What changed in Avalon.

Avalon now gives the most space to work that needs a decision. Mail, Flowboard, calendar, Cleanup, Automation, settings, and Avalon AI share the same navigation and interaction patterns.

• Flowboard puts On You first — On Them, Resolved, and No Action stay visible without crowding the work that needs a response.
• Pulse stays in reach — Ranked priorities live in the sidebar. Search remains a separate tool for finding mail, people, and events.
• The whole app fits together — Shared controls, clearer states, stronger keyboard support, and responsive layouts now carry across every main workspace.

Mail, Flowboard, calendar, onboarding, and bulk operations now show what completed, what failed, and what the user can do next.

• Drafts save while you write — Compose keeps the draft on screen and shows an error when autosave fails.
• Bulk work reports the whole result — Partial failures identify completed and unfinished work so the queue can resume safely.
• Old tabs stop quietly going stale — Avalon asks for a reload when a newer deployment is available.

Custom prompts now stay with the mailbox they belong to, across browsers and devices. The chat also makes tool progress, completed work, interruptions, and failures easier to follow.

• Presets follow the mailbox — Save a recurring prompt once and use it wherever you sign in.
• Long chats are easier to navigate — The revised message layout keeps the current task and its result in view.
• Tool failures explain themselves — Interrupted and failed actions leave a clear result with a next step.

Avalon now separates India and international pricing, with the currency and checkout provider shown before payment begins.

• India plans are billed in INR — Razorpay handles subscriptions and its supported local payment methods.
• International plans are billed in USD — Stripe continues to handle international subscriptions.
• Account deletion closes the subscription first — Avalon cancels active billing before deleting the account.

Saved Gmail and Microsoft 365 drafts now open as drafts. Change the recipients, subject, or message, then send or delete the draft without returning to the provider.

• Continue where you stopped — Existing provider drafts open in the compose experience with their content intact.
• Finish the draft in Avalon — Edit, send, or delete it from the Drafts folder.

Avalon now loads the selected conversation first and leaves heavier background work until the mailbox is usable. Flowboard also catches up sooner after a reply is sent.

• The active thread gets priority — Mail content appears before slower counts and background work finish.
• Replies update responsibility — Sending from Avalon refreshes the related conversation and its Flowboard state.
• Calendar weeks start where you expect — Choose Sunday or Monday in calendar settings.

Avalon now explains its protected-data model as layered security: database encryption at rest, per-tenant application encryption for sensitive data, private passphrase unlocks, and tenant-aware KMS key wrapping.

• Private passphrase boundary — Protected mailbox content stays unreadable while locked; unlocked AI automation is called out as the explicit, narrow processing exception.
• Tenant-aware KMS wrapping — Security copy now describes KMS wrapping as an application-layer boundary rather than relying only on database-level logical separation.
• Honest assurance wording — The public security page avoids overstating SOC 2, pentest, or end-to-end encryption claims before external evidence exists.

Mailbox navigation, search, counts, calendar context, and message rendering now behave more consistently across Gmail and Microsoft 365.

• Gmail navigation has better signals — Inbox and draft counts are more accurate, nested labels are easier to use, and search loads faster.
• Outlook threads recover more cleanly — Sent messages, quoted replies, mailbox actions, and meeting context are less likely to fall out of step.
• Cleanup has its own workspace — Archive and unsubscribe tools now live together.

One Avalon account can now hold multiple Gmail and Microsoft 365 mailboxes. Each mailbox keeps its own identity, routes, and protected-data session.

• Add another mailbox without another Avalon user — Linked accounts stay available from the same sign-in.
• Protected content stays with the right mailbox — Unlock state remains scoped to one account.
• Existing mailboxes link safely — Avalon reuses the mailbox record and avoids duplicates.

Avalon is now available for Gmail and Microsoft 365. Flowboard shows who owns the next move, Avalon AI helps with the work, and users can choose where the model runs.

• Flowboard turns mail into responsibility — On You, On Them, Resolved, and No Action show where each conversation stands.
• Avalon AI works from the conversation — Ask questions, find context, prepare a reply, or take a reviewed mailbox action.
• Teams choose the model path — Use Avalon-hosted AI, bring a provider key, connect a private endpoint, or run a local model.

Flowboard now counts the conversations and tasks that need attention, then separates unread work, pending replies, overdue follow-ups, and open tasks.

• Exact counts replace estimates — Each lane reflects the work Avalon has actually found.
• The reason is visible — Users can tell whether an item is unread, waiting, overdue, or still open.

Pulse pulls recent mail, calendar pressure, and Flowboard movement into a short view of what changed and what may need attention next.

• Recent activity stays close — New mail and changing work appear without a separate reporting screen.
• The day has context — Calendar pressure and responsibility state sit beside the active work.

Deadline changes, packed schedules, and movement across Flowboard can now change what Avalon brings forward.

• Deadlines affect priority — A conversation can rise when its due date gets close.
• Schedule pressure is part of the picture — A crowded calendar can change what belongs at the top of the day.

Long email threads now open with the decisions, open questions, people involved, and next steps pulled into a shorter read.

• The useful parts come first — Avalon condenses repeated replies and quoted history into the latest state.
• Open work stays visible — Questions, commitments, and follow-ups appear before the full thread.

Avalon AI can prepare a reply from the current thread, earlier conversations, the recipient, and the user's preferred tone.

• The draft starts with context — Avalon reads beyond the last message before proposing a response.
• The user still sends it — Every generated draft waits for review and approval.

Drafts can now draw from facts, policies, and writing guidance that a user or team has approved.

• Recurring answers stay consistent — Avalon can reuse the same approved facts across later replies.
• Writing guidance carries forward — Saved preferences can shape later replies without another round of setup.

