Skip to content
Latest news

Yuma 2.0: one Handbook runs every customer conversation

Yuma 2.0 runs every customer conversation from one written Handbook: email, chat, SMS, WhatsApp, social DMs, outbound outreach and phone calls. You manage it in plain language, and every playbook is tested before it goes live.

YumaSeptember 28, 20265 min read
Yuma 2.0: one Handbook runs every customer conversation

Yuma 2.0 is an AI customer service platform where one written Handbook, managed in plain language, runs every customer conversation across email, chat, SMS, WhatsApp, social DMs, outbound outreach and phone calls.

That sentence is the whole release. The rest of this post explains what changed, why we built it this way, and what it means if you're already on Yuma.

The document you already have

Every support team has a document. How we handle returns. What to say when a parcel goes missing. Who gets a refund without a fight. You wrote it to train new hires, and it's probably the most accurate description of how your support actually works.

Here's the thing: none of the automation you've bought over the last three years runs on that document. It runs on configuration. Intent trees, workflow builders, per-scenario setup that takes weeks and drifts the moment a policy changes.

Yuma 2.0 runs on the document. We call it the Handbook.

What the Handbook is

The Handbook is a single, versioned source of truth for how your brand does customer service. It has three kinds of pages:

  • Brand pages define how Yuma sounds, per store, per channel, per language if you need it.
  • Playbooks say what to do in each customer situation.
  • Procedures hold the exact steps behind a playbook, written once and referenced everywhere.

You don't write it in a text editor. You describe how you do support, upload the SOP you already have, or point at real tickets, and Ask Yuma writes the pages. Every change lands as a draft. Nothing reaches a customer until you publish, and once you do, the next conversation follows the new rule with nothing to redeploy.

Maintaining your automation becomes maintaining your documentation. That's the entire mental model.

Tested, playbook by playbook

This is the part we're proudest of, and the part we haven't seen anywhere else.

Every playbook in the Handbook carries its own tests: example situations with the outcome you expect. Ask Yuma writes most of them for you while it builds the Handbook and whenever you fix a mistake. You can also point at a real conversation and turn how it played out into a test.

Change a return policy and Yuma rewrites the pages it touches, reruns the whole suite as simulated conversations (nothing real is touched), and shows you what passed. All green, you publish with confidence. Any red tells you exactly which page is unclear, and Yuma fixes it and reruns until it's green.

If you've worked with LLM evals, that's what these are. We just call them tests, because that's what a support lead would call them.

One Handbook, inbound and outbound

Support has always been reactive. A customer writes in, the AI answers. Yuma 2.0 flips that.

Event playbooks start conversations from what happens in your business: an order placed, a parcel delivered, a checkout abandoned, or a signal from your own systems. Yuma sends the message itself over email or SMS, and it follows up over days, scheduling each step on its own.

A typical sequence: reach out the moment an order is placed, follow up when it's delivered, and check in two days later to make sure everything's fine and offer proactive help. Same Handbook, same brand voice, same policies as inbound. Merchants running sequences like this tend to see fewer returns and a natural opening for a relevant upsell.

And now it picks up the phone

Yuma Voice answers a real phone number, takes backend actions mid-call, and reads from the same Handbook as every written channel. No separate script to maintain. If your returns playbook changes, it changes on the phone too.

You can hear it in the browser demo.

Managed from wherever you work

Ask Yuma is the interface to the whole system: ask questions, investigate a ticket, change a policy, add a test, schedule a weekly report. It lives in the Yuma dashboard, and through MCP it also lives in Claude and ChatGPT. Connect once and manage your Handbook from the tool you already have open. Changes still arrive as drafts for you to review.

You set the pace

Roll a new playbook out at 10% of matching conversations, with the rest handled as drafts your team reviews in the helpdesk. Raise it as trust builds. Set it to zero to pause.

Where the Handbook is silent, Yuma doesn't improvise. It hands the conversation to a person and writes the gap down as an open question for you to answer once.

What this looks like in production

Merchants on Yuma today automate a large share of their support volume. EvryJewels automates 89% of its tickets, Petlibro 79% and Clove 70%. At Tediber, Yuma is involved in 64% of all customer queries.

Yuma is SOC 2 Type II compliant. Read more on our security page.

If you're on Yuma 1.0

Your Auto-Pilots, processes and guidelines retire. The Yuma team runs the migration for you: Ask Yuma rewrites your existing setup as Handbook pages, and where two old rules conflict, the conflict becomes a task for you to settle rather than a silent choice.

Flows, custom actions and helpdesk macros keep working. A Handbook page names them where they apply.

Get started

Yuma 2.0 is available now. Open your Handbook from your account, or book a demo if you're new to Yuma.

Book a Demo | Open your Handbook

Stay in the loop

Get the latest on AI-powered customer service, straight to your inbox.