Talavine is a personal and business data hub. Review how it brings information together, supports answers with original records, and lets you change supported AI models while retaining your data. Use a demo or test profile for published captures.
Fact sheet
The basics, checkable.
What it is
A personal and business data hub that brings chosen emails, documents, notes, and calendars together for smarter answers and useful work such as client briefings and follow-up drafts. Users can inspect the supporting records and change supported AI models.
Platforms
Windows and Apple Silicon macOS. (No Intel Mac or Linux build today.)
Price
Free during the public preview. No paywall on the desktop app.
Channels
Gmail, Microsoft/Outlook, and IMAP email; Apple Messages (iMessage/SMS, macOS); Slack, Telegram, Discord. Works alongside existing mail clients via standard IMAP sync.
AI
Bring your own: local models via Ollama or LM Studio (free), or cloud keys for OpenAI, Anthropic, Gemini, DeepSeek, Grok, and others. Claude Code / Codex subscription sessions work as routes too. Many AI jobs can run on local models; review each workflow's settings, because changing the chat model alone does not disable every cloud connection.
Where data lives
On the user's machine, in per-account local databases. No hosted service holds the mail, and Talavine doesn't use it to train models. Cloud AI providers you configure receive relevant request context and handle it under their own terms.
Family
Part of the Swarm, with Talavine CRM (hosted, for teams), Vibes (plain-English app builder), and SwarmSpan (Mac↔PC clipboard/file utility).
Story angles
Six claims — and how to test each one yourself.
We'd rather you verify than quote us. Each angle below pairs the claim with the fastest way to check it in the shipping app.
1 · Evidence-backed answers carry receipts
The claim: claims grounded in your records link to the supporting email or message, while inference and missing evidence remain visible instead of being presented as sourced fact.
Verify it: ask something only your mail knows ("what did the plumber quote me?"), click the citation, and read the original message. Then ask something outside the record and inspect how the answer labels its evidence limits.
2 · Your saved information stays when you change AI
The claim: you can choose another supported AI while keeping your Talavine data and conversation. The available models differ in capability, cost, and privacy settings; answers and citations can differ too.
Verify it: ask a question, choose another configured model, and ask again. Check that your saved records remain available and compare the answers against those records. For local AI, review the settings for the whole workflow; changing the chat model alone does not disable every cloud connection.
3 · Related information can come from different places
The claim: a useful answer may need a text message as well as an email or document. Talavine brings supported sources together, subject to account access, device permissions, and workplace policies.
Verify it: on a Mac, connect Apple Messages with the required permissions and ask a question that spans a text conversation and an email thread. Open the supporting records and check whether anything is missing.
4 · The agent writes its own tools, then stops needing the model
The claim: for repeated jobs, the AI authors a small, typed C# tool once — then the job runs as deterministic code, and a refresh needs no model call.
Verify it: in Vibes, describe a stock or weather tracker, let it build, then open the app's “Inside” view: read the generated tool's code, its declared parameters, and its last-run record — and refresh the rows without a single model call.
5 · The engineering fails closed, and the docs admit the sharp edges
The claim: work pinned local-only can never fall back to a cloud model (it stops and says so instead); unproven models can't silently take over the mail pipeline (calibration-gated routing); paying per message is off by default; and the docs state plainly that Claude Code subscription routes do not count as "local."
Verify it: read the white paper and routing page, then check the behavior in the app — set a function to local-only with no local model installed and watch it stop instead of silently going to the cloud.
6 · AI helpers turn information into work you can review
The claim: give a helper a clear job, the information it needs, and limits on what it may do. It can prepare work, gather updates, and report what needs your attention. Desktop work requires an awake, connected computer.
Verify it: use a test profile with client emails and meeting notes. Ask a helper to draft a follow-up without sending it. Review the draft, compare it with the records, and check the activity log and pause controls. The client-work example shows a starting request.
On the record: “I started building Talavine because I found the agent gateways everyone was excited about unusable. This is the version I could trust on my own machines.” — Scott Crossen, creator of Talavine. Quote freely.
Hands-on
Connect one account, ask, and inspect the evidence.
First · Install and connect one inboxConnect Gmail, Microsoft, or standard IMAP. Recent mail becomes available first while the rest indexes in the background.
Next · Ask, then click the citationAsk something only your mail knows. The citation click-through is the moment that separates this from a chatbot — you're reading the actual email behind the answer.
Then · Try another supported AIChoose another configured model and ask again. Your saved information stays in Talavine. Compare the answers, check the source records, and review the privacy settings for each connection.
Go deeper when you are readyOn a Mac, connect Apple Messages and ask across texts + email. Build a tracker in Vibes and open its “Inside” view to read the tools it wrote for itself. Or point Claude Code at the built-in MCP server and query your own history from your terminal.
This is a public preview. It's a real, daily-drivable app, but you will find rough edges — and we'd genuinely like to hear about them.
Local AI wants real hardware
The app and indexing run fine on a normal laptop. Local models want RAM and ideally a recent GPU or Apple Silicon; on weaker machines, cloud models carry the assistant.
First index takes time
Years of mail take a while to sync and index. Recent history is askable quickly; the deep archive fills in as the background pass completes.
iMessage is macOS-only
Apple Messages support reads the local Messages database on a Mac and needs Full Disk Access granted once. There is no iMessage path on Windows — Apple doesn't offer one.
Platform coverage
Windows and Apple Silicon Mac today. No Intel Mac or Linux. The iPhone and iPad companion is on the App Store and pairs with the desktop app; Android is coming soon.
Answer quality tracks the model
The grounding and citations are the app's; the prose is the model's. A small local model gives terser, occasionally clumsier answers than a frontier model — the receipts are the same either way.
Boilerplate & assets
Copy, paste, publish.
Boilerplate
Talavine is your personal and business data hub for Windows and Apple Silicon Mac,
free during the public preview. It brings together supported email accounts, documents,
notes, calendars, and selected imported history for answers, client briefings, and
follow-up drafts. AI helpers can collect updates and prepare work. Your data stays yours as you change
supported AI models, with options to export selected knowledge or move a Talavine profile.
The profile is stored locally; enabled cloud AI may receive relevant information during
processing and answering. Talavine CRM, Vibes, and SwarmSpan are separate products with
explicitly supported connections.
More at talavine.com.