User
Manual.

Install

WINDOWS

Two builds on GitHub Releases: HARBOUR-AI-1.3.13-Setup.exe for a normal install (Start Menu entry, auto-updates), or HARBOUR-AI-1.3.13-Portable.exe to run with no installation — from a USB stick or a locked-down machine. Then install Ollama from ollama.com and pull models:

Both Windows builds keep your data in C:\Users\<name>\.harbour-ai\, so the portable build shares its database with an installed copy on the same machine.

LINUX

Install Ollama first, then run the AppImage or deb:

AppImage won't launch: sudo apt install libfuse2

DOCKER / SERVER DEPLOYMENT

HARBOUR AI also runs headless on a VPS, private cloud, or on-premise server — your whole team accesses it via browser. The Docker image ships a multi-stage build, health checks, resource limits, an nginx reverse proxy with TLS termination, and a persistent /data volume for the database, models, uploads, and RAG index.

Server deployment is provided to licence holders. Email Gregorymoores@proton.me and we’ll send the deployment bundle (docker-compose.yml, DOCKER.md, and your licence-keyed .env template) plus setup guidance.

First Launch

An Onboarding Wizard walks you through use case → model setup → quick-start tips. After that, choose:

• Start Free 14-Day Trial — one click, no payment, one per machine
• Enter Licence Key — paste your HARBOUR-XXXX-XXXX-XXXX-XXXX key

The first registered user becomes admin.

The Agents

Toggle agents on/off in the left panel. Add custom agents with Create Agent, or install more from the Marketplace — both under ☰ WORKSPACE → Agents & Automation.

All agents know your name and today's date automatically — no manual context needed.

The Shell & Appearance

Everything HARBOUR does is reachable three ways: the header bar, two menus, and the command palette. There is no long toolbar — the old scrolling row of 84 buttons was replaced in v1.2.0.

THE HEADER BAR

CTRL+K — THE COMMAND PALETTE

Press Ctrl+K (⌘K on Mac) anywhere and type. It searches every menu entry by name and by keyword — "gdpr" finds the DPIA generator, "churn" finds Client Sentiment. ↑↓ to move, ↵ to open, esc to close.

The palette and the two menus are generated from one registry, so they cannot drift apart.

☰ WORKSPACE

⚙ SYSTEM

Menu entries carry live state: a green dot means the feature is on, a number is a live count. Entries you have no licence or role for are not shown at all.

APPEARANCE

🌙 NIGHT / ☀ DAY — night is the default dark cockpit; day is a full light theme where accents darken for light panels. Danger, warning and positive colours stay red / amber / green in both, so a risk banner never changes meaning with the theme.

Palette — Neutral (default), Ocean, Teal, Amber, Rose, Violet. Accent family only; layout and safety colours are untouched. If you used one of the old 16 named themes, your choice is migrated automatically, and night + neutral is pixel-for-pixel identical to the old default.

UI scale — ⚙ SYSTEM → Display → larger / smaller / reset (80%–160%). Saved per machine.

Core Features

• Multi-agent chat — active agents respond in parallel with streaming output
• EMMA Orchestrator — toggle on for a synthesised final verdict after all agents respond. EMMA has full context when giving her verdict: your persistent memories, Company Knowledge Base, user profile, and today's date are all injected automatically.
• Debate Mode — agents challenge each other before EMMA delivers a verdict (⚔ marks debate messages)
• EMMA Agentic Mode — ReAct loop up to 20 steps (default; configurable up to 30) with autonomous tool use. Enable with the AGENTIC button. Tools: run commands, read/write files, scrape URLs, web search (SearXNG + DuckDuckGo), create folders, system stats, list directory, plugins, MCP — all restricted to safe directories. Recommended model: llama3.1 or better.
• EMMA Voice Commander — say "Hey EMMA" (or your custom wake phrase). Pulsing pink dot when listening. Change the phrase in EMMA panel → WAKE PHRASE
• Web Search — enable per-agent via the WEB toggle. Uses DuckDuckGo by default. For fully private search: docker run -d -p 8080:8080 searxng/searxng
• Document RAG — upload PDF, DOCX, TXT, or CSV per session. Agents cite sources, stored locally
• Vision — webcam or screen capture analysed by llava. Enable VISION in the header
• Voice transcription (dictation) — click 🎤 beside the message box to dictate, again to stop. Transcribed on-device by faster-whisper (offline, multilingual); no audio or text leaves the machine. Model size under ⚙ SYSTEM → Models → Model & Providers
• Spoken replies — the header 🔊 toggle reads responses aloud with your operating system voice. HARBOUR's own neural voice (Kokoro) is used for Live Voice and Audio Overviews
• Tool Calling — enable TOOLS. Agents can read/write files, run safe commands, scrape URLs, call plugins
• Chat Export — ↓ MD for Markdown, ↓ PDF for print dialog
• Global Hotkey — Ctrl+Shift+H from anywhere opens the EMMA Quick Ask overlay

Attaching Documents

Two ways to put a document in front of an agent, both doing the same thing:

• 📎 in the send bar — pick a file.
• Drag it straight onto the chat panel — the panel turns green and shows DROP TO ATTACH while you are holding a file over it. Let go and it attaches.

