth £500,000 per acre or more. That same 50 acres is now worth £25 million.
The problem is that getting planning permission in the UK is incredibly difficult. The system is “discretionary” — meaning unlike American zoning (where rules tell you what you can build), every UK planning application is judged on its merits by a human planner. Every local council operates differently, timelines shift constantly, and the process involves years of regulatory submissions, environmental assessments, legal negotiations under something called Section 106 (bespoke legal agreements covering affordable housing, schools, infrastructure — negotiated individually for every single site), and public hearings. All of this is governed by a national policy document called the NPPF (National Planning Policy Framework), which itself gets revised every year or two. Only about 40% of English councils even have an up-to-date local plan.
Land promoters are the people who navigate this system. They either buy the farmland outright and promote it themselves (the freehold model — high risk, massive reward), or they sign promotion agreements with landowners where they do all the planning work in exchange for 15–30% of the eventual sale price. Either way, once planning consent is secured, they sell the consented land to housebuilders like Barratt, Persimmon, or Taylor Wimpey.
The value uplift is staggering. A 50-acre site bought for £500K can sell for £25 million with consent. Even after £1.5 million in consultant costs (planning, environmental, legal), the net profit can be £22 million on a single deal.
This is why Raoul operates from a premium Marylebone office (2 Portman Street, W1) despite having only 5 employees. The deals are enormous. The margins are extraordinary. And the bottleneck isn’t capital — it’s the bandwidth to manage the pipeline.
Raoul Ragoowansi is an ex-JP Morgan banker who founded Land First Investments in 2016. He built a team of heavy hitters from the UK housebuilding industry: a former Regional MD of Bovis Homes (40+ years, responsible for £250M+ in land acquisitions), a former partner at Barton Willmore (the UK’s biggest planning consultancy, now Stantec), and a former Land Director from Taylor Wimpey. Five people with over 100 years of combined experience.
They currently have 8 active projects across England and have completed 7 past developments, including a flagship 150-unit scheme at Padgbury Lane, Congleton that went through the full planning cycle (initial approval in principle, then detailed design approval, then sale to a housebuilder in 2020).
How Raoul sources deals: 45% come from what he calls “bespoke software” (likely a custom land search tool, not operational software), 45% from land agent relationships, and 10% from open market. His appraisal process involves planning policy analysis, regional valuation, environmental and technical review, and residual financial appraisals — all done manually with PDFs, Excel, and Word documents.
When I visited on March 31, Raoul sat me at his computer and walked me through his entire operation. What I saw was:
“The deal flow is increasing. The velocity of work is increasing. I’m about to hire someone. But instead of hiring someone, why don’t I just try AI out and see if it could scale me out?” — Raoul Ragoowansi, in-person meeting, 31 March 2026
He came to me via referral from Rohin Bhuptani, who told him I was “the go-to guy for AI.” We’ve had a Google Meet call and an in-person meeting. He’s booked in for a hands-on session this Easter (April 6) where I’ll sit with him for 1–2 hours and demonstrate what’s possible.
Raoul is not unusual. He is a textbook example of an entire industry segment.
The UK has a trade body called the LPDF (Land, Planning and Development Federation) with 67 member companies and 70 affiliate members. These are the land promoters. They have an annual lunch, a summer party, and a member directory. It is a small, tight, relationship-driven community.
I researched as many of these companies as I could find. Not a single company website mentions CRM software, project management tools, or any operational technology. Every site talks about “relationships” and “experience.” They all describe their processes in manual terms. The government is actually digitizing the council side of planning faster than these companies are digitizing themselves.
The site sourcing market is well-served. Tools like LandInsight (£200–500/mo, 5,000+ users), Nimbus Maps (£150–390/mo, 10,000+ users), and Searchland (£195/mo+) help land promoters find sites. That’s the easy part. What nobody has built is the system for managing those sites through a 3–7 year planning journey across multiple local authorities, consultant teams, legal processes, and stakeholder relationships.
Land promoters control 41% of planning permissions on strategic sites (100+ units) in England. They punch far above their weight relative to their size. The sector is small in headcount but enormous in value — the aggregate land value flowing through promotion agreements runs into billions annually.
The market structure breaks into three tiers. At the top are institutional players like Gladman (acquired by Barratt for £250M in 2022), Hallam Land (Henry Boot), and Catesby Estates (Wellcome Trust) with 30–100+ staff. In the middle are firms like Ainscough, Pigeon, and Lone Star (acquired by Persimmon in October 2025) with 10–30 people. And at the bottom — our target — are 40–60 boutique operators like Land First with 3–15 people, founder-led, privately funded, managing portfolios of 5–50 sites.
