Skip to main content

How it works

How Convira works.

Start with a result you want. Convira gathers the context you choose, keeps the full list visible, shows the work as it happens, and saves the finished file on your computer. Six chapters follow: what you see while it works, then the architecture underneath.

01Request to result

You describe the result. Convira handles the steps.

Follow one message from send to saved result. Convira gathers the material, chooses the tools, shows the work as it runs, and leaves you a file you can open straight away.

  1. You ask

    Describe the result and add any files, screenshots, or folders Convira should use. The source material remains visible before you send.

    Q3-report.xlsx · attached
  2. Convira chooses

    Convira reads the request and decides whether it needs a direct answer, a plan for review, or completed work. There is no mode to pick - a permission dial sets how much it asks before acting.

    approach · complete the task
  3. You set the context

    Convira gathers only the sources you selected: attachments, Active History, Project references, editable Memory, and reusable Skills.

    what Convira can see · 42% used
  4. The AI plans

    The AI runs where you chose - Cloud, this computer, or your Private Box - and proposes specific actions for the task.

    AI location · on this computer
  5. Every action is checked

    Permissions and approvals are checked before an action runs, and original files are backed up before Convira's file and document tools modify or remove them.

    update Q3-report.xlsx · backup created
  6. You get the result

    The workspace shows the file changing beside the conversation. Stop or redirect the work, then open the finished result from your computer.

    complete · Q3-client-review.pptx

Visible workspace

The answer is not hidden in a progress spinner.

When Convira touches a document, the conversation stays beside the file. You can see cells, pages, or slides change as each step lands, then open the saved result in the app you already use.

02What Convira can see

You choose what Convira can see for this task.

Dropping files in is the easy part. Knowing the full list is the point: the history and attachments going into the next run are visible, chosen by you, and removable before you send. Memory rides along too - you can read it in the app and edit it in Settings.

Use the right context without opening the whole computer

AttachmentsFiles and folders you drop in - read from your disk for this run.
Active HistoryPrior runs, grouped per runtime, each group yours to include or exclude.
ProjectsDurable sources and instructions that follow every session in a project.
MemoryFacts extracted across sessions. Openable, editable, deletable.
SkillsYour reusable playbooks, surfaced when they match the task.
Connectors1,000+ services the agent can act on once you connect them.

Connected services

1,000+ services under the same permission model.

Connectors extend what the agent can see and do - email threads, CRM records, chat channels. Whatever a connector reads lands in the visible context above, and anything it sends or changes waits for your approval. A slice of the catalog:

03Active History

The same session can continue without carrying everything.

Active History separates continuity from accumulation. Keep the decisions that matter, leave stale attempts behind, and prevent on-device material from quietly crossing into a cloud run.

01One lane per runtime

Cloud, Local, Private - every run sits in the lane of the runtime that produced it, with its exact token count.

02Nothing rides along silently

Toggle a lane out of the next run, or drag a single run between lanes. The model gets exactly what you leave in.

03Crossing to Cloud

Dragging on-device history toward Cloud stops the move mid-air and asks. Keep it on device with one click.

04Compact, don't forget

Old runs fold into a short recap - the gauge drops from 73% to 30% and the facts come with it.

04Saved on your device

Sessions, plans, and file backups live on your device.

Convira makes local storage understandable. See how much space your work uses, choose how long to keep it, and restore an original file that was backed up before a session changed it.

The run record on this device is the authoritative copy. Changing or ending a plan never takes your work away: the encrypted data stays on your disk, readable in the app for 14 days after a subscription ends and fully available again the moment you resubscribe.

05Where the AI runs

Choose where the AI processing happens.

Use powerful hosted models, run AI on this computer, or connect a larger machine you control. Flip through the three choices to see the exact path your request takes.

YOUR DEVICE

Desktop + local model

  • Inference happens here
  • Works with Wi-Fi off, up to 24h at a time
CONVIRA

Managed tool service

  • Web, media, connectors
  • Receives only the tool call's input
ONLY IF INVOKED
SELECTED SERVICE

Tool provider

  • Receives that tool input
  • Returns that tool result
TOOL CALL ONLY

Local inference runs entirely on your computer. The dashed path appears only when you choose to run a managed online tool such as web research, media generation, or a connector.

06What reaches a teammate

A teammate can read it. Our servers cannot.

Team Hub is the one place where content Convira stores lives on our servers, which makes it the sharpest version of this page's question. It is sealed on your device before it is sent: what we hold is ciphertext, and no Convira server has a key that opens it.

01Sealed before it leaves

Messages, attachments, bulletins and even conversation names are encrypted on your device. The keys are made there too, and reach our servers only sealed for your teammates' devices - never in a form we can open.

02A key per teammate's device

The keys come from each scope's own MLS group: every member device holds a leaf in that group and derives the epoch secret from it. They rotate when the group changes, so a device that has left cannot read what comes next.

03Search stays on the device

There is no server-side search to build over ciphertext, so the index is local - and the app says so under every result instead of leaving you to work it out.

04What delivery still needs

Our servers see what routing a message requires: which workspace and conversation, a sequence number, an epoch, a content type, a size, a hash, which account uploaded it, and when.

Deleting a message removes our copy and the copies on your teammates' devices, everywhere the app holds it. To deliver messages, our servers keep which accounts are in a direct message or group, and which of them run it. A new computer joins only after one of that person's own computers approves it, or, if every one of them is lost, after someone holding the workspace's 24 recovery words or a workspace owner or admin approves it, with everyone told. Read the full security model →

07Safety and architecture

Four layers stand between a request and your computer.

Every proposed action is checked against your rules and approvals, and filesystem scope, network policy and native process isolation each add a separate layer.

PROPOSED ACTION
A request naming exactly what it touchesread this file, update this sheet, or open this page - with validated details
YOUR PERMISSIONS
Rules and approvals check the requestallowed folders, websites, and capabilities - with a pause for sensitive actions
CONTAINED WORKSPACE
The host narrows the execution scopeallowed folders, tool permissions, and network rules are each checked before anything runs
OPERATING SYSTEM
Native containment adds another boundarythe operating system's own sandbox, and Convira shows the isolation each action ran under

Want the implementation details? Read the full security model →

Now watch it work on yours.

Start with a file you already know and a result you can judge. Convira will show you the work from the first step to the saved result.

Get Convira Read the security model →