Supported: PDF, DOCX, TXT, MD, CSV, XLSX, JSON, HTML. The file is parsed on your machine and its text is put in front of your next message as context. A green badge above the input shows the filename and character count; ✕ removes it before sending. The attachment clears after each send — re-attach to use the same document again.

One file at a time. Drop several at once and HARBOUR attaches the first and tells you exactly which ones it ignored. It will not quietly swallow the rest.

Integrations

⚙ SYSTEM → Connections → Integrations. Three tabs:

EMAIL / OUTLOOK

Connect any IMAP mailbox. Enter host, port, username, app password → SAVE EMAIL CONFIG → FETCH to load inbox. Everything processed locally.

• Outlook: IMAP imap.outlook.com:993 · SMTP smtp.office365.com:465
• Gmail: IMAP imap.gmail.com:993 · SMTP smtp.gmail.com:465 (use an App Password)

MICROSOFT TEAMS

Paste an incoming webhook URL from your Teams channel (Channel → Connectors → Incoming Webhook) → SAVE TEAMS CONFIG. Agents can then post summaries and alerts directly to your channel.

CALENDAR

Paste an ICS URL (Google Calendar, Outlook, Apple iCal) → SAVE & LOAD TODAY. EMMA fetches today's events for your morning briefing.

Email Ghost-Writing

EMMA learns how you write emails — your tone, vocabulary, greeting style, sign-off, sentence length — and drafts new emails that sound like you wrote them. Open ✉ MAIL → 👻 GHOST.

Step 1 — Train EMMA on your voice

Paste 2–5 emails you have actually written into the training box. Separate each email with --- on its own line. Click 🧠 TRAIN EMMA. EMMA analyses your samples and saves a style profile — this only needs to be done once.

Step 2 — Draft an email

• Topic (required) — what the email is about, e.g. "chasing an overdue invoice — second reminder"
• Recipient (optional) — who you're writing to, e.g. "a client named James"
• Extra context (optional) — any background detail that shapes the tone
• Reply-to email (optional) — paste the email you're responding to

Click 👻 GHOST-WRITE EMAIL. EMMA writes the complete email — subject line, greeting, body, sign-off — matching your style.

Managing your profile

Your style profile is saved permanently. Click RETRAIN to replace it with new samples (useful if your style has changed or you want to train for a different context). The style summary is visible at the top of the draft form so you can verify EMMA understood your voice correctly.

Morning Briefing

EMMA panel → ☀ MORNING BRIEFING → toggle ON → set time → SAVE. Each morning EMMA pulls together today's calendar, unread emails, and a key priority. The briefing appears as a pink banner on open. Dismiss with ✕.

Click ▶ RUN NOW for an immediate briefing at any time.

Meeting Pre-Brief

Fifteen minutes before each calendar event, EMMA automatically reads your Company Knowledge Base, Company Brain, and memories for anything relevant to that meeting — then delivers a sharp, contextual brief as an Always-On alert. EMMA panel → 📅 MEETING PRE-BRIEF → toggle ON.

Setup

Requires a calendar ICS URL connected under 🔌 INTEGRATIONS → Calendar. EMMA checks every 5 minutes and fires a brief for each event starting within the next 5–20 minutes.

What the brief covers

• Who is attending and what the meeting is about (from calendar title and description)
• Relevant KB entries — client files, company notes, documents that match the event
• Relevant Company Brain entities — people, projects, or companies referenced in your Brain
• Your own memories related to the topic
• 2–3 specific things to be prepared for or raise

Manual trigger

Click ▶ RUN NOW to immediately generate a brief for any event in the next 2 hours — useful before an unplanned call or to test your setup.

Briefs appear as purple alerts in the 🤖 ALWAYS-ON alerts tab. Each event is only briefed once per session to avoid duplicates.

Compliance Autopilot

Click 🛡 AUTOPILOT (turns green). Select a framework:

• UK GDPR
• FCA Consumer Duty
• KCSIE 2023
• CQC Fundamental Standards
• UK Employment Law

After each agent response, HARBOUR AI scans it async. If an issue is found, a coloured banner shows the risk level (LOW / MEDIUM / HIGH) and specific findings. Zero impact on response speed.

Scheduled Reports

⏰ REPORTS → + NEW REPORT. Set agent, prompt, schedule (daily / weekly / monthly), and optional email recipient. Click ▶ RUN NOW for an immediate one-off. Results saved and viewable inline.