Avalon AI now has one chat for questions about mail, calendars, workflows, and setup. It can also help carry out the mailbox work behind the answer.

• Ask from one place — The chat can work across the mailbox, calendar, and Flowboard.
• Move from answer to action — Avalon AI can prepare the next step after answering the question.

Search by person, date, topic, attachment, decision, or commitment without remembering the exact words in the email.

• Describe what you remember — Avalon searches the meaning of the request along with ordinary mailbox fields.
• Bring old decisions back — Earlier threads can return when they matter to the work happening now.

Ask Avalon AI to draft, label, organize, or archive mail from the same conversation where the work was discussed.

• Instructions become mailbox work — The chat can prepare supported actions from a plain-language request.
• Review stays in the path — Consequential changes remain visible and controlled before they run.

Avalon can pull due dates, promised follow-ups, and time-sensitive requests out of a conversation and place them beside the work.

• Commitments become visible — A promise buried in an older reply can still appear as an open obligation.
• Deadlines reach Flowboard — The date stays beside the work that depends on it.

Avalon can inspect a calendar, find workable openings, and use those times in a scheduling reply.

• Open slots appear together — Users can compare practical meeting windows without checking each day by hand.
• The reply is ready to review — Suggested times can flow into a clear email draft.

Upcoming events can bring together the calendar invite, relevant email, travel details, weather, flights, and reservations in one briefing.

• Related details stay attached — Avalon connects the event to the messages and plans around it.
• Preparation starts from one page — Users can review the useful context without opening each source separately.

Before a meeting, Avalon can gather the purpose, attendees, recent email, earlier decisions, and likely next steps into a short brief.

• Know why the meeting exists — The brief connects the invite to the conversations behind it.
• Carry the last decision forward — Relevant history arrives before the call starts.

When weather matters to a meeting or trip, Avalon can include current conditions and the forecast in the event briefing.

• Conditions appear with the plan — The forecast sits beside the event it may affect.
• Forecast changes stay relevant — Upcoming conditions can shape the preparation before departure.

Avalon can estimate the travel window for a calendar event and include enough buffer for the expected route and conditions.

• Departure time becomes part of the event — The calendar shows when travel needs to begin.
• The estimate follows the plan — Route timing stays connected to the meeting or reservation it supports.

Avalon can identify flight details in travel mail and bring current status information into the trip briefing.

• Flight details come from the conversation — Relevant numbers, dates, and routes can be picked up from the mailbox.
• Status changes stay near the itinerary — The briefing can reflect new flight information when it is available.

Restaurant, hotel, and event confirmations can now follow the calendar entry or trip they belong to.

• Avalon recognizes reservation mail — Confirmation details can be pulled from the message and attached to the plan.
• The current status is easier to find — Users can see whether a reservation is confirmed without searching the inbox again.

Avalon AI can remember useful preferences, frequent collaborators, writing patterns, and recurring ways of working.

• Preferences survive the next conversation — Saved guidance carries into later work.
• Familiar people bring familiar context — Recurring collaborators can shape later drafts and recommendations.

Rules can classify, route, prioritize, and follow up on recurring email patterns without asking the user to repeat each step.

• Triage runs from a known pattern — Repetitive mail can move to the right place automatically.
• Follow-ups keep their state — A rule can help carry an active conversation through its next step.

When a rule handles a message badly, tell Avalon AI what should have happened. The rule can be adjusted from that explanation.

• Corrections use plain language — Users can describe the mistake without rebuilding the rule from scratch.
• The changed behavior stays visible — The repaired rule can be reviewed before it handles more mail.

Users can ask Avalon AI to change supported preferences and feature settings in plain language.

• Configuration starts with the request — Tell Avalon AI what to change in plain language.
• Changes stay explicit — Avalon shows the setting being changed before applying it.

Avalon can flag spam, phishing-like messages, and suspicious senders before they enter focused work queues.

• Sender and message signals work together — Avalon weighs several clues before flagging a message.
• The warning arrives before the action — Users can review a suspicious message before replying or following a link.

Avalon can identify unsolicited sales mail and keep it away from the conversations that belong in Flowboard.

• Common outreach patterns are recognized — Repeated pitch language and sender signals help identify cold email.
• Real work stays easier to scan — Unwanted outreach no longer needs the same manual triage as active conversations.

Review recurring senders, choose what no longer belongs, and unsubscribe or clean up the selected mail together.

• Subscriptions appear by sender — Users can see the sources creating repeated inbox noise.
• One review handles the batch — Cleanup no longer requires opening every newsletter separately.

Avalon supports Model Context Protocol, an open standard for connecting approved tools and data sources to AI applications.

• External context can join the workflow — Avalon AI can use information from a connected system when the task allows it.
• Connections follow one standard — Teams can add supported tools without a separate integration design for every source.

Teams can receive Avalon updates, ask questions, and start supported actions from Slack.

• Updates reach the team workspace — Relevant mailbox and workflow changes can appear where the conversation is already happening.
• Questions can start in Slack — Users can ask for context or begin an Avalon action without opening another tab first.

Avalon can connect to Salesforce through Model Context Protocol and use customer, account, and opportunity context while working with email.

• Customer context sits beside the conversation — The user can see relevant CRM information without searching Salesforce separately.
• Follow-ups start with the account history — Drafts and summaries can use the connected context when the workflow permits it.

Inbox analytics show message volume, response patterns, sender load, unresolved work, and how those signals change over time.

• Volume has a source — Users can see which senders and categories create the most work.
• Unresolved work has a shape — Response time and open conversations show where pressure is building.

Teams can use their own provider key, private endpoint, or local model when they need more control over cost, privacy, or deployment.

• Bring your own provider key — Model usage and billing can stay with the provider the team already uses.
• Keep the model closer — Private and local model paths support stricter deployment and data-handling requirements.