These acquisitions are important context: housebuilders are paying hundreds of millions of pounds to acquire land promoters because the planning pipeline is that valuable. This validates the sector’s worth. But it also means independent operators like Raoul need to move faster and manage better to compete — and they’re trying to do it with Excel.
Five reinforcing factors keep this industry manual:
Here’s the critical nuance: the relationship part cannot be automated, and we’re not trying to. Raoul will always drive to a farm to shake a landowner’s hand. What we automate is everything surrounding that relationship: the document processing, the pipeline tracking, the planning monitoring, the report generation, the deadline management. The human does the deal; the AI manages the system around the deal.
Our total addressable prospect list. Green = our client. Blue = London-based targets we’ve identified as closest to Raoul’s profile.
We need to understand how much money flows through Raoul’s business, because that directly determines what we can charge and which pricing model makes sense.
A single Land First deal on the freehold model looks like this: buy 50 acres for £500K, spend £1.5M on consultants over 3–5 years, sell the consented land for £25M. Net profit: roughly £22 million. That’s an 11x return on capital. On the promotion agreement model (working for a landowner), the promoter takes 25% of net proceeds — about £4.4M profit on the same site.
Why this matters for us: Raoul is benchmarking AI against hiring another person (£60–80K/year). But his actual economics operate on a completely different scale. If our system helps him close one additional deal, or reduces the cycle time on an existing deal by six months, the value to him is measured in millions, not thousands. This is why outcome-based pricing is the right model — we should tie our compensation to measurable results, not hourly rates or flat retainers.
He also spends £500K–1.2M per site on consultants (planning: £50–150K, environmental impact assessment: £150–250K, planning application fees: ~£290K, legal: £30–100K). Any automation that reduces consultant dependency or accelerates their work is worth a percentage of those savings.
Now that we understand the money flowing through this industry, let’s look at why nobody has built software for the core workflow.
I mapped every PropTech and PlanTech tool I could find. The site sourcing category is well-served (LandInsight, Nimbus Maps, Searchland, LandHawk). Planning application monitoring has some tools (Trackanapp, PlanWatch). Development appraisal has Aprao for SMEs and ARGUS for enterprises.
But the core land promotion workflow — the thing these companies spend 90% of their time doing — has zero purpose-built tools. Here are the seven specific gaps:
The closest AI competitor is PlanningBot — an early-stage tool that orchestrates 15 parallel searches across government APIs and cross-references against NPPF guidelines. It connects to AI tools via an open protocol called MCP (Model Context Protocol — the same standard our agents use). But it’s focused on planning consultants, not land promoters. It helps with one-off analysis, not ongoing pipeline management. It’s a tool, not a system.
How the 7 gaps map to our Easter experiments: Gap #1 (no CRM) → Experiment 2 (pipeline view). Gap #3 (no document AI) → Experiment 1 (document ingestion). Gap #5 (no risk dashboard) → also Experiment 2. Gap #6 (no drafting assistance) → Experiment 4 (document generation). Gaps #2, #4, and #7 are Tier 2–3 capabilities we’d build over weeks and months, not in a single session.
There is also a tailwind from government: the Planning & Infrastructure Act 2025 is pushing councils to digitize, and nearly 200 councils are participating in the Open Digital Planning programme. As council data becomes more structured and accessible, the opportunity to build AI on top of it grows. We’d be building on a foundation that the government is actively creating.
Our approach is service-first, not software-first. We embed AI agents into Raoul’s operation. We don’t hand him a login to a dashboard and walk away — we deploy a system that actively works alongside his team. The agent knows his deals, monitors his deadlines, extracts data from his documents, and surfaces what needs attention.
This is exactly what Alavida already builds: an AI agent deployed with domain-specific skills, a data layer, and communication channels (WhatsApp, Slack, email). The technology stack doesn’t change. What changes is the knowledge layer — instead of generic GTM or content skills, we train agents on the UK planning system, NPPF policy, local authority processes, and Land Registry data.
The Easter session is not a demo. It’s an experiment. We’re testing whether our stack can deliver real value in this domain, and we’re learning what matters most to him. Here are the specific things we want to try:
After the session, we learn what resonated and what didn’t. This shapes the engagement model. If document ingestion is the killer feature, we lean into that. If the pipeline view matters more, we build there first. We let Raoul’s reactions determine the product, not our assumptions.