Conversation Routing

🔀 AUTO (turns purple) routes each message to the best agent automatically. Built-in rules:

Custom rules: click ROUTE → add keyword + target agent. Custom rules run first.

Company Knowledge Base

🏛 KB → upload PDF, DOCX, TXT, or Markdown. All agents reference it automatically in every conversation. Upload and delete require Power User or Admin role.

Agent Memory

🧠 MEM → add facts agents recall in every conversation. Examples: "I'm a solicitor specialising in employment law", "Always reply in formal British English". On Business licences, scope facts to PERSONAL or TEAM.

• Auto-Memory — HARBOUR AI extracts key facts from your conversations automatically every 10 messages and saves them. No manual action needed.
• 💾 Save to Memory — click the 💾 button on any agent response to save it as a memory fact instantly.
• /remember — type /remember <fact> in the chat input to save anything directly to memory.

All three methods write to the same store and take effect on the next message.

Sector Packs

📋 TMPLT → Sector Packs — installs templates, prompts, and UK law memory facts in one click:

Email Assistant

✉ MAIL → paste the email → select a task → ▶ RUN:

• ANALYSE — tone, intent, key points, risk flags, urgency
• DRAFT REPLY — ready-to-send UK English reply
• SUMMARISE — 3–5 bullet summary
• COMPLIANCE — review against UK GDPR, FCA, Employment law, ACAS

Click ↓ TXT to save the result.

Data Analysis

📊 DATA → upload CSV or XLSX → optional question → ▶ ANALYSE DATA. Returns insights, trends, anomalies, data quality issues, and recommendations. Fully offline.

Contract Comparison & Compliance

⚖ ANALYSE:

• CONTRACT COMPARE — upload two documents (A and B) → COMPARE DOCUMENTS. Returns differences, unique clauses, risk flags, and recommendation
• COMPLIANCE CHECK — upload a document, select a framework → RUN COMPLIANCE CHECK. Returns risk rating (HIGH / MEDIUM / LOW / COMPLIANT) plus specific gaps

Frameworks: UK GDPR · ICO Children's Code · FCA Consumer Duty · UK Employment Law · KCSIE 2023 · CQC Fundamental Standards. Both tools support ↓ MD export.

Document Generation

📄 DOCS → select a template → describe the parties and situation → GENERATE & DOWNLOAD DOCX.

Built-in templates: NDA · Letter Before Action · Employment Offer Letter · Client Engagement Letter · Employee Written Warning.

Meeting Assistant

🎙 MEETING → START RECORDING → hold your meeting → STOP RECORDING. Transcribed locally via Whisper. EMMA produces: attendees, key points, decisions, action items (owner + deadline), follow-ups. Export via ↓ MD.

Post-Meeting Auto-Actions

When a meeting recording ends, EMMA automatically extracts every action item and follow-up email from the transcript — no manual review needed. ☰ WORKSPACE → Meetings → Meeting Actions; the menu entry shows a live count of pending actions.

What gets extracted

• Action items — specific tasks with assigned owner and deadline where mentioned
• Follow-up emails — identified emails that need to be sent, with suggested recipient and subject

Managing actions

Click ✅ ACTIONS to open the panel. Each action shows: task, meeting title, owner, and deadline. Click ✓ DONE to dismiss it once complete.

Ghost-writing follow-up emails

Email-type actions have a 👻 DRAFT FOLLOW-UP EMAIL button. EMMA writes the complete email using your M1 voice profile if trained, or professional UK English if not. Copy to clipboard or download as TXT.

Manual extraction

You can also paste any transcript into POST /meeting-actions/extract via the API to extract actions without going through the recording flow.

Screen Memory

Screen Memory gives EMMA ambient awareness of your workday. Every few minutes, HARBOUR AI captures the title of your active window and stores it locally. EMMA silently includes your recent activity in every conversation — so when you ask for help, she already knows the context.

EMMA panel → 🖥 SCREEN MEMORY → toggle ON. Everything stays in local SQLite. Nothing ever leaves your machine.

What it captures

• Active window title (e.g. Chrome — Johnson Contract v3 — Google Drive)
• No screenshots, no keylogging, no clipboard — window title only by default
• Configurable capture interval (default 5 minutes)
• Configurable retention (default 30 days — older entries auto-deleted)

Using Screen Memory

Click VIEW next to the toggle to open the recall panel. From here you can:

• ASK EMMA — natural language: "what was I working on Tuesday afternoon?"
• 🔍 Search — keyword search across all captured activity
• CLEAR ALL — wipe the entire history instantly

Privacy

Screen Memory is disabled by default and must be explicitly enabled. HARBOUR AI filters out its own window from captures. The full activity log is visible at any time — there are no hidden captures. This is the privacy-first version of what Microsoft Recall attempted: local-only, opt-in, fully auditable.

