The Planner’s Assistant is an open-source framework designed to assist with two interrelated domains of planning practice: development management and Local Plan design. It is not a product or platform in the commercial sense, but a structured intervention into how planning systems operate — particularly how they justify decisions, structure discretion, manage constraint, and hold institutional memory. It models the planning system not as a linear data pipeline, but as a complex assemblage of documents, actors, geographies, and evolving obligations.
In development management, the system supports officers in interpreting policy, assessing site-specific constraints, constructing planning balances, and preparing draft decision texts. It treats planning not as data retrieval, but as an entangled process of reasoning, referencing, spatial analysis, and documentation — across policies, material considerations, and the granular messiness of sites and schemes. The architecture reflects this: modular chains combine document classification, policy logic extraction, discretionary modelling, and drafting support. Outputs are not answers, but legible, overridable scaffolds for professional judgement.
In Local Plan design, the system supports early-stage drafting, coherence checking, and spatial strategy development. It includes automated indexing of policy text, semantic classification of objectives and requirements, and structured plan logic tracing — designed to surface inconsistencies, latent contradictions, and misalignments between proposed allocations, infrastructure capacity, and overall intent. At the site level, it draws together constraints, delivery trajectories, and evidential baselines to identify risks of overcommitment or political unviability. The aim is not to automate plan-making, but to provide a reasoning substrate — a set of procedural and analytical tools that enhance institutional capacity under regulatory and political strain.
The AI components within the system are deliberately traceable and falsifiable. Each inference is linked to its inputs — whether a paragraph of policy, a geometry, or a modelled threshold — and each structured output is designed to be challengeable. This reflects a core theoretical assumption: that planning is not merely a regulatory function, but a system of justification operating under democratic constraint. Tools that obscure reasoning — however optimised — risk undermining legitimacy. Tools that surface, document, and expose it may not eliminate disagreement, but they allow it to be situated and contested.
The Planner’s Assistant draws from spatial decision support systems, computational planning theory, public law, and software engineering — particularly principles of explainability, modularity, auditability, and failure resilience. It is not agnostic: it is aligned with a normative view that planning decisions should be explainable, defensible, and open to scrutiny — even (or especially) when discretion is involved.
This tool does not aim to “fix” planning in the reductive sense. Instead, it offers a framework for working within its institutional limits: to help the system retain coherence, traceability, and public defensibility — even when political, procedural, and evidential pressures pull in opposing directions.