AI OS for Customer Service Teams & AI Assistants
Answer every customer from one shared memory
Ditto connects to any app your support team uses and remembers what each customer told the team. Your team's AI assistants work from that memory, and changes wait for your approval by default.
- Gmail
- Outlook
- Slack
- Teams
- Jira
- Notion
- Stripe
- Shopify
- Any other app
Where customer service teams lose time and customer data privacy
What your support team knows is spread across people and disconnected systems, so agents lose time chasing answers across apps and channels. An AI assistant for customer service needs to reach every app the support team uses.
-
Customers repeat themselves
What a customer told one agent or sales rarely reaches the next agent on the ticket.
-
Answers sit in other teams' apps
Bug fixes and refund approvals live in engineering and finance tools, away from the ticket.
-
Know-how leaves with agents
The workarounds senior agents know leave when they move on.
-
Customer data ends up in personal AI accounts
Agents paste ticket threads into ChatGPT or Claude to get help, outside anything the business controls.
What is an AI OS for customer service?
An AI OS for customer service is an AI operating system with three layers. Your support team's apps sit underneath and the AI assistants and AI agents your team uses sit on top, with one memory in the middle. It is the software that joins them into one platform for the support team. Ditto reads every app you connect and keeps that memory. Every AI assistant and AI agent you connect works from it, and anything Ditto wants to change waits for your approval by default. Ditto is a central brain set up for a support team.
Central brain for teamsThe three layers of an AI OS
How Ditto keeps customer data private
You choose which models see your customer data and what Ditto may change.
Where your team's data goes
- Private inference in the Ditto appModels inside the Ditto app run through encrypted, private inference, so your data is not handed to the big AI companies.
- Zero-data-retention modelsChoose models whose provider keeps no copy.
- Temporary chatsStart a chat that is not saved to memory.
- Not used to train AI modelsWhat Ditto remembers about your customers is not used to train AI models.
What Ditto can do
- Read-only by defaultDitto starts by reading only the apps and accounts you connect. Disconnect an app and Ditto stops reading it.
- You approve every changeAnything Ditto wants to write waits for a person to review it first. You can let routine actions run on their own.
- Open any memoryEvery memory shows where it came from. Edit or delete it at any time. You can also export your memory.
- Team memory stays when someone leavesWhat an agent added to the team's memory stays when they leave.
Ditto's memory harness and integrations are open source.How Ditto handles your data
AI assistants for every role on your support team
Each role hands Ditto the work around a customer reply. Ditto prepares it from the support team's memory, and a person approves it before anything changes. Each tab shows the outcome for that role.
Support agents
Before you reply, ask Ditto what this customer told the team before, in any conversation in a connected app. Ditto answers with the source for each point and drafts the reply from that. The reply waits for your approval.
AI for customer support that starts from what the customer already said.
Team lead
Ask what is open in the queue and the decisions made on each case. Ditto answers from the tickets and team chats you connect, with sources. Each Monday, Ditto prepares a summary of open escalations in the Ditto app. If you want it posted to Slack or Teams, the post waits for approval.
The lead spends the week on the escalations that need a decision.
Escalations
When an agent escalates a complex issue, the next person needs what engineering said about the bug. They also need what the customer was told. Ditto finds both across Slack and Jira and drafts the customer update for approval.
Each escalation starts from what engineering and the customer already said.
Knowledge and training
New agents ask Ditto how the team handled a similar case and get the answer from past tickets and internal policies, with its source. Ask Ditto to draft a help article from a solved case, and the draft waits for review.
New agents find the team's past answers without asking a senior agent.
Billing questions
A billing question needs what the customer was charged in Stripe and what an agent promised them. Ditto finds both and drafts the reply and any refund, which wait for approval.
Billing replies match what the team already agreed with the customer.
How Ditto works in a customer service team
Ditto starts read-only in your support team's apps. You decide what it may change.
Connect your support team's apps
Sign in to any app your support team uses. Ditto starts read-only in each one.
Ditto builds one memory
It keeps what each connected app holds in one memory of the support team, and keeps it current.
Ask, then approve
Ask in the Ditto app or in ChatGPT or Claude. Ditto answers with sources and waits for your approval before it changes anything.
Ditto vs other customer service technology
Support teams use AI for customer service from their help desk platforms and from general AI assistants. Ditto keeps one memory of the support team under the AI assistants you connect.
-
Help desk AI agents
Zendesk and Intercom offer AI agents, such as Intercom's Fin, that answer customer questions.
-
ChatGPT and Claude
General AI assistants keep memory per person or per project.
-
Ditto
One memory under the AI assistants you connect and your support team's apps. It prepares replies and fixes and waits for your approval. Private inference in the Ditto app, and what Ditto remembers is not used to train AI models.
