01 · Workflow agents
Zapier · n8n · Make
Cheap and reliable, until they aren't.
StrengthPredictable for known, well-defined flows.
LimitRigid. Breaks on the first edge case. No judgement, no memory, no learning.
The platform
How Lifter runs your team's AI in one place.
The specialists, the tools they reach for, the memory they build, the controls you keep — and where it all runs.
01Positioning
The spectrum of agent capability.
AI agents aren't one thing. They sit on a spectrum — from rigid task runners to general assistants to a new third category Lifter is built for.
02 · Personal assistants
Claude · ChatGPT · Copilot
Universal, conversational, broadly capable.
StrengthHelpful on any question, with judgement and reasoning.
LimitPer-person. No continuity, no team context, no governance, no shared knowledge.
03 · Operator agents
Lifter
General-purpose work, with persistent memory and governance.
Built forAgents organisationally embedded, learning over time, with the controls a real company needs.
The pointA new category that operates the business alongside your team.
Less autonomy
More autonomy
02Centralised by design
One platform for every AI capability in your business.
Connectors, skills, memory, agents, approvals, audit — all in one place. Set up once. Distribute across the org. Govern centrally.
Connectors
Managed in one catalog. Connect Shopify, Gmail, Slack, Stripe once. Allocate to the right teams and agents. Revoke instantly.
Skills
Shared library. Pre-built handbooks for common operations. Author your own. Distribute to every relevant agent across the org.
Memory
Organisational, not personal. Agents share institutional knowledge — house style, customer history, decisions made — not one person's chat log.
Governance
Single pane. Access control, permissions, data policy, audit — one place to set the rules and one place to see what happened.
The compounding effect. Every skill written, every connector configured, every lesson learned accrues to the company — not a single inbox.
03The specialists
A handbook, a tool list, a memory.
Three parts your team can read and edit. A specialist is a self-contained unit with its own permissions, its own memory, and a clear way of working — not a black box.
A · Handbook
How it thinks about the work.
A written brief. How to triage a ticket. When to escalate. Your tone of voice. Your refund policy. Authored by retail operators, editable by anyone on your team.
B · Tools
What it can touch.
An allow-list of connectors. Shopify orders. Klaviyo flows. Gorgias tickets. Each tool gets a permission stance: allow, ask, or deny.
C · Memory
What it remembers.
Plain text the specialist reads and edits over time. Products, customers, policies, the tone your last reply used. Versioned. Auditable. Yours.
04Connects to what you already pay for
100+ connectors. The tools your team already uses.
Native packages for the tools we hand-tune to retail: Shopify, Klaviyo, Gorgias, Cin7, Dotapparel. MCP (the open standard for AI tool access) for the rest of the ecosystem. New connectors land every week.
Live now
Shopify
Klaviyo
Gorgias
Cin7
Dotapparel
Slack
Gmail
Stripe
Xero
HubSpot
+ many more
See all integrations →
Shopify
05Workflows
Agent teams & humans, co-owning the same board.
Customer-defined workflows. Customer-defined columns. Typed transitions. Some auto-fire and wake the next agent; some require human approval.
Triage
MAgent
Refund request · order #2487
Replacement enquiry · SKU mismatch
Bulk return · B2B
Verify
TSpecialist team
Refund eligibility check
Inventory hold & restock
Approve
HHuman
$340 refund · awaiting manager
Loyalty credit · $50
Resolve
MAgent
Refund issued · customer notified
Replacement shipped · tracking sent
Closed · resolved by Holly
Built on building blocks already in the runtime. Capability tools, confirmations, scheduling. Workflows aren't a separate product; they're the shape your team gives the platform.
06Dashboards
Dashboards your agent composes.
Ask the agent what matters about your store today. It composes a dashboard from the tools and data it has access to. Persistent, shareable, updated on a schedule.