Workflow Builder

⚡ FLOW → + NEW → add steps (agent + prompt, use {{input}} for previous output) → SAVE. To run: select the workflow, enter starting text, click ▶ RUN WORKFLOW. Export as MD.

OCR — Document Scanner

🔍 OCR → upload an image or use the webcam. HARBOUR AI preprocesses the image and runs Tesseract OCR locally. Edit the extracted text, then click SEND TO EMMA ▶.

Requires Tesseract:

• Linux: sudo apt install tesseract-ocr
• Windows: download from the UB-Mannheim installer page, then restart HARBOUR AI

Session Review Queue

⚙ ADMIN → SESSIONS — assign any session to an admin or power user. Assignees view their queue via 📋 REVIEW and read the full conversation inline.

Browser Extension

Install from browser_extension/ in the repo (Chrome / Edge, Manifest V3):

1. Open chrome://extensions, enable Developer Mode
2. Click Load Unpacked, select browser_extension/

Select text on any page → ⚓ Ask HARBOUR button appears. Right-click for context menu options. Set auth token and default agent in the popup ⚙ SETTINGS tab.

Mobile Companion

1. Find your machine's local IP (ip addr on Linux, ipconfig on Windows)
2. Install Expo Go on your phone
3. From mobile/: npm install && npx expo start
4. Scan the QR code, enter http://192.168.1.x:8000, sign in

Features: browse sessions, read history, send messages to EMMA with live streaming. 2FA fully supported. Phone and HARBOUR AI machine must be on the same Wi-Fi network (or backend reachable via VPS).

Real-Time Collaborative Sessions

Click the 👥 icon next to a session name → share the session ID with a colleague → they join via the same icon on their instance. All messages broadcast in real time. Requires Business licence.

MCP (Model Context Protocol)

⚙ Settings → MCP Servers → add a server:

Tools auto-inject into EMMA at connection time.

AI Provider

Defaults to Ollama. To connect LM Studio, vLLM, or any OpenAI-compatible API: ⚙ SYSTEM → Models → Model & Providers → AI PROVIDER → set type, base URL, optional API key → TEST → SAVE. Admin only.

Anthropic Claude — select type Anthropic, enter your API key, click SAVE. Recommended models: claude-sonnet-4-6 (fast, balanced), claude-opus-4-7 (most capable).

Per-agent provider routing — each agent can use a different provider independently. ⚙ SYSTEM → Models → Model & Providers → PER-AGENT PROVIDER → click any agent to expand → set type, API key, optional model override → SAVE. Classic setup: EMMA on Claude API, the four squad agents on local Ollama. Set to Global to revert an agent to the global provider. Changes take effect immediately.

Image Generation

Connect a local AUTOMATIC1111 instance (./webui.sh --api --nowebui), then ⚙ SYSTEM → Models → Model & Providers → 🖼 IMAGE GENERATION → set URL → TEST → SAVE → enable ON. Ask naturally — "generate an image of...", "/imagine ...". Renders inline in chat.

Two-Factor Authentication

🔐 2FA → ENABLE 2FA → scan QR code in your authenticator app → enter 6-digit code → ACTIVATE 2FA. On login, a second screen prompts for the current TOTP code. To disable, enter a valid code and click DISABLE 2FA.

Team Mode (Business — £999)

Shared memories, shared prompt library (TEAM badge), session review queue, and 👥 TEAM panel (member activity, 7-day usage histogram, per-member favourite agents). Power User role can manage the Knowledge Base and review assigned sessions. Admins promote/demote via ⚙ ADMIN → Users.

Enterprise SSO / Active Directory

Requires Enterprise licence. Configure via ⚙ ADMIN → SSO:

• LDAP — direct Active Directory bind. Test connection before saving
• SAML 2.0 — supports Entra ID, Okta, Google Workspace. Download SP metadata for IdP setup. XMLDSig-verified

Users auto-provisioned on first SSO login.

Admin Dashboard

⚙ ADMIN → tabs: Users · Sessions · Stats · Analytics · Logs · Audit · PII · Branding · Reseller · SSO · Model Cache · Quotas.

• Audit — every interaction logged (timestamp, user, agent, content hash, IP address, tokens). Export as CSV. Click VERIFY CHAIN to confirm the entire log has not been tampered with.
• PII — toggle PII Auto-Redaction on/off; view stats and the full redaction event history.
• Quotas — set per-user daily and monthly message caps. When a cap is reached the user gets a clear limit message instead of a response.
• Branding — app name, tagline, logo, primary colour, agent name overrides
• Reseller — enter HARBOUR-RSL key, generate HARBOUR-CL client keys, track activations
• Sessions — list all sessions, assign any to a reviewer, bulk export
• Model Cache — shared network Ollama path for team installs (NHS / slow-broadband networks)

Quota Management

Admins can set per-user message limits via ⚙ ADMIN → Quotas.

