Dedoctive

What is Dedoctive?

Dedoctive is a UK neuro-symbolic AI platform that answers questions only from curated, verified evidence and links every claim to its exact source. It is built for organisations whose decisions must survive an audit, an appeal or a safety review. Instead of trusting a language model’s memory, it first turns documents, data tables and images into a checked knowledge model.

The platform wraps any large language model inside BPMN process models, so each AI step follows a mapped route with human checkpoints where judgement is needed. Results can appear as Response Maps, custom reports or formal diagrams such as GSN safety cases, and each one keeps hyperlinks back to the evidence.

The site credits Dedoctive Ltd and Collaboration Tools Ltd in Somerset, and founder Keith Harrison-Broninski writes its blog. Development has been backed by Innovate UK, DASA, ITEA 4, Ufi VocTech Trust and NOCN, and the Ministry of Defence has published a case study on its use. The product was live and accepting pilot requests in October 2026.

Key Features

Dedoctive’s most important features are source-level provenance and governed workflows, because together they let a reviewer see both the evidence and the steps behind an answer.

• Source-level provenance: Every answer carries hyperlinks to the passage, table cell or region of an image that supports it. Reviewers can click through and check the original evidence in seconds, rather than taking the model’s wording on trust or searching the files by hand.
• Curated knowledge models: Documents, data, tables and images are indexed into a structured knowledge model before any question is asked. Content is checked against the GOV.UK data quality dimensions, so material is meant to be correct, complete, consistent, current, concise and checkable.
• No-answer fallback: The system declines to reply when the knowledge model holds no supporting evidence. This trades coverage for safety, which matters in compliance or clinical work where a confident but invented answer could cause real harm.
• Agentic BPMN workflows: AI agents run inside visual BPMN process models that mix fixed rules with agent calls. The DedoctiveDesigner builds these flows from UserTasks, ScriptTasks, AgentTasks and DocumentAgent calls, and dynamic debugging helps you test changes quickly.
• Human-in-the-loop guardrails: Workflows can pause at set points and hand a decision to a named person before continuing. This keeps accountability with staff, and the BPMN map shows auditors exactly where a human approved or changed an output.
• Response Maps and diagrams: Results can be presented as Response Maps, custom reports or formal diagrams such as GSN safety arguments and mind maps. Each node keeps its hyperlinks, so a safety case or compliance audit can be checked line by line against its sources.

Pros

The main reason to choose Dedoctive is traceability, since every claim can be checked against a specific source before a decision is signed off.

• Traceable answers: Reviewers can verify each claim against its source before relying on it. Auditors, regulators and appeal panels receive a clear evidence trail instead of a block of unexplained model text.
• Model choice: Dedoctive works with whichever large language model you prefer. Teams can switch providers as prices or policies change without rebuilding their knowledge models or redrawing their workflows.
• Sovereign hosting: On-premises and private cloud deployments keep sensitive documents inside your own network. Optional automated redaction and Cyber Essentials certification add further protection for public bodies and regulated firms.
• Open standards: The platform supports BPMN, MCP, A2A, ADK, JSON and REST rather than proprietary formats. This lowers lock-in risk and makes it easier to connect agents and tools you already run.
• Low-risk start: Free micro-pilots let you test the platform on one real, document-heavy process. You see results on your own material before committing budget, which suits cautious public-sector buyers.

Cons

The biggest drawback is the lack of public pricing, which makes it hard to budget for Dedoctive or compare it with other tools before speaking to sales.

• No public prices: Enterprise costs are only available on request, and directories list no figures either. Procurement teams must go through a sales conversation and usually a pilot before they can estimate total cost.
• Curation effort: Answers are only as good as the knowledge model behind them, so documents must be indexed and checked first. Organisations with messy or scattered records will need preparation time before the system pays off.
• Technical setup: The Developer Edition needs Docker, API keys and some familiarity with BPMN modelling. Business users without technical help will find the designer harder to approach than a simple chat assistant.
• Restrictive free licence: The free Developer Edition allows only individual, non-commercial use. Team sharing, commercial services and reverse engineering are forbidden, so any business deployment needs a paid enterprise agreement.
• Small public footprint: Few independent reviews or named customers exist beyond the Ministry of Defence case study. Buyers have limited outside evidence on performance, so a pilot remains the main way to judge results.

Who is Using Dedoctive?

Dedoctive is aimed mainly at regulated and public-sector teams that need decisions they can defend, along with developers building evidence-grounded agent workflows.

• Defence and safety engineers: Engineering teams use the Safety Navigator to extract safety arguments and produce hyperlinked GSN diagrams. The Ministry of Defence has published a case study on validating complex systems this way.
• Planning and emergency teams: Place Navigator, Crisis Navigator and RouteMAIT target these users. The tools check planning applications against local rules or support disaster routing with reasoning that stays traceable.
• Compliance and risk officers: Compliance staff run audits and build bowtie risk diagrams from policies and regulations. Provenance links let them show which clause supports each finding.
• Healthcare reviewers: Clinical teams are the audience for the Healthcare Navigator, which summarises patient records and analyses implications for diagnosis or treatment. Answers stay tied to the underlying record and guidance.

Pricing

Dedoctive is quote-based for enterprise use, with a free micro-pilot and a free non-commercial Developer Edition (official site, October 2026).

Toolify lists no prices, while There’s An AI For That labels it freemium from $0 a year, which reflects only the free entry points.

Disclaimer: Quotes depend on scope and hosting, so ask Dedoctive for current terms.

What Makes Dedoctive Unique?

Dedoctive’s provenance links reach down to a single table cell or part of an image, which few AI assistants attempt. A GSN safety case or compliance report produced by the platform can therefore be checked click by click.

The neuro-symbolic design also sets it apart from typical retrieval tools. A curated knowledge model sits between your documents and the language model, and BPMN control keeps the AI on a mapped route. Auditors see both the evidence and the sequence of steps, and the system stays silent when evidence is missing.

Compatibilities and Integrations

Dedoctive connects through open standards rather than a marketplace of app connectors, so integration relies on APIs, MCP tools and agent protocols.

• Deployment: Enterprise cloud, on-premises or private cloud, with optional automated redaction.
• Language models: Designed to wrap any LLM provider; the Developer Edition takes API keys for the models you choose.
• Agent protocols: MCP, A2A and ADK are supported, and workflows can call MCP tools directly.
• Data and APIs: JSON and REST interfaces, with ingestion of documents, data tables and images.
• Developer tools: The Developer Edition runs in Docker, with Windows start-up scripts supplied.

Dedoctive Tutorials

The quickest way to try Dedoctive is the free Developer Edition, which runs locally in Docker.

1. Install Docker Desktop on your machine.
2. Download the Developer Edition from the GitHub link on dedoctive.ai.
3. Add your email address and LLM API keys to the settings file.
4. Pull the image and run the supplied start-up script.
5. Open DedoctiveDesigner and build a flow from UserTasks, AgentTasks and ScriptTasks.
6. Follow the official YouTube tutorial series, then request a micro-pilot for real data.

How We Rated It

Dedoctive earns 3.6 out of 5 from us, with strong marks for traceability and privacy held back by setup effort and unpublished prices.

Summary

Dedoctive suits regulated organisations, defence suppliers and public bodies that need AI answers they can defend in an audit. Its provenance links and BPMN workflows make each step visible, which generic chat assistants do not offer. The main limitation is opacity on price, together with a small public track record beyond the Ministry of Defence case study. Smaller teams may also find the setup heavier than they need. Start with a free micro-pilot on one evidence-heavy process, and bring a clean document set so the knowledge model has good material to work from.