Alma by Olivares.AIv21Version updatev21May 1, 2026Alma v21 — Background Agents, a Notion-style Workspace, and Every Source You Already Use
Alma is no longer just a chat. Background agents now run multi-step research, monitor sources, process batches and ship scheduled reports — while you do something else.
A new Notion-style workspace lands with nested pages, rich blocks, slash commands and embedded databases you can view as tables, boards, calendars or galleries.
Connect Google Drive, Gmail, Notion, iCloud, Fastmail, Nextcloud, IMAP and any REST endpoint, and Alma builds memory from the documents that matter to you. Magic-link and Apple sign-in join the front door. The whole product wears a refined, calmer look.
Background agents — work that runs while you do not
Four autonomous workflows: Research (multi-step web research with citations), Monitor (watch sources, ping when something changes), Process (long-running batch transformations), Report (scheduled deliverables like weekly digests).
Brief any agent in plain English; Alma classifies the intent and shows a transparent cost estimate (median + P95) before anything runs.
Agents share your monthly budget — no per-agent subscription, no separate credit currency.
Predictable lifecycle: queued, running, paused, waiting on user, waiting on budget, completed, failed, cancelled — you always know where a job stands.
A critique step on Max double-checks every report before it lands in your inbox, and revises automatically when something is off.
Agents read your memory layer, so they pick up your context, your tone and your priorities without re-briefing every time.
A Notion-style workspace inside Alma
Rich block editor: paragraphs, headings, bullet and task lists, toggles, callouts, code blocks, quotes, images, files, tables and dividers.
Slash commands and drag-and-drop reordering.
Nested pages up to three levels deep, sidebar tree, breadcrumbs, instant search across titles and body.
Embedded databases with table, board, calendar, gallery and list views — filter, sort and group without writing a query.
Memory blocks and agent output blocks live inside the same pages — what Alma knows and what Alma produces share a single canvas.
Existing notes migrate automatically. Nothing to redo.
Every source you already use
WebDAV providers: iCloud Drive, Fastmail Files, Nextcloud, Box, Apache, any standards-compliant server.
IMAP for any email provider.
Custom REST endpoints when your data lives in a private API.
Email-to-Alma forwarding alias: send anything from any inbox to your personal alias and Alma ingests it (signature checks and sender allowlist enforced).
Per-source agent templates — Drive, Gmail and Notion each get a hand-tuned briefing so retrieval is calibrated for that source.
Sign in your way
Magic-link sign-in is the new default — one-time link by email, no password.
Apple Sign-In joins Google and GitHub.
Password remains optional: enable or disable from Settings → Security at any time.
Two-factor (TOTP) and recovery codes apply to every method.
Voice that actually talks back
Conversational voice mode with natural turn-taking, interruption-aware, low-latency.
Pause and resume the mic without losing the session.
Mobile haptics on key interactions where the device supports them.
Battery-aware: voice degrades gracefully in low-power mode.
Notes & Calendar in the same workspace
View, create and reschedule events from chat or directly in the calendar grid.
Google Calendar two-way sync with conflict resolution.
CalDAV support for iCloud, Fastmail, Nextcloud, GroupWise.
ICS subscription for read-only feeds (public calendars, sports schedules, shared work calendars).
Per-calendar colour overrides and a connector health badge.
Smarter, faster memory retrieval
Hybrid memory search blends semantic and keyword rankings with a rank-fusion algorithm — better matches on long, conversational queries.
Procedures (saved how-to steps) gained their own semantic index — they surface even when the wording does not match.
Embedding layer hardened with retry, dimension health checks and a transparent fallback path.
Smart context caching expanded across chat, extraction and embeddings — repeat context is virtually free on follow-up turns.
Studios — image, video and music polish
Image Studio: upscale, in-paint canvas, bulk generation up to 8 variants, ETA bar.
Video Studio: backend ETA, 6 starter templates, batch queue, drag-and-drop scene stitching, YouTube preview modal.
