Keep the work together
Bring the relevant files, instructions, tasks, and outputs into one project.
Give AI the whole job.
Amplio is a desktop app where AI does real work across your files and tools. It returns the presentation, document, or spreadsheet the job needs, with the model you choose.
Windows coming soon
+50Free to start · Your own keys, free · Pay as you goQ3 market research
Message Amplio…
3 filesClaude Fable
Amazon
Apple
Intercom
Raycast
PostHogA good answer is only the beginning.
Real work has sources, steps, calculations, decisions, reviews, and deliverables. Amplio keeps them together so the agent can move from your request to finished work without making you rebuild the context in every conversation.
Use the tools the job requires
Work with connected apps, the browser, local files, scripts, and specialist skills.
Finish with something useful
Create the presentation, document, spreadsheet, analysis, or external action the job needs.
Ask for the result, not the AI feature.
Describe what you need in plain language. The app coordinates the context, tools, steps, and outputs while you review the decisions that matter.
Describe the outcome
Ask for the completed work in the language you already use.
Add what it needs
Bring files, project context, connected apps, examples, and instructions.
Review the work
See what the agent is doing, inspect sources, and guide important decisions.
Receive the result
Open the finished artifacts, continue the work, or make the process reusable.
The output is not another chat bubble.
Create and refine the formats your work already depends on.
pptx
Presentations
Turn research, analysis, and recommendations into a structured deck ready for review.
docx
Documents
Create briefs, reports, proposals, policies, and supporting narratives with sources attached.
xlsx
Spreadsheets
Build models, calculations, reconciliations, trackers, and tables with auditable logic.
Actions
Update the CRM, file the tickets, draft the emails, and send the follow-ups. Every step approved by you.
+ moreUse the model that fits the work.
Bring your own provider keys or connect a managed provider. Switch models per task without rebuilding how your team works.
50+ model providers
Use supported models from Anthropic, OpenAI, Google, Moonshot, and others.
Local-first desktop
Work on your Mac with your local files and direct provider connections. Windows is coming soon.
Self-hosted or managed
Enterprise teams can choose their infrastructure and inference path.
100+ modelsBalanced for cost, quality, and speed
Your best work should not stay in one person's chat.
Share the skills, plugins, connections, and instructions behind good work. Teammates can ask for the outcome in plain language and use the approved setup without rebuilding it themselves.
Share once
Package reusable capabilities for a person, team, or organization.
Start ready
Give teammates the right tools and context without a setup guide.
Different work. The same path to a finished result.
Consulting and professional services
Turn evidence, methodology, and calculations into recommendations, decks, and workbooks.
Example: a market study across 6 research files becomes an 18 slide recommendation with the workbook behind it.Marketing and communications
Move from brief and research to campaign assets, channel variants, and content calendars.
Example: one campaign brief becomes channel variants, a calendar, and the copy deck for review.Finance and accounting
Analyze, reconcile, explain exceptions, and produce controlled reporting outputs.
Example: 1,482 vendor lines reconciled, 9 exceptions explained, one reporting pack produced.Human resources and operations
Coordinate policies, people, documents, approvals, and repeatable operational work.
Example: a policy update becomes the revised document, the approval list, and the announcement.Legal and compliance
Review source material, apply approved guidance, flag decisions, and create review-ready documents.
Example: a 14,000 word agreement reviewed against your playbook, with the open decisions flagged for you.Every team follows the same path: describe the outcome, add what it needs, review the work, receive the result.
Keep the conversation. Add everything the job requires.
Typical AI chatAmplio
Produces an answerProduces reviewable work and artifacts
Starts from the current promptWorks with selected project context and sources
Leaves execution to youUses approved tools, scripts, and connections
Repeats setup across peopleShares reusable skills and plugins
Ties the experience to one providerLets you choose supported models and providers
Make AI work the way your company works.
Give teams approved models, connections, skills, and reusable processes while you control identity, deployment, access, and the infrastructure behind them.
Standardize the work
Turn trusted methods, templates, and controls into capabilities teams can run.
Control access
Manage members, teams, providers, models, and assigned capabilities from one place.
Choose the deployment
Use a managed rollout or deploy in your own environment with your inference path.
We build it with you
Our engineers sit with your team, learn how the work really runs, and turn one real workflow into a skill your team owns.
First workflow live in 2–4 weeksYour deployment. Your provider. Your data boundaries.
Desktop mode keeps files and task history on your machine and sends prompts directly to the model provider you choose. Self-hosted enterprise deployments keep the control plane and storage inside your environment.
Read about security and data handlingWhere the work runs
DesktopFiles and task history local · prompts to your provider
Self-hosted enterpriseControl plane and storage in your environment
Start free. Pay for what you use.
Start free with included models and your own keys. Pay only for the tokens you use. No subscription.
How billing works: usage is measured per token at each model's fixed rate and billed monthly. Your own keys are billed by the provider directly.
Free
$0No card requiredThe full app. Use the included free models, or bring your own provider keys and pay providers directly.
Included free models · limited timeDownload for MacPay as you go
From $0.40Per 1M tokens · no subscriptionEvery popular model through one Amplio account. A fixed price per model, billed only for what you use.
+50Download for MacEnterprise
CustomAnnual agreementEverything in Pay as you go, plus the controls your company requires.
- Single sign-on and identity controls
- Private deployment in your environment
- Your inference path and provider agreements
- Admin controls for members, models, and capabilities
Questions
What is Amplio?+
Amplio is a desktop app for doing work with AI agents across your files, tools, and connected services. It helps you move from a request to finished artifacts and actions while choosing the models and infrastructure you use.
How is this different from Claude Cowork or ChatGPT for Work?+
Those products give you one vendor's model inside that vendor's environment. Amplio is model-neutral: bring your own provider keys and switch between Anthropic, OpenAI, Google, Moonshot, and others per task. It also runs as a desktop app working directly on your local files, and teams can share the skills, plugins, and connections behind good work so a setup isn't rebuilt person by person.
How is this different from a normal AI chat?+
A normal chat is often optimized for answers. Amplio is designed for work that also needs context, tools, multiple steps, review, and finished outputs such as presentations, documents, and spreadsheets.
Do I need to know how agents, plugins, or MCP work?+
No. Describe the result you need in plain language. Technical users can configure models and capabilities, while teammates can use approved setups without understanding how they were assembled.
Which AI models can I use?+
The app supports models available through OpenCode, including providers such as Anthropic, OpenAI, Google, and Moonshot. You can bring your own provider keys or use an available managed option.
Do my files leave my computer?+
In desktop mode, files and local task history stay on your machine. Prompts and selected context are sent to the model provider you choose when a task requires inference. Connected services and optional hosted workers follow their respective data paths. See the Trust page for details.
Can my team reuse the same setup?+
Yes. Teams can share approved skills, plugins, connections, and configuration so people can start with the capabilities they need. Exact sharing and editing controls depend on the plan and deployment.
Can it create presentations, documents, and spreadsheets?+
Yes, when the relevant built-in capability is enabled. The app can create and revise these artifacts as part of a task.
Can I start on my own?+
Yes. Download the individual app and use your preferred model provider. Team and enterprise capabilities follow the current pricing page and deployment terms.
Can we self-host it?+
Self-hosted enterprise deployment is available according to the current enterprise offering. Contact the team to discuss infrastructure, identity, inference, and rollout requirements.
Give AI the work. Get back something useful.
Start locally for free, choose your model, and take a real task from request to result.
Windows coming soon · Bring your own key · No subscription