Research
Gemini is best at real-time research, vision, and image generation for on-brand visuals.
Every frontier model · one subscription · GDPR-friendly proxy
One AI workspace.
Private by proxy.
Remembers everything.
Nevel is a focused AI workspace that runs GPT-5.4, Claude Opus 4.7, Gemini 3.1 Pro and Grok 4.1 behind a privacy proxy that anonymises every prompt, plus a long-term memory that actually remembers you and your projects.
Try for free
7-day trial · No credit card · All models included
Every vendor is best at one thing. Nevel knows which.
30% of our customers were paying for three or four AI subscriptions before Nevel — and they were right to.
Different labs are genuinely better at different jobs. Nevel puts them all behind one chat, one bill, and one memory — with auto-routing built in.
Pick the right model.
Or let Nevel pick for you.
Switch models mid-conversation without losing the thread. Router mode picks the best model for each message automatically — reasoning, long context, vision, code.
-
-
Writing
Claude Opus is the most thoughtful, careful writer of the four — for long documents, contracts, sensitive email.
-
Reasoning
GPT-5.4 is the sharpest thinker — for strategy, analysis, architecture, and hard maths.
-
Live info
Grok 4.1 reads the live web with the fewest filters — for news, market moves, opinionated takes.
Not notes. A memory that remembers.
Every other assistant forgets the moment the tab closes. Nevel doesn't. It builds a real, queryable model of you and your projects — who your clients are, what you're working on, what you've already decided, how you like to write — and pulls the right slice into every reply. Automatically.
Organised by project, not by chat
Related threads live under one project. Ask about it three weeks later — Nevel pulls in the right context, not your entire history.
Remembers without being told
Nevel reads every conversation and quietly extracts what matters — people, companies, decisions, deadlines, preferences. No "please remember this" ritual.
You're in control
Open the memory, read it, edit or remove any fact. Plain text you own — not a black box you have to trust.
Asks the right questions back
"What did we decide about Acme's pricing?" — Nevel pulls the exact thread, the exact decision, the exact date. No scrolling, no guessing.
Drop in a .docx. Get back an edited .docx.
Most assistants turn your Word document into Markdown, lose the formatting, and hand you back a draft you have to redo.
Nevel works on the real thing — tables, styles, comments, tracked changes, slide masters — and returns a file your colleagues can open without asking where the headings went.
-
Word (.docx)
Rewrite, translate, redline a contract. Styles, headings, comments and tracked changes preserved.
-
Excel (.xlsx)
Edit values, write formulas, reshape tables, add sheets. Formatting and named ranges survive.
-
PowerPoint (.pptx)
Rewrite slides, swap images, apply a template, generate a full deck from a brief. Masters and layouts intact.
Your data never leaves the building as itself.
Every prompt goes through Nevel's proxy before reaching any AI model.
It strips names, emails, phone numbers, IBANs and custom patterns on the way out — and puts them back on the way in. All in under 40ms.
The model never sees the real data.
Automatic depersonalisation
PII detectors for 14 entity types — people, orgs, emails, phones, addresses, card numbers, IBANs, passports — plus custom regex and dictionary rules per workspace.
-
Reversible, not redacted
Tokens are swapped, not blanked. Replies come back fully personalised because the mapping lives in your workspace — never on the model provider's side.
-
No training, ever
All upstream requests are sent with "no training" flags where providers support them, and the proxy strips identifying headers regardless.
-
Privacy-first regions
Choose where your data is stored. EU residency available by default. Compliance certifications in progress for teams on Advanced.
Share the work, not just the prompts.
Most AI tools are private by default — what one person figures out stays in their tab. Nevel makes context, automations, chats and artifacts shared objects, so the team builds on each other's work instead of starting over.
Shared contexts
Client profiles, brand voice, PRDs. Any teammate's reply is grounded in the same source of truth.
Shared automations
Save a workflow once — weekly report, inbound triage, QBR deck — and the whole team runs it with one click.
Shared chats
Drop a thread into a channel. Teammates pick it up, branch it, continue it. Comments and mentions built in.