• Set a Daily Limit and/or Monthly Limit per user
• When a cap is reached the user gets a clear limit message — no silent failure
• Leave blank for no limit; limits can be removed at any time
• Today's and this month's counts shown live in the Quotas table

PII Auto-Redaction

Enable via ⚙ ADMIN → PII → toggle ON. Every incoming chat message is scanned for 11 UK PII types before reaching any AI model. Matched values are replaced with [REDACTED:TYPE].

Every redaction event is logged. The PII admin tab shows total events, total items redacted, breakdown by type, and full event history. Off by default; non-blocking — messages are never dropped on detection failure.

Tamper-Proof Audit Trail

Every logged action is cryptographically chained — any tampering is immediately detectable.

• Each entry stores a SHA-256 hash of prev_hash|timestamp|user_id|action|content_hash
• The first entry anchors to GENESIS — creating an unbroken chain from day one
• Modifying, deleting, or inserting any row breaks the chain at that point

What is logged: every chat interaction · login success and failure (with originating IP) · admin user actions (disable, enable, delete, reset password, change role) · PII setting changes.

Verify the chain: go to ⚙ ADMIN → Audit → VERIFY CHAIN. HARBOUR AI re-walks the full log from GENESIS and either confirms all entries are intact or reports the exact entry ID where the chain breaks.

Export the full log as CSV at any time for offline archiving or ICO evidence packs.

HARBOUR AI Box

A pre-configured hardware appliance for organisations needing zero-setup, air-gapped AI with no IT involvement.

• Self-registration is disabled — the admin pre-provisions all user accounts
• Activated by setting HARBOUR_APPLIANCE_MODE=true in the environment
• All HARBOUR AI features remain fully available — no capability is removed

→ Full specs and ordering

Referral Programme

Earn 20% commission per sale. Go to harbour-ai.co.uk/refer, enter your licence key, get your personal REF-XXXXXX link.

• £29.80 per Solo sale (20% of £149)
• £199.80 per Business sale (20% of £999)

Self-Test — Prove It Works on This Machine

⚙ SYSTEM → Diagnostics → Self-test. It runs as soon as you open it.

Every check exercises the feature end to end and asserts on the result — it indexes a real document and then retrieves a sentinel only that document contains, asks the model a real question and reads the reply, encrypts a secret and decrypts it back. A check that cannot fail when the feature is broken is not a check.

Each row is ✓ pass, ! note, ✕ fail or – skipped. Failures come with a suggested fix, and every row has its own re-run link.

⬇ SUPPORT BUNDLE collects logs, versions, the results grid and any failed requests the browser saw into one JSON file, with email addresses, tokens and file paths stripped. It is saved to your machine — nothing is uploaded.

THE HEALTH STRIP AND THE PROBLEMS TRAY

Health strip — a bar across the top, visible only when something needs attention: amber for a note, red for a real problem, each item with its fix beside it. Re-checks every 30 seconds.

Problems tray — a red ⚠ n problems pill in the bottom-left corner, and only when there are any. Every failed backend request is recorded with time, endpoint, status and detail, by an interceptor around every request — so a failure cannot be hidden by a swallowed error. COPY puts the list on your clipboard for a support email.

If a button ever appears to do nothing, open the tray — the reason is in there.

The Egress Meter — "0 bytes sent externally"

Bottom-left of the cockpit is a live counter of everything this session has sent to any host that is not local. When it reads 🔒 0 B sent externally, that is measured, not asserted.

Click it for the detail: total bytes, a per-host breakdown, and a kill-switch — refuse all external requests — that makes HARBOUR reject any outbound call to a non-local host outright. Your local model, RAG and anything on your own LAN never count towards the meter.

This is the number that matters when someone asks you to prove the privacy claim. It is also the reason three call-homes buried inside bundled libraries were found and closed in v1.3.5–v1.3.7: a document embedder contacting huggingface.co on every start, telemetry inside the vector database, and the speech model phoning home on every spoken word. Code review had found none of them. The meter did.

DATA RESIDENCY ATTESTATION

⚙ SYSTEM → Governance → Data residency attestation downloads a signed JSON document recording what this install has sent externally, signed with your install's own Trust Root key. Hand it to a client, an auditor or a DPO: it is verifiable without HARBOUR, and it says something a policy document cannot — not "we promise nothing leaves", but "here is the measured egress, signed".

Privacy & GDPR

All data stored locally in ~/.harbour-ai/ (Linux) or C:\Users\<name>\.harbour-ai\ (Windows). No telemetry, no cloud, no crash reporting. Passwords hashed PBKDF2-SHA256 (260,000 iterations). JWT sessions with 7-day expiry. Delete everything by removing the ~/.harbour-ai/ folder.

Don't take that on trust — measure it. The egress meter in the bottom-left corner counts every byte sent to any non-local host this session, the kill-switch inside it refuses outbound requests entirely, and ⚙ SYSTEM → Governance → Data residency attestation gives you a signed record of the same thing. The self-test's "Nothing has left this machine" check asserts on that meter.

