For operations teams, agencies, and solopreneur founders

Your AI team,
running every morning.

Walli is the managed AI agent platform where every agent knows your business, runs scheduled workflows, and tells you what got done — without API key juggling, AWS bills, or weeks of integration work.

Public launch coming soon. Get notified the moment trials open — no application form, no founders' cohort.

Sarah's EA agent · awake
create a workflow that summarises my email at the end of every day
I'll set it up — daily 11:55 PM, "agent_outputs" as input and "summary" as the markdown output. Open the builder to review and save.
📝
Open workflow builder
Daily Email Summary · 55 23 * * *
5 min
to first deliverable
No keys
Walli Spark — managed inference, no API keys to juggle
Recurring
scheduled workflows that fire on cron
Memory
that learns your preferences over time
What makes Walli different

Three things every agent on Walli does that nobody else's does.

Hosted AI is a commodity. The wrapper around it isn't. Walli ships the wrapper that turns "smart chat" into "your AI team running real recurring work."

⟪⟫

Walli-aware agents

Every agent knows it's inside the Walli portal. It speaks your vocabulary — workflows, schedules, agents, runs — and emits one-click CTA cards in chat. Ask "how do I create a workflow?" and the agent opens the builder for you, prefilled. Not a feature you have to configure. Default behavior.

Fleet, not Agent

Mission Control gives you a single view of every agent in your tenant: what's running, what just completed, what's queued. Schedules in standard cron. Workflows that chain. Reasons to keep one agent that built up six months of memory rather than spawn a new one for every task.

$

Pricing you can see

No application form. No founding-X program. No "request a demo" gate. Pick a tier, start a trial, see your first deliverable in five minutes. The whole pricing matrix is on this page, including the per-token math, the multipliers, and the cached margin we keep.

Day in the life

Three agents. Three roles. One quiet team.

Walli ships pre-built workflows for the EA, support, and ops jobs that should run without anyone touching them. Connect Google (Gmail, Calendar, Drive) or Slack; the workflows below come live in minutes.

SK
Sarah's EA
Executive assistant agent
0 8 * * 1-5 · weekdays @ 8:00 AM

Daily Meeting Brief

Walli reads Sarah's calendar at 8 AM, pulls notes from prior meetings, drafts prep snippets for each slot, and emails the brief before her first standup. She lands on agenda — never on inbox triage.

Output: markdown email, ~600 words, 4-6 meeting blocks
MR
Marc's Inbox
Email triage agent
30 6 * * 1-5 · weekdays @ 6:30 AM

Email Triage

Categorizes overnight emails into four actionable buckets — reply now, schedule, FYI, archive — and proposes a draft reply for the high-stakes ones. By the time Marc opens his inbox, the triage is done.

Output: 4-bucket summary + 2-3 ready-to-send drafts
AT
Alex's EOD
End-of-day summary agent
55 17 * * 1-5 · weekdays @ 5:55 PM

EOD Summary

Recaps today's meetings, captures decisions and follow-ups, and previews tomorrow's priorities. Alex closes the laptop at 5:55 with a one-page summary already in his inbox — and one in his Slack DMs.

Output: markdown summary + Slack DM, ~400 words
Integrations

Connect the tools your team already uses.

OAuth, bot tokens, or API keys — Walli agents connect to your stack in a few clicks. No configuration files. No webhook debugging.

📧
Gmail
Google OAuth
Live
📅
Google Calendar
Google OAuth
Live
💾
Google Drive
Google OAuth
Live
💬
Slack
Bot OAuth v2
Live
📬
Outlook
Microsoft OAuth
Soon
🔔
Discord
Bot token + webhook
Live
✈️
Telegram
Bot token
Live
🔀
n8n
API key · workflow automation
Live
🐙
GitHub
PAT · catalog
Live
📝
Notion
API key · catalog
Live
🎯
HubSpot
API key · catalog
Live
🏃
Jira
API key · catalog
Live
Mission Control

One view. Every agent. Every schedule.