Shared artifacts
Documents, decks, spreadsheets, images generated inside Nevel live in a team library — versioned, searchable, reusable.
What people actually use Nevel for.
Not "boost productivity." Concrete things our customers ship every week — from marketing briefs to board decks to 50-page contract reviews.
-
Consultants & analysts
Docs in, decks out
Upload a strategy memo, get a 12-slide keynote outline, speaker notes and an editable Google Slides export.
-
Developers
Pair programming
Connect a repo, get code review, refactors and PR descriptions. Secrets and customer data get stripped on the way out, restored on the way back.
-
Researchers & Analysts
Deep Research
One prompt, ten sources, one report with citations. Drops a clean Markdown brief into your project folder — ready to edit.
-
Designers & Marketers
Brand-consistent images
Generate, edit and iterate on images with the same models you use for text — with your brand palette and product photography as references.
-
Legal & Procurement
Contract review
Drop a 50-page PDF. Get a risk summary, redlines, and a plain-English version. Sensitive clauses stay depersonalised upstream.
-
Support
Agents you can trust
Save a workflow once; run it with a single click. Each agent runs through the same privacy proxy and writes to the same project vault.
The boring parts, done well.
-
Workspaces
Shared history, prompts and agents for your team. SSO (SAML, OIDC), centralised billing, audit logs.
-
BYO keys
Bring your own OpenAI, Anthropic or Azure keys — we handle routing, quotas and failover.
-
Audit logs
Every prompt, every depersonalisation mapping, every model call — exportable to your SIEM.
-
Native mobile
iOS and Android apps with the same memory, the same proxy, and offline drafts.
-
Everywhere
Web, macOS, iOS and Android. Same memory, same proxy, same artifacts — with offline drafts.
-
Integrations
Slack, Gmail, Google Drive, Notion, Linear, GitHub, Figma. More every month.
Three layers, one calm interface.
- 1. You write in Nevel
A single chat workspace in the browser, on macOS, or on iOS. Pick a model, ask anything. Switch models without losing the thread.
- 2. The proxy depersonalises
PII is swapped for tokens before the prompt leaves our infrastructure. The upstream model sees [PERSON_1], not your customer.
- 3. Memory writes it down
Anything worth remembering lands in the project's memory as plain text. Editable, portable, and linked to the right project.
Ultra · invite only
Stop keeping up.
Let us keep up for you.
You're tired of tracking releases, rotating betas, testing which model is best this month, deciding which agent to use for what. Ultra is a concierge tier: we run the frontier for you. New models swapped in the day they ship. New agents configured the moment they're useful. New capabilities turned on automatically. You just talk to Nevel.
- 1. Always the newest
Frontier models the day of release. Private preview access via our lab partnerships. Nothing to configure.
- 2. Maximum automation
Auto-router picks the model. Auto-memory curates the context. Auto-agents trigger the workflow. You approve, not operate.
- 3. A real human, on call
A dedicated AI engineer from our team. 1h response. Quarterly reviews of what's working and what to automate next.
Request invite
Invite-only. We onboard 10 customers per month.
Three plans. One bill for every model.
Seven-day free trial on Pro, no credit card.
Ultra is invite-only.
-
Pro
For individuals getting started.
$20 /month- GPT-5.3, Opus 4.7, Gemini 3.1 Pro, Grok 4.1
- Unlimited messages (fair use)
- Projects and chat history
- Image generation with every vendor
- Web, iOS, Android
-
Most popular Advanced
For power users and small teams.
$60 /month- Everything in Pro
- Privacy proxy with depersonalisation
- .docx / .xlsx / .pptx round-trip editing
- Smart context memory across projects
- Deep Research + agents
- Team sharing · contexts, automations, chats, artifacts
-
Invite only Ultra
For professionals who live in AI.
$200 /month- Everything in Advanced
- Day-zero access to new models and agents
- Maximum automation · auto-router, auto-memory, auto-agents
- Dedicated AI engineer · 1h response
- Quarterly workflow reviews
- White-glove onboarding
Stop switching.
Start shipping.
Seven-day free trial on Pro. No credit card.
The proxy and the memory are on from message one.
Try for free
7-day trial · No credit card · All models included