Revenue · yesterday
$48,210
Orders
421
AOV
$114.51
Refund rate
2.3%
Holly's narrative
Yesterday's lift came from the winter-essentials bundle. Three SKUs accounted for 38% of revenue. Worth checking inventory depth before Friday's email send.
7-day revenue
Different from generic BI. The agent already knows your business (memory) and can act on what it sees (tools). The dashboard is an artifact of the work, not a separate product to maintain.
07Memory
The longer you use Lifter, the more useful it gets.
Plain text your team can audit and edit. Memory belongs to your business — not the model, not us.
i · Remembers
Your business, in writing.
Products. Customers. Tone of voice. Refund policy. Launch process. What worked last quarter and what didn't.
ii · Readable
Your team owns every word.
Memory is plain text, organised the way a person would organise a wiki. Open it up, change what's wrong.
iii · Portable
Survives a model swap.
Switch the underlying model and the specialist keeps everything it learned. Memory belongs to your business.
08Permissions & approvals
Three states. One inbox for the hard calls.
Every tool a specialist can use gets one of three stances. The risky ones route to the channel your team already uses.
Allow
Low-risk, reversible, audited. Runs without human approval. Drafting an email, building a dashboard, summarising a thread.
Ask
Pauses for a human in the channel your team already uses. Default for anything that touches a customer, costs money, or changes inventory.
Deny
The tool isn't available to that specialist at all. They don't see it. They can't try to use it.
#cx-approvals · today
CS
Maya
Order #48201 arrived damaged. Photo on file. Refund $186.50?
Approve
Deny
S
Sam W.
Approved. Add a $20 store credit too.
CS
Maya
Done. Refund processed, store credit issued, customer notified.
09Where it runs
The Gateway runs without the Hub.
Hub is optional. On-prem, customer VPC, air-gapped: pick the deployment that survives your security review.
Hosted
Lifter-managed Hub. Gateway in your cloud.
Multi-tenant operator UI, OAuth flows, connector catalog: all in Hub. Your data stays in your Gateway.
Self-managed
No Hub. Gateway in your VPC. No outbound.
Gateway loads its own config, holds its own credentials, exposes its own tools. Nothing leaves the perimeter.
When a security team asks where the data goes, the answer is: wherever you put the Gateway. We do not require a phone-home.
10Takes your brand
Your company's AI operations centre.
The Hub is fully white-label. Your brand, your colours, your typography. To your team it feels like a tool your company built, because for all intents and purposes, it is.
Default
L
Lifter Hub
Six specialists working on the business today.
The Lifter-default brand. What you'd see if you signed up today without customising anything.
Customer-branded
A
Acme Operations
Four agents handling today's tickets.
The same platform, configured for Acme. Different palette, different typeface, different name. Same engine underneath.
Logo & mark
Your wordmark and app icon throughout.
Brand colour
Your accent applied across the UI.
Typography
Choose from a curated typeface set.
Custom domain
ai.yourco.com instead of ours.
11Extensibility
No lock-in to a model, or an agent.
Today we ship our own purpose-built operational agent. The platform is designed so other AI agents — Claude, Codex, Copilot, and the ones that don't exist yet — can plug into the same controls without rebuilding anything.
Lifter Agent
Claude managed agents
Codex / OpenAI agents
The Lifter platform
One platform, every AI.
Whichever agent runtime you plug in, the controls below stay the same.
Tool access. Every agent sees the same connectors, with the same scopes.
Permissions. Allow / Ask / Deny applies to every agent equally.
Credentials. Held by the Gateway. Never delivered to any model.
Approvals. Same approval queue, regardless of which agent raised it.
Audit trail. One ledger covers every action by every agent.
Copilot-style agents
Open-source agents
Your own internal agent
The model changes; the guarantees don't. Memory, approvals, audit, scheduling — all stay the same when you swap the AI underneath.
Founding cohort, open
Want to see the platform on your stack?
A 30-minute call. We'll show you Lifter running against the integrations your team already uses, and tell you what it would take to deploy.