Music Studio: studio-grade tracks up to 10 minutes with vocals and lyrics, all on the unified monthly budget.
Favorites across all studios — keep your best generations one click away.
Better organisation across the workspace
Folders nest up to three levels, with smart criteria, archive and reorder.
Bulk actions on conversations: move, archive, delete, tag.
Default folder per environment.
Pinned and starred items move to the top automatically.
Bring your own providers — BYOK on Max
Seven providers supported: Anthropic, OpenAI, Replicate, Leonardo, ElevenLabs, Google AI, Azure OpenAI.
Health dashboard with reachability and per-provider usage trends.
Calls billed to your own provider account — they do not consume your Alma monthly budget.
Encryption rotation built in.
A refined look across the product
Dark mode rebuilt around a calm, neutral palette — Alma orange remains the single accent.
Light mode keeps the warm ivory you know, with neutral text and tighter contrast for long sessions.
Softer corners on buttons and cards.
Marketing pages — landing, pricing, agents, philosophy, FAQ — rewritten to match.
A smoother mobile experience
Safe-area handling improved on iPhone notch and Dynamic Island devices.
Offline state, visibility changes and slow networks handled gracefully.
Modal sheets and overlays respect device chrome.
Faster initial paint thanks to lazy-loaded language packs and route-level code splitting.
International checkout
Polar Software, Inc. is now our Merchant of Record — VAT, sales tax and FX handled for you.
Local currency at checkout: USD, EUR, GBP, CAD, CHF, AUD, JPY, NOK, SEK, SAR, QAR.
Card and digital wallet support.
Customer portal for subscription, invoices and top-ups without a support ticket.
Onboarding & Skills
Six-step onboarding captures your role, your style and your goals — turned into a starter Soul block automatically.
Skills sheet shows what each skill does, what tools it adds, and which plan unlocks it.
Skills preview their tool effect before you switch.
Import what you already have
ChatGPT, Claude and Gemini exports with live preview before commit.
Streaming progress while large imports run in the background.
Automatic deduplication against your existing memory store.
ZIP archive support for bulk imports across multiple sources.
Hardened against abuse
CAPTCHA on forgot-password, reset-password and magic-link requests.
Per-email and per-IP rate limits on every authentication endpoint.
Single-use, short-lived magic-link tokens — no replay, no leakage by URL sharing.
Same anti-enumeration guarantees on every auth surface.
Pricing
Starter $14 · Pro $29 · Max $99 — one monthly budget per plan, no add-ons, no surprises. Max unlocks BYOK, native web search and the Critic step on agent reports.
Try it: https://alma.olivares.ai
Overview
One subscription. One monthly budget. No add-ons.What Alma actually is:
One memory layer across every surface — chat, the four studios, voice, agents, workspace, and your IDE.
A workspace where pages, databases, memories and agent outputs live on the same canvas.
Background agents — Research, Monitor, Process, Report — sharing your memory and your monthly budget.
A configurable identity so her tone, expertise and boundaries match yours.
Your own provider keys on Max — calls billed to your account, never the Alma budget.
This is Alma. AI that works the way you do — across everything.
Supported features
Key Features
- Persistent Memory (memories, Episodes, Procedures)
- Soul Engine (12+ Identity Blocks)
- Automatic Knowledge Extraction From Conversations
- Mcp Server (21 Tools, 9 Resources)
- Full Rest Api (200+ Endpoints)
- Vscode Extension
- Voice Input (stt) And Text-to-speech (tts)
- Ai Image Generation (flux Pro + Leonardo Ai)
- Ocument Generation (docx, Xlsx, Pptx, Md, Txt, Html, Svg)
- Byok Support (anthropic, Replicate, Leonardo Api Keys)
- Hybrid Search (keyword + Semantic Embeddings)
- 9 Ai Agent Tools (web Search, Deep Research, Canvas, Documents)
- 15-language Localization
- Aes-256-gcm Encryption + Totp 2fa + Oauth
- Full Data Export/import (.alma Format)
- Conversation Folders And Tags
- Context Environments (work/personal Separation)
Releases
Alma is no longer just a chat. Background agents now run multi-step research, monitor sources, process batches and ship scheduled reports — while you do something else.