The things that do legitimately reach the internet, and only when you turn them on: web search, cloud AI providers, Companies House lookups, HMRC MTD, Twilio, WhatsApp, Slack/Teams, cloud drive sync, and update checks. Everything else — the model, RAG, embeddings, speech-to-text, speech, OCR, vision — runs on your machine.

Full compliance statement: harbour-ai.co.uk/gdpr

Keyboard Shortcuts

Troubleshooting

Client Sentiment Monitor

EMMA reads every inbound client email daily and scores the relationship — Positive, Neutral, At-Risk, or Churning — using a local language model. Results are stored in the Company Brain. A weekly digest surfaces any At-Risk or Churning relationships so you can act before the client leaves.

Knowledge Decay Alerts

Knowledge Base entries that have not been reviewed or updated for a configurable number of days are flagged as stale. EMMA sends a weekly digest listing every stale entry with its age and the person who created it. This ensures your Company Brain stays accurate as your business evolves.

NHS DSPT Compliance Check

Every AI output in the dashboard can be checked against NHS Data Security and Protection Toolkit standards in real time. Non-compliant outputs are flagged inline before they leave the system. The check runs locally — no data transmitted to any cloud service.

BNF Drug Interaction Checker

The British National Formulary drug interaction database, offline and embedded within EMMA. Enter any combination of drugs and EMMA flags interactions by severity — HIGH, MEDIUM, LOW — with the clinical effect and a plain-English explanation. No data leaves your device.

CQC Self-Assessment

EMMA guides care providers through a structured CQC self-assessment against the five key questions: Safe, Effective, Caring, Responsive, and Well-led. Answers are saved to the Company Brain. EMMA produces a formatted self-assessment report ready for submission or internal governance review.

SRA Conflict of Interest Checker

Before onboarding any new client, run a conflict check across your entire Company Brain — every conversation, document, and relationship EMMA has indexed. Returns CLEAR, POTENTIAL CONFLICT, or CONFLICT, with specific matches and an explanation. Every check is logged with a timestamp for your compliance records.

Court Form Auto-Fill

EMMA fills N1 (money claim), ET1 (employment tribunal), and N244 (application notice) from a plain-English description of the matter. Every required field is mapped automatically. Exports as DOCX ready for review and filing.

MTD VAT Return Assistant

EMMA reads your transactions from connected accounting software, groups them by VAT category, and populates the nine MTD VAT return boxes automatically. Reviews flagged items before submission. Exports the completed return as a CSV for your bridging software or accountant.

Automated Compliance Reports

EMMA generates scheduled compliance reports — weekly, monthly, or quarterly — covering GDPR obligations, regulatory deadlines, outstanding actions, and audit trail summaries. Reports are exported as formatted PDFs and emailed to configured recipients automatically.

Regulatory Change Monitor

EMMA monitors GOV.UK, the ICO, FCA, CQC, SRA, and other configured regulatory bodies for new guidance, consultations, and legislative changes. When a relevant update is detected, EMMA generates a plain-English summary and adds a compliance action item to your dashboard.

Cash Flow Forecasting

EMMA models your cash position week by week for the next 90 days using real outstanding invoices, historical payment patterns, and recurring costs from your connected accounting software. An Always-On alert fires when your projected runway drops below your configured threshold.

Auto Invoice Chasing

EMMA sends polite, escalating payment reminders automatically — polite at due date, firm after 14 days, formal at 30 days. Tone adjusts per client relationship. Stops the moment payment is received. Every chase is logged to the Company Brain.

R&D Tax Credit Assistant

EMMA scans your conversations, commits, and documents for HMRC-qualifying R&D activity — technical uncertainty, systematic investigation, advance in science or technology. Estimates the credit value and drafts the technical narrative for your accountant to submit.

Pricing Intelligence

EMMA monitors competitor pricing pages, job boards, and Companies House filings on a schedule. Hiring patterns reveal revenue direction. Companies House filings reveal financial health. Together they produce a quarterly pricing recommendation with specific evidence.

Slack / Teams Bot

@EMMA lives in your Slack workspace or Microsoft Teams. Team members ask questions, get Company Brain lookups, and run research — all processed on your local HARBOUR AI instance. Slack and Teams only relay the message text. No AI processing happens in their cloud.

Embeddable Chat Widget

Embed EMMA on any website using a single script tag. Visitors chat with a customisable AI assistant backed by your Company Brain. All processing runs on your local HARBOUR AI instance — visitor conversations never touch a third-party AI cloud.

Public API

Expose HARBOUR AI as a developer API. Issue hbr_ prefixed API keys, set monthly request quotas, and receive HMAC-signed webhooks when events fire. As of v1.0.121 the API covers 10 endpoints across chat, Company Brain, agents, workflows, document generation, and data analysis.

