"AI-powered" has become one of those phrases that tells you almost nothing on its own. Every land sourcing platform seems to have some version of it now, which is either a sign the industry has genuinely moved forward, or a sign that "AI-powered" has become the new "cloud-based," and everyone's slapping it on the box regardless of what's underneath.
The honest answer is a bit of both. There are a few genuinely different ways platforms are building AI into land sourcing right now, and the differences between them matter far more than whether a tool has "AI" in its feature list at all.
Here's how the main approaches actually compare, and where LandInsight's AI Assistant sits.
A general-purpose AI model (think ChatGPT or Claude under the bonnet) is wrapped in a chat window inside the platform's interface. You ask it a question about a site, and it answers based on whatever it's been trained on. It's often the quickest feature for a platform to ship, since it doesn't require building any planning-specific logic. The trade-off is that the chatbot has no particular allegiance to your platform's own data, and no built-in way to show you where an answer came from.
Rather than embedding a chatbot, the platform exposes its data to an external AI model via a connector, commonly MCP (Model Context Protocol), and lets that model do the reasoning. This is a genuinely clever piece of engineering, and it means you get the full capability of a frontier AI model working against real data. The catch is that the reasoning still happens outside the platform's own guardrails, which introduces a risk of drift - the model's answers becoming less reliably grounded in the platform's actual data over time - and puts more of the burden on you to phrase things precisely to get a reliable answer.
The AI logic is built natively, using the platform's own data and rules written specifically for planning and land acquisition, rather than borrowing a general-purpose model's default behaviour. It's the slower approach to build, and arguably the least flashy in a demo, but it's the only one of the three that can realistically guarantee every claim traces back to a specific, checkable source, because the whole thing was built around that requirement from day one.
None of these is inherently the "AI is smarter or dumber" question. It's a question of where the accountability sits, and whether that accountability was designed in from the start or bolted on afterwards.
|
Embedded Chatbot |
Connected Model (MCP) |
Purpose-Built Engine |
|
|
Data source |
Model's general training data |
External AI model, connected to platform data |
Platform's own institutional data, kept current centrally |
|
Verification |
Rarely offered; conclusions stated without sourcing |
Inconsistent; depends on the connected model's behaviour |
Source-level citations built into every output |
|
User effort |
Low setup, but answers can be inconsistent |
Higher; benefits from prompt engineering skill |
Low; built to accept plain, unpolished instructions |
|
Workflow |
Usually a separate chat window or tab |
Often requires a separate AI client or interface |
Native to the platform; no export or re-upload |
|
Document handling |
Typically a transient chat thread |
Depends on the external client used |
Persistent document saved against the site |
|
Cost |
Often bundled into the platform subscription, though usage limits or upsell tiers are common |
Requires a separate AI subscription alongside the connector, plus usage costs - effectively a double cost |
Included natively; no separate AI subscription or token meter |
The AI Assistant, live now for LandInsight Unlimited subscribers, was built as the third approach: purpose-built, anchored exclusively to LandTech's own institutional-grade property data, rather than reasoning from a general AI model's training data or an external connection.
Every document it generates carries source-level citations built in, so a claim about flood risk, a listed building designation, or an NPPF tilted balance test links straight back to the actual record behind it, rather than asking you to trust the summary. It's also built to accept plain, ordinary descriptions of your project, because the planning logic was hardcoded by our in-house property planners rather than left for you to prompt your way toward a usable answer. And because it's native to LandInsight, every document it produces is saved against the site it belongs to, without exporting anywhere or juggling a second subscription.
The AI Assistant is designed to speed up the early stages of site sourcing and appraisal. It's not a replacement for the judgement your team applies at the sharp end of a high-value transaction, and it was never built to be.
Whatever platform you're evaluating, the more useful question isn't "does it have AI?" It's "which of these three approaches is it actually using, and can I check its working?" Once you know the answer to that, most of the rest of the comparison sorts itself out.
Try the AI Assistant on your own saved sites inside LandInsight Unlimited, and see how it holds up against that standard.