Connect Ditto with over 1,400 apps
Connect any app your support team works in and it joins the team's memory. Examples include Gmail and Shopify. Don't see your app? Ditto connects to any app through native connectors and MCP.
Browse all Ditto integrationsPlans for customer service teams
Enterprise is a custom build: Ditto for enterprise support teams. Memory uses credits and funded storage, with no monthly read or write count quotas.
New plan preview. Rollout is in progress; check the offers available in your account before purchasing.
Pro
$97/month Monthly subscriptionShared capacity and credits for you and your team.
Plan includes:
- 10 spaces
- Paid chat connections
- Ditto Review
- 1 org · 4 included licensed seats
- Choose models manually
- Custom domains
Enterprise
CustomA custom Ditto build for your business: our team writes the spec, builds it and supports it.
Frequently asked questions
What are AI assistants for customer service?
AI assistants for customer service are artificial intelligence tools that help a support team with the work around each reply, such as finding what a customer said before and drafting the answer. With Ditto your AI assistants and AI agents work from one memory of your support team. Every change Ditto wants to make waits for your approval by default.
What does an AI operating system do in a customer service team?
It keeps one memory of the support team under the apps and AI assistants your agents already use. Take a customer who emails about a bug. Ditto checks Jira and the team's tracker, then answers with its sources. It drafts a reply to the customer plus a tracker update, both waiting for a person to approve.
Is an AI OS safe to use with customer data?
Ditto starts read-only and only reads the apps you connect. A person reviews every change Ditto wants to make, unless you let a routine action run on its own. What Ditto remembers is not used to train AI models. You can choose zero-data-retention models and keep a separate memory per client. Temporary chats are not saved to memory.
How do you use AI for customer support?
Using AI for customer support with Ditto starts with email and team chat. That is where customers and teammates write. Connect those apps, then ask Ditto about a customer before you reply. It answers with sources and drafts the reply for your approval, which keeps human oversight on what customers receive. Add the apps engineering and billing use next.
Why do AI assistants forget what our customers told us?
AI assistants work from what fits in their context windows: the recent turns of the conversation history and whatever you paste in. ChatGPT and Claude keep memory per person or per project, so what one agent told them doesn't reach the rest of the support team. Ditto keeps one memory of the team across every AI tool you connect.
How does Ditto's memory work?
Ditto saves what it learns as memories and organises them into a knowledge graph of subjects such as a customer or a known bug. When someone asks a question, Ditto finds the relevant memories by semantic similarity in its vector database. It ranks them and adds the best matches to the model's context window, after the system prompt.
Can we choose which tasks run without approval?
Yes. By default, every change Ditto wants to make in a connected app waits for a person on your support team to approve it. You choose which routine tasks to automate. Everything else waits in the approval queue until a person on your support team approves it.
Can I use ChatGPT or Claude with Ditto for customer service?
Yes. Connect ChatGPT or Claude to Ditto over MCP. Inside them, Ditto recalls the support team's memory and saves what it learns, so each chat starts from what your team already knows about a customer. In the Ditto app you can also use every major foundation model.
Which AI models does Ditto use?
Ditto doesn't train its own foundation models. In the Ditto app you choose from foundation models such as GPT and Claude, or bring your own model key on Pro and Team. What Ditto remembers about your customers is not used to train AI models.
Which apps does Ditto connect to?
Any app. Ditto has native connectors for common apps and connects to the rest over MCP. Gmail and Slack are common starting points, for example. Each integration adds that app to the support team's memory, so an answer can draw on any app you connect.
Does Ditto connect to our help desk?
Your existing help desk stays the support team's record of tickets, and Ditto works alongside it. Ditto connects to any app through native connectors and MCP, including MCP servers your team approves. Replies Ditto drafts wait for an agent's approval. Talk to us about your setup.
Will AI replace customer service agents?
Ditto does not replace your agents. It prepares the work around each reply. That can be the answer to the customer plus the fix in a connected app. By default a person on your support team approves each draft before anything changes, and what the team decides is saved to its memory.
Is our data used to train AI models?
No. What Ditto remembers about your customers and your support team is not used to train AI models. You can edit or delete any memory at any time. You can also export your memory, or choose zero-data-retention models, whose provider keeps no copy.
What does Ditto cost?
Ditto offers Pro and Enterprise plans for support teams. Enterprise is a custom build for larger support teams, planned with your team. Memory uses credits and funded storage, with no monthly read or write count quotas. See Ditto pricing for the current plans and what each one includes for a support team.
Give your support team a memory.
Ditto connects to any app your support team uses and keeps one memory under the AI assistants you connect.
Read-only by defaultNo card to start