Cron expressions, last-run results, queued work, token usage — all on one screen. Your fleet, not a single chat window.

Mission Control · Acme Operations

Agents Workflows Runs Schedules
Sarah's EA
Executive assistant
Marc's Inbox
Email triage
Alex's EOD
EOD summary
Pipeline Bot
Sales digest
Research Helper
Daily news digest
Support Triage
Manual only
Onboarding Bot
New-customer flow
+ Add agent
7 of 9 used

You shouldn't have to apply
to use an AI agent platform.

Walli is built for operators who want to ship work this week — not impressionable enterprise buyers chasing a "founding cohort" that turns into a sales motion. Pick a plan. Start a trial. Cancel anytime. We mean it.

Self-serve at launch. No application form. No founders' cohort. Pick a plan and go.
Pricing on this page. Per-tier, per-token, per-multiplier. Math you can verify.
BYOK on Growth+. Use your own Anthropic / OpenAI keys if you'd rather pay your provider directly.
Cancel anytime. Workspace retained 30 days for re-subscription, then deleted.
Open source where it matters. Built on OpenClaw — no proprietary lock-in on the agent runtime itself.
Pricing

Pick a tier. Start a trial. See your first deliverable.

Token quotas are denominated in Haiku-equivalent tokens. Sonnet uses 3× quota per token; Opus 5×. One number to track regardless of which model you pick.

Hobby

$29/mo

Personal experimentation, single agent.

  • 1 Walli Spark agent
  • 10M Haiku-eq tokens / mo
  • Claude Haiku 4.5
  • 5 GB workspace storage
  • EA starter pack (Daily Brief, Email Triage, EOD Summary)
  • 3-day trial, credit card required
Get on the waitlist

Growth

$199/mo

Agencies + ops teams running an AI fleet.

  • 2 Walli Spark agents + 7 BYOK agents
  • 200M Haiku-eq tokens / mo
  • Haiku, Sonnet, Opus selector
  • 40 GB workspace storage
  • SSO via Google Workspace + Microsoft Entra
Get on the waitlist

Business

$399/mo

Departments + serious automation programs.

  • 5 Walli Spark agents + 15 BYOK agents
  • 1B Haiku-eq tokens / mo
  • All models + priority routing
  • 100 GB workspace storage
  • SSO via Okta + SAML + OIDC
Get on the waitlist

Need more? Token packs 25M for $49 · 100M for $179 · 500M for $799. Extra Spark agent $25/mo (with 7.5M tokens). Extra BYOK agent $15/mo. Need a custom contract? sales@walli-ai.ai.

Security & isolation

Tenant isolation across seven layers. Verified, not asserted.

Walli runs every tenant in an isolated ECS service with its own EFS access point, KMS encryption, dedicated ENI, and per-agent IAM. Each layer is independently enforced — not just policy-documented.

Layer 1 · Compute
ECS task per tenant, cgroups + namespaces
Layer 2 · Network
awsvpc per-task ENI, no cross-tenant visibility
Layer 3 · Storage
EFS access points, tenant-scoped uid/gid
Layer 4 · Database
DynamoDB partitioned by TENANT#id
Layer 5 · Secrets
Per-agent Secrets Manager, KMS-encrypted
Layer 6 · Identity
Cognito JWT + tenant claims, MFA, step-up
Layer 7 · Application
HMAC defense on internal Lambda invokes
Audit
Every ops action logged; 90d retention
Frequently asked

Direct answers to the questions buyers actually ask.

