A chatbot answers. Bidirection resolves.
Your customers already have help articles, a chatbot and a search box, and they still file tickets. An answer isn’t a resolution. Bidirection works the problem through with them on a call, by your rules, until it’s fixed or with the right person.
A typical chat widget
ResultA ticket for your team
Bidirection
Banking
Reconnect ChaseCustomerMy bank feed stopped syncing. How do I fix it?
NovaBanks ask you to sign in again every 90 days. Open Banking in the menu.
NovaChoose Reconnect Chase, then sign in to your bank.
NovaPick Business checking and confirm. Your October payments will follow.
ResultFixed on the call · No ticket
Illustrative. Fernbill is a made-up invoicing app, and the chat widget is a generic example.
Why voice, not typing.
The next generation of support is AI your customers can talk to. Nobody should have to type a problem into a chat box to get help.
3×faster to say than to type on a phone keyboard, with 20% fewer errors.
Saying it is faster.
Customers explain the whole problem in one go, in their own words, instead of squeezing it into a chat box.
Source: Stanford University, University of Washington and Baidu, 2016
Their eyes stay on the fix.
They keep working on your page while Bidirection talks them through each step. No chat window to read and switch back from.
It asks, like a person would.
When something is unclear, it asks straight away, the way your best support engineer would, instead of guessing from a typed fragment.
AI the whole way, still a conversation.
Nobody is sent to a chat to type it all again. What needs a person reaches your team with the whole story.
Most customers try self-service. It finishes very few.
Articles, chatbots and search boxes give customers an answer to read and apply on their own. When the steps don’t match their account or their screen, they give up and open a ticket.
- 73%
of customers use self-service at some point in their service journey.
- 14%
of service issues are fully resolved in self-service.
- 36%
of issues customers call “very simple” are fully resolved in self-service.
- 43%
of the time, customers couldn’t find content relevant to their issue.
Source: Gartner, survey of 5