The future of RICS reporting: where AI fits and where it stops

A year or so into building Otto, we have a clearer view of where AI can improve the RICS reporting workflow and where it should not be involved at all. This post is our honest account of that distinction.

What AI is good at in this context

RICS reports have a fixed structure. Level 2 HomeBuyer Reports and Level 3 Building Surveys both follow standard element frameworks. The sections are defined; the required content for each section is defined; the condition rating scale is defined. That structure is well-suited to AI assistance, because the task is not open-ended.

Given a set of observations (from dictation or photographs), the job of mapping them to the right section and drafting clear prose for a non-specialist reader is something AI models do well. It is not a judgement call. It is translation and organisation.

Similarly, learning how a specific surveyor writes and applying that pattern to new content is a tractable problem. The surveyor's voice is learnable because it is consistent. That is what StyleProfileService does.

Extended context handling, which became more capable in Otto v2, means the model can hold a complete complex survey in working context and maintain consistency across sections. References to earlier observations, consistent terminology throughout, appropriate cross-referencing: these are things that become easier with better context handling, and they are things surveyors currently have to manage manually.

Where AI should not be in control

The condition rating is the clearest line. Rating an element Condition 1, 2, or 3 is a professional judgement with legal standing. It requires physical presence, contextual knowledge, and the kind of pattern recognition that comes from years of inspections. A photograph, however good, cannot substitute for being in the building. Neither can a transcription.

We built RatingFidelityChecker precisely because this line must be enforced mechanically, not just as a policy. No AI output in Otto can alter a condition rating the surveyor has set. That is not a limitation we imposed reluctantly. It is the correct design.

The broader judgement calls in a Level 3 Building Survey follow the same logic: whether a crack is structural or cosmetic, whether damp is rising, penetrating, or condensation, what the appropriate sequence of remedial works would be. These require the surveyor's training, experience, and physical assessment. An AI model reading a description of a crack is not equipped to make that call reliably enough for a professional document.

What Otto v2 changed

Otto v2, which launched in June 2026, brought two significant changes. Extended context handling allows complete Level 3 Building Surveys to be drafted with greater consistency across the longer document structure. Auto-mode reduces the number of manual steps between completing an inspection and having a reviewable draft, by applying the dictation structuring, image analysis, and style profile in a single pipeline rather than sequential manual triggers.

What did not change in v2 is the surveyor's role as the professional author of every report. Faster and more consistent does not mean less supervised.

Our view on where this goes

The surveyor profession is not being automated. The inspection, the judgement, the professional relationship with the client, and the liability for the document are all things that sit with a chartered surveyor and will continue to do so.

What is changing is the ratio of inspection time to desk time. AI drafting tools, when well-designed, shift that ratio in favour of the inspection. A surveyor who spends less time converting notes into sentences and more time doing thorough inspections is better placed to do good work.

That is the direction we are building in. Not to produce reports without surveyors, but to make the report-writing part of the job take less of the day.

You can read more about Otto and the rest of what Home builds for the surveying profession at RICS surveyors on Home.

Further reading: RICS on home surveys.