The private AI assistant for UK SMBs that actually remembers what you said last week — without sending it to OpenAI's training pipeline. One installer. Your machine. Your data. Your Telegram bot.
Private OR remembers. Frontier OR local. Free OR usable. We give you all three.
Same product, three levels of hand-holding. Pick what fits your team's capacity.
All pre-built. Bring your own keys, the installer wires them up.
DM your bot, get a reply with full conversation context — last 20 turns prepended automatically. Includes /forget and /memory commands.
Telegram → ClaudeSend a voice note. Local Whisper transcribes it on your machine, AI extracts a bulleted action list. Audio never leaves your network.
Whisper (local) → ClaudeDrop a PDF in the chat. Get back: overview, key points, dates/deadlines, action items. Handles up to ~12 pages per upload.
PDF → ClaudeEvery 2 hours during weekdays, scans Outlook unread. AI classifies + drafts replies into your Drafts folder. Never sends without you.
Outlook → Claude → DraftsPaste text into a local web form. Get a sector-aware risk report flagging PII, GDPR, SRA/ICAEW/FCA exposure. Decision support, not sign-off.
Form → Claude → HTML reportn8n is a full workflow editor. The bundled credentials are reusable — drag in new nodes, ship internal automations on top of the same stack.
Extend, don't replaceThe bundled system prompts know about your sector's regulators — and what the AI shouldn't claim to do. Pick yours during install. Swap any time.
The installer walks the buyer through every step. No YAML, no Docker commands, no manual config.
Installer checks Windows version, Docker Desktop, disk, internet. Opens the Docker download page if needed and waits.
Asks for your AI key + Telegram bot. If you don't have them, opens the vendor's signup page and waits while you paste back. DPAPI-encrypts everything.
Generates docker-compose, pulls images, brings up n8n + optional Whisper / Ollama, waits for healthcheck. Binds to localhost only.
Imports 5 workflows into n8n, activates the chat bridge, sends a welcome message to your Telegram. You're done.
Your messages → Telegram's infra → your machine → the AI provider (Anthropic or OpenAI). Nothing else. VantagePoint never sees anything — we sold you the installer, full stop. If you pick local Ollama, even the AI provider sees nothing.
Yes for decision support, drafting, and summarisation — treat AI output as a junior assistant's draft, always review. Not appropriate for unsupervised client communication or storing UK GDPR Article 9 special-category data without a DPIA. Enterprise tier includes a Data Processing Schedule add-on drafted by a UK data-protection lawyer.
ChatGPT trains on your conversations by default — not appropriate for client work. ChatGPT Temporary chat doesn't train but forgets every session. AI In Your Pocket gives you trained-by-no-one privacy AND persistent memory. That's the gap.
You're a direct customer of the AI provider — we don't mark up. If they raise rates, you pay them more. Switching providers takes ~5 minutes in n8n. Or use local Ollama and pay nothing per-message.
All sales final. Digital-content waiver under UK Consumer Contracts Regulations 2013 Reg 36 — required to check at checkout. Once you've downloaded the installer, the value is delivered. If something genuinely doesn't fit, email support — we'll typically credit you against a different VantagePoint product.
v1 is Windows 10/11 only — that's where 90% of UK SMB endpoints actually live. macOS and Linux are on the v2 roadmap. If you're a Mac shop and want this badly, email us — we'll fast-track if there's volume.
Yes. v1 runs on your machine. For always-on, the reference guide includes a one-page "move to a £6/mo Hetzner VPS" walkthrough. v2 will offer a managed hosted option.
Free for paid customers. We email you when v2 ships, you download the new installer, it upgrades in place. Memory and n8n data survive the upgrade.
£1,497 to run it yourself. £3,497 if you'd rather we did. Either way, 20 minutes from purchase to live bot.