Webhooks: register a URL to receive signed events. HARBOUR AI sends HMAC-SHA256 signed POSTs with the X-HARBOUR-Signature header. Events: chat.completed, brain.asked, workflow.completed, doc.generated.

White-Label Engine

License the HARBOUR AI engine to other software companies. They ship "AI by HARBOUR" to their users under their brand. Configure the brand name, logo, colour scheme, and feature set exposed. Revenue share applied per active seat in their deployment.

EMMA Hiring Assistant

Post a role, upload CVs, and EMMA scores every candidate against your job description (0–100) — ranked Strong Yes, Yes, Maybe, or No — with specific strengths, weaknesses, and tailored interview questions per CV. Automated acknowledgement emails sent. Everything logged to the Company Brain.

Company Health Dashboard

EMMA synthesises everything she monitors — email sentiment, cash flow, compliance flags, at-risk clients, regulatory changes, competitor moves — into a single weekly health score (0–100) across five domains: Finance, Compliance, Operations, Clients, and Knowledge.

AI Business Simulation

Ask EMMA "What if we lose our biggest client?" and she models the financial and operational impact using your actual Company Brain data. Revenue and cost change percentages, cash position impact, headcount effect, key risks with likelihood ratings — all from your real business data, entirely offline.

Kiosk Mode

Lock the interface to a role-specific panel set. A nurse sees only DSPT-compliant clinical tools. A receptionist sees only PHONE and RECEPTIONIST panels. Configured by admin per role, applied per user — the employee never sees what is hidden.

EMMA as AI Director

Configure EMMA as a named director of your company. She attends every meeting via transcript, monitors every regulatory deadline, and produces a monthly board pack automatically — P&L narrative, risk register with owners and mitigations, action tracker, KPIs, forward look — exported as DOCX and emailed to the board.

Features Without a Panel — Ask EMMA, or Call the API

Not every capability has a window of its own. Around forty of them are wired straight into EMMA as platform tools, so the way you use them is to ask:

EMMA calls the same endpoint a panel would have called, under your own account — so your licence tier, quotas and audit trail apply exactly as if you had clicked it, and any draft is saved where the matching panel would have put it. These tools only draft, analyse and look up; they never send email, submit to HMRC, or delete anything.

Everything below is also a plain REST endpoint you can call directly — see Public API. The full endpoint list for each feature is in the shipped USER_MANUAL.md.

IR35 Determination Assistant

HARBOUR guides you through 15 HMRC-aligned questions across six assessment areas: Substitution, Control, Mutuality of Obligation, Financial Risk, Integration, and Business on Own Account. EMMA returns a determination — Inside, Outside, or Borderline — with a confidence rating, key factors on both sides, and a recommended action plan. All assessments are saved with full reasoning.

MTD Income Tax Self-Assessment

Enter quarterly self-employment income and allowable expenses. HARBOUR calculates your full tax estimate using 2025–26 rates and saves the submission to the quarterly log.

DSAR Processor

Log an incoming Data Subject Access Request with the subject's name and email and the date received. HARBOUR auto-calculates the 30-day statutory deadline, searches the Company Brain, knowledge base, and inbox for all data held on that individual, and generates a compliant response letter.

Contract Renewal Watchlist

Paste contract text and EMMA extracts: contract name, counterparty, start/end dates, renewal date, notice period, auto-renewal flag, and annual value. Contracts inside the notice window are flagged based on urgency.

HARBOUR Redact

Paste any text to remove personal data in two passes: a fast regex sweep followed by a local LLM PII sweep. All processing stays on your machine.

Ofsted Inspection Prep

Answer seven questions per domain across the four EIF inspection areas. EMMA returns a readiness score out of 10, evidence gaps, and recommended actions. Sessions are saved for comparison.

GP Referral Letter Generator

Select specialty and urgency, enter the clinical summary. EMMA generates a DSPT-compliant NHS referral letter. Patient names are never stored — only specialty, urgency, age, and sex are retained for audit.

Care Plan Generator

Enter the resident/service user ID, care setting, and assessment notes. EMMA generates a person-centred care plan across all five CQC domains. Plans are versioned — each review creates a new linked document.

Charity Sector Pack

Four tools built for the charity sector, all SORP FRS 102 aligned.

HARBOUR Voice Assistant

Transcribe audio locally with Whisper — no audio leaves the machine. Supports WAV, MP3, M4A, OGG, WEBM up to 25 MB. EMMA responds in concise, TTS-ready language.

HARBOUR Translate

37 languages translated by the local LLM — no data sent to Google Translate, DeepL, or any external API. Welsh, Irish Gaelic, and Scottish Gaelic supported.

EMMA Code Reviewer

Paste code and select the language. EMMA reviews across seven categories with severity ratings and an overall score out of 10.