A new Notion-style workspace lands with nested pages, rich blocks, slash commands and embedded databases you can view as tables, boards, calendars or galleries.
Connect Google Drive, Gmail, Notion, iCloud, Fastmail, Nextcloud, IMAP and any REST endpoint, and Alma builds memory from the documents that matter to you. Magic-link and Apple sign-in join the front door. The whole product wears a refined, calmer look.
Background agents — work that runs while you do not
Four autonomous workflows: Research (multi-step web research with citations), Monitor (watch sources, ping when something changes), Process (long-running batch transformations), Report (scheduled deliverables like weekly digests).
Brief any agent in plain English; Alma classifies the intent and shows a transparent cost estimate (median + P95) before anything runs.
Agents share your monthly budget — no per-agent subscription, no separate credit currency.
Predictable lifecycle: queued, running, paused, waiting on user, waiting on budget, completed, failed, cancelled — you always know where a job stands.
A critique step on Max double-checks every report before it lands in your inbox, and revises automatically when something is off.
Agents read your memory layer, so they pick up your context, your tone and your priorities without re-briefing every time.
A Notion-style workspace inside Alma
Rich block editor: paragraphs, headings, bullet and task lists, toggles, callouts, code blocks, quotes, images, files, tables and dividers.
Slash commands and drag-and-drop reordering.
Nested pages up to three levels deep, sidebar tree, breadcrumbs, instant search across titles and body.
Embedded databases with table, board, calendar, gallery and list views — filter, sort and group without writing a query.
Memory blocks and agent output blocks live inside the same pages — what Alma knows and what Alma produces share a single canvas.
Existing notes migrate automatically. Nothing to redo.
Every source you already use
WebDAV providers: iCloud Drive, Fastmail Files, Nextcloud, Box, Apache, any standards-compliant server.
IMAP for any email provider.
Custom REST endpoints when your data lives in a private API.
Email-to-Alma forwarding alias: send anything from any inbox to your personal alias and Alma ingests it (signature checks and sender allowlist enforced).
Per-source agent templates — Drive, Gmail and Notion each get a hand-tuned briefing so retrieval is calibrated for that source.
Sign in your way
Magic-link sign-in is the new default — one-time link by email, no password.
Apple Sign-In joins Google and GitHub.
Password remains optional: enable or disable from Settings → Security at any time.
Two-factor (TOTP) and recovery codes apply to every method.
Voice that actually talks back
Conversational voice mode with natural turn-taking, interruption-aware, low-latency.
Pause and resume the mic without losing the session.
Mobile haptics on key interactions where the device supports them.
Battery-aware: voice degrades gracefully in low-power mode.
Notes & Calendar in the same workspace
View, create and reschedule events from chat or directly in the calendar grid.
Google Calendar two-way sync with conflict resolution.
CalDAV support for iCloud, Fastmail, Nextcloud, GroupWise.
ICS subscription for read-only feeds (public calendars, sports schedules, shared work calendars).
Per-calendar colour overrides and a connector health badge.
Smarter, faster memory retrieval
Hybrid memory search blends semantic and keyword rankings with a rank-fusion algorithm — better matches on long, conversational queries.
Procedures (saved how-to steps) gained their own semantic index — they surface even when the wording does not match.
Embedding layer hardened with retry, dimension health checks and a transparent fallback path.
Smart context caching expanded across chat, extraction and embeddings — repeat context is virtually free on follow-up turns.
Studios — image, video and music polish
Image Studio: upscale, in-paint canvas, bulk generation up to 8 variants, ETA bar.