This is the question that determines whether Land First is a one-off consulting gig or a vertical market play. The answer: there are at least 40–60 companies that match Raoul’s profile almost exactly.
| ICP Attribute | What It Means | Why It Matters |
|---|---|---|
| 3-15 employees | Lean team, founder makes all decisions | No procurement committee. We sell to one person. |
| Privately funded | Family office or self-funded, not VC-backed | Comfortable spending on tools that work. Don’t need board approval. |
| 5-50 active sites | Enough complexity that Excel breaks down | This is the pain threshold. Below 5 sites, Excel works. Above 5, it doesn’t. |
| Decision maker is ex-finance or ex-housebuilder | Understands ROI, comfortable with numbers | Outcome-based pricing resonates. They think in returns, not hourly rates. |
| LPDF member | Connected to the industry network | One happy client generates referrals to peers at LPDF events. |
| No operational technology | Excel, email, shared drives, maybe LandTech for sourcing | Low switching cost (from nothing), high impact from any improvement. |
I identified 9 London-based operators that match this profile closely: Danescroft (W1S), Welbeck (EC2Y, £140M under management), Sigma (WC1B), LSL Partners (£200M invested), Generator Group (JV with Topland), Valorem (£80M deployed), St Congar (9,000+ home pipeline), Landhold Capital, and RO Land. All W1/EC/WC addresses — financially sophisticated, ex-finance or ex-housebuilder directors. Raoul’s peer group.
Beyond the core 40–60, there are 50+ adjacent firms with identical problems: planning consultancies (DLP, Boyer, Edgeplan, Marrons), rural land agents (Carter Jonas, Strutt & Parker, Bidwells), and strategic land divisions within large agents (Savills, Knight Frank). These are expansion targets, not first customers.
The GTM is a straight line because the market is small, concentrated, and relationship-driven. There is one trade body. One network. One set of events. If we deliver for Raoul, we have a case study. If we have a case study, we have access to the LPDF annual lunch and summer party. If we present there, we’re in front of the entire addressable market at once.
Easter session. Hands-on, in-person. We run the experiments described above. If it works, we agree on an engagement. The first month is about delivering real value and building the case study.
Once Raoul is seeing results, we ask him directly: “Who else in the LPDF has the same problems?” He knows these people. They see each other at industry events. A warm intro from a peer is the highest-converting sales channel in a relationship-driven industry. Target: 2–3 additional clients.
Present the case study at an LPDF event. Position Alavida as “the AI system built for land promoters.” Direct outreach to remaining LPDF members with the case study as proof. We need to verify this is possible — does the LPDF allow non-member presentations? Can Raoul sponsor our presence?
Planning consultancies, rural land agents, small-to-mid developers without in-house tech. Same problems, different business model. The skills and knowledge we build for land promoters transfer directly.
We should not lead with a flat monthly retainer. That commoditizes what we do and caps our upside. The right models for this market are:
The Easter session is where we start measuring. We watch what Raoul reacts to, we quantify the time savings on specific tasks, and we build the pricing model around real data — not assumptions.
If we validate the model with 3–5 clients and move to verticalize:
The TAM ceiling on land promoters alone is modest (£3.6M). This is a beachhead, not the end state. The adjacent verticals are larger. And the playbook — embedding AI agents into document-heavy, relationship-driven, high-margin professional services — is repeatable far beyond property.
| Risk | Severity | Our Thinking |
|---|---|---|
| Raoul’s engagement doesn’t convert to paid work | Medium | Even if the Easter session is free, we learn whether our stack works in this domain and we get case study material. The downside is 2–3 hours of time. |
| The TAM is too small to matter | Medium | £3.6M in land promoters alone is modest. But adjacent verticals expand it 3–5x, and the playbook transfers to other industries. This is a beachhead play. |
| The industry is too traditional to adopt AI | Medium | We take this seriously. 81% of real estate firms cite data complexity as a barrier. But Raoul is already asking for AI. The “hire vs AI” decision is happening now. We’re not pushing — he pulled. |
| A PropTech player builds the CRM before us | Low | LandInsight and Nimbus focus on sourcing. They’d need a major pivot. PlanningBot is the closest AI entrant but targets consultants, not promoters. We have a head start if we move now. |
| Regulatory change reduces deal flow | Low | The UK has a structural housing shortage. The government is making planning easier, not harder. The Planning & Infrastructure Act 2025 is increasing deal velocity. |
Read this. Understand the opportunity. Come to our next sync with a point of view on:
The meeting is Easter Sunday or Monday. I’ll update this document with results.