HARBOUR Forms

Build no-code AI-processed forms. Each form gets a public submit URL and an embeddable widget. EMMA processes every submission according to your instructions. Webhooks fire HMAC-SHA256 signed on each submission.

GDPR RoPA Builder

Build and maintain your Article 30 Record of Processing Activities. AI generates processing activity entries from plain-English descriptions. Exports ICO-ready PDF or Excel. Every change is audit-trailed with timestamp and user ID.

Breach Response

AI-guided 72-hour ICO notification workflow. Classify the breach, assess risk to data subjects, auto-draft ICO notification and subject communications, and track remediation actions through to closure.

Pen Test Analyser

Upload penetration test reports in PDF, DOCX, or JSON (Burp Suite, Nessus, OpenVAS, Metasploit, Nuclei, OWASP ZAP, Qualys, Rapid7, Nmap, CrowdStrike). AI extracts findings, maps to OWASP/CVE/CVSS, and generates an executive summary with prioritised remediation plan.

Business Valuation Engine

AI-powered business valuation using four methodologies simultaneously. Input three years of financials; receive a blended valuation range with full workings, comparable transactions, and an investor-ready narrative.

M&A Due Diligence Pack

Structured deal workspace for buy-side and sell-side M&A transactions. AI analyses uploaded documents across seven due diligence workstreams, flags red flags, and populates a live DD tracker with risk ratings.

Accounts Production

Prepare statutory accounts under FRS 102, FRS 105 (micro-entities), or IFRS. Import trial balance from CSV or Open Banking (Section 98); AI applies the correct disclosure framework and drafts all notes automatically.

Competitive Market Intelligence

Continuous AI-powered monitoring of competitors, market trends, and industry signals. Track up to 50 competitors across 10 intelligence dimensions with automated weekly briefings delivered to your inbox or Slack.

Clinical Decision Support

NICE Guidelines-aligned decision support for clinical teams. Enter presenting symptoms, patient demographics, and current medications; AI returns evidence-based assessment pathways, drug interaction flags, red flag alerts, and referral recommendations — all citing the source NICE guideline and publication date.

CRM

Full customer relationship management with AI-powered engagement scoring, opportunity tracking, and pipeline analytics. Contacts, companies, deals, activities, and communications in one place — with EMMA able to update records from meeting notes automatically.

HRIS

Full Human Resource Information System covering the entire UK employment lifecycle. From offer letter to leaver reference — AI drafts all documentation, calculates entitlements, and ensures UK employment law compliance throughout.

Payroll (RTI)

HMRC Real Time Information payroll for 2025-26. Calculate PAYE, NIC, student loan deductions, and pension contributions. Submit Full Payment Submissions (FPS) and Employer Payment Summaries (EPS) to HMRC via the PAYE API.

Project Management

AI-powered project management with RAG health scoring. Gantt, Kanban, and list views. EMMA monitors project health daily and alerts when tasks are at risk — before they become late.

Natural Language Analytics

Query any connected data source in plain English. No SQL required. EMMA translates your question into a query, executes it, and returns a chart or table with a plain-English explanation of the result.

Learning Management System

AI-powered LMS with CPD tracking. Create courses, assign learning paths, track completion, and issue certificates. EMMA can generate course content from your existing documents and knowledge base.

RAG Connectors

Connect external data sources to HARBOUR AI's Retrieval-Augmented Generation engine. EMMA searches across all connected sources when answering questions, citing the source document and relevance score for every retrieved chunk.

Marketplace

Browse, install, and manage third-party extensions, sector packs, prompt libraries, and agent templates. Published by HARBOUR AI and verified community developers. All marketplace items are code-reviewed before listing.

Advanced Regulatory Horizon Scanning

Monitors 50 UK and international regulators in real time. AI classifies each publication by impact, urgency, and applicability to your sector configuration. Daily digest with action-required items flagged at the top.

Enterprise SSO

Configure SAML 2.0 or OIDC federation with your organisation's identity provider. Supports Azure AD, Okta, Google Workspace, ADFS, PingFederate, and OneLogin. Just-In-Time (JIT) user provisioning creates HARBOUR AI accounts automatically on first SSO login.

Open Banking

Connect to 15 UK banks via the Open Banking OBIE / PSD2 framework. Retrieve real-time account balances and transaction history with your explicit consent. AI categorises transactions, analyses cash flow, and flags covenant breaches automatically.

NHS API (FHIR R4)

Integration with NHS Digital APIs using the FHIR R4 standard. Perform PDS patient demographic lookups, raise e-Referral Service requests, issue EPS prescriptions, and retrieve NHS terminology — all within the secure HARBOUR AI environment.

AI Video Interview Platform

End-to-end video interview platform powered by AI. Create interview templates, generate STAR-framework questions, invite candidates with unique secure tokens, collect asynchronous video responses, and receive a structured hire/no-hire scorecard — without scheduling a single call.