Video Studio: backend ETA, 6 starter templates, batch queue, drag-and-drop scene stitching, YouTube preview modal.
Music Studio: studio-grade tracks up to 10 minutes with vocals and lyrics, all on the unified monthly budget.
Favorites across all studios — keep your best generations one click away.
Better organisation across the workspace
Folders nest up to three levels, with smart criteria, archive and reorder.
Bulk actions on conversations: move, archive, delete, tag.
Default folder per environment.
Pinned and starred items move to the top automatically.
Bring your own providers — BYOK on Max
Seven providers supported: Anthropic, OpenAI, Replicate, Leonardo, ElevenLabs, Google AI, Azure OpenAI.
Health dashboard with reachability and per-provider usage trends.
Calls billed to your own provider account — they do not consume your Alma monthly budget.
Encryption rotation built in.
A refined look across the product
Dark mode rebuilt around a calm, neutral palette — Alma orange remains the single accent.
Light mode keeps the warm ivory you know, with neutral text and tighter contrast for long sessions.
Softer corners on buttons and cards.
Marketing pages — landing, pricing, agents, philosophy, FAQ — rewritten to match.
A smoother mobile experience
Safe-area handling improved on iPhone notch and Dynamic Island devices.
Offline state, visibility changes and slow networks handled gracefully.
Modal sheets and overlays respect device chrome.
Faster initial paint thanks to lazy-loaded language packs and route-level code splitting.
International checkout
Polar Software, Inc. is now our Merchant of Record — VAT, sales tax and FX handled for you.
Local currency at checkout: USD, EUR, GBP, CAD, CHF, AUD, JPY, NOK, SEK, SAR, QAR.
Card and digital wallet support.
Customer portal for subscription, invoices and top-ups without a support ticket.
Onboarding & Skills
Six-step onboarding captures your role, your style and your goals — turned into a starter Soul block automatically.
Skills sheet shows what each skill does, what tools it adds, and which plan unlocks it.
Skills preview their tool effect before you switch.
Import what you already have
ChatGPT, Claude and Gemini exports with live preview before commit.
Streaming progress while large imports run in the background.
Automatic deduplication against your existing memory store.
ZIP archive support for bulk imports across multiple sources.
Hardened against abuse
CAPTCHA on forgot-password, reset-password and magic-link requests.
Per-email and per-IP rate limits on every authentication endpoint.
Single-use, short-lived magic-link tokens — no replay, no leakage by URL sharing.
Same anti-enumeration guarantees on every auth surface.
Pricing
Starter $14 · Pro $29 · Max $99 — one monthly budget per plan, no add-ons, no surprises. Max unlocks BYOK, native web search and the Critic step on agent reports.
Try it: https://alma.olivares.ai
Top alternatives
-
Persistent memory infrastructure for AI agents with Real Time Analyticsmanlike9🙏 6 karmaMar 31, 2026@OctopodaBrilliant system and feels fairly easy to use, one issue with sign in took me 5 mins, but i think its fixed now? Not perfect, but a step in the correct direction!
-
AI memory layer for smarter agents.Memoripy is a GAMECHANGER. I see why it has so many stars on GitHub- we use it for all of our agents to ensure they are context aware! Great job 👏
-
Give any AI tool persistent organizational context.With the AI memory of aiqbee I can build any type of AI context (a Brain) and use it with any LLM model to produce context specific results instead of guess work. I can also combine different brains or use brains from a marketplace to enrich the output quality.
-
Cure AI amnesia: Remember everything you tell it.I'm using MemoryPlugin since a year. It is brilliant how can it improve the work with LLM-s. With any LLM-s. Asad did a lot of improvement, he is open for discussion about enhancements.
-
Shared memory for all your AI tools.
-
Waitlist
Shared memory for AI agents and teams.I'm in early beta and I'm managing to orchestrate my 3 openclaw agents with my 2 claude code sessions. It's cool, maybe this is the future.