Do I bring my own Anthropic / OpenAI API key?
Either way works. Walli Spark is the default — managed Bedrock inference, you pick Haiku / Sonnet / Opus per agent and pay one Walli tier price. BYOK is available on Growth+ if you'd rather route inference through your own provider account. You can mix per-agent: some agents on Spark, some BYOK.
How is "Walli-aware" actually different from a normal LLM chat?
Standard LLMs respond with prose. Walli agents emit structured action blocks at the end of replies — those render as one-click CTA cards under the chat bubble in the portal. Tell the agent "create a workflow that summarizes my email daily" and the response comes with an "Open workflow builder" button that opens the builder pre-filled. Plus RAG over our help docs, plus persistent memory, plus platform vocabulary baked into every reply. Stuff you'd otherwise be configuring tools for.
What integrations are available?
Google suite (OAuth): Gmail, Calendar, Drive — connected once, usable by any workflow. Slack (bot OAuth): agents can send messages, read channels, and post deliverables direct to DMs or channels. Discord & Telegram: bot token channel routing — agents receive commands and deliver outputs wherever your team lives. Microsoft Outlook (mail + calendar) is live in limited beta.

Catalog integrations (API key or PAT): GitHub, Notion, HubSpot, Jira — connected per-agent in 30 seconds. n8n: tenant-wide workflow automation bridge, API key config.

All integration credentials are encrypted at rest with your tenant's KMS key and never leave your AWS region.
How fast is the trial setup?
Sign up → pick an occupation → optionally connect Gmail, Calendar, or Slack → run a pre-built workflow. Five minutes to first deliverable. The first agent cold-start is ~70 seconds because of an ECS task launch; subsequent runs are instant. The trial runs 3 days; a credit card is required up front and converts automatically at the end of day 3 unless you cancel.
What about team members — can I invite my team?
Yes. Multi-user is supported with four roles — Owner, Admin, Editor, Viewer — and per-action authorization. Growth tier includes SSO via Google Workspace and Microsoft Entra; Business tier adds Okta, SAML, and OIDC. Invite by email, accept on first sign-in. Seat limits scale with the tier; extra seats are $19/mo on Growth+.
What models are available?
Walli Spark gives you Anthropic Claude Haiku 4.5 (default), Sonnet 4.6 (Starter+), and Opus 4.7 (Growth+) on AWS Bedrock with cross-region inference profiles. Quota is denominated in Haiku-equivalent tokens with 1.0× / 3.0× / 5.0× multipliers for the three models. BYOK supports Anthropic, OpenAI, Google direct.
What happens when I hit my monthly token quota?
Token Meter widget warns at 70%, 90%, and 100%. At 100%, agent responses pause and you're prompted to either upgrade your tier or buy a one-time token pack (25M for $49, 100M for $179, 500M for $799). In-flight runs complete; we don't kill mid-task.
How is Walli different from self-hosting OpenClaw?
Self-hosting OpenClaw gives you an agent runtime. Walli gives you the agent runtime PLUS: managed Bedrock inference (or BYOK), conversational workflow builder, recurring workflow scheduler, mission control across multiple agents, EA starter-pack workflows out of the box, Walli-aware action protocol, RAG over help docs, persistent memory with Titan v2 retrieval, multi-channel routing, auto-recovery + circuit breaker, ops console, audit log, MFA, and capability flags. You can absolutely self-host. Most teams shouldn't.
Where does my workflow data live?
Walli runs in AWS us-east-1. Workflow definitions live in DynamoDB (tenant-partitioned, KMS-encrypted, point-in-time-recovery enabled, 35-day window). Conversation history and deliverable bodies live on per-tenant EFS access points (encrypted, daily AWS Backup snapshots). Long-term memory chunks — embedded snippets the agent uses for cross-session recall — live in a per-agent SQLite database on the same EFS, scoped strictly to your tenant.

Gmail and Calendar are read fresh per-run; Walli does not maintain a separate cache of inbox or event data. Anything the agent references during a conversation may show up in that agent's memory index for future recall (also tenant-scoped). All data uses your tenant's KMS key.
Can I cancel anytime?
Yes. Cancellation takes effect at the end of your current billing cycle; you keep your tier until then. Workspace data is retained 30 days post-cancellation for re-subscription, then deleted. Self-serve data export is on the post-launch roadmap; in the interim, contact support@walli-ai.ai to request a workspace export.

Stop pretending your AI is your team.
Make it your team.

Public launch is around the corner. Get on the waitlist and we'll notify you the moment trials open — no application form, no founders' cohort, no sales call.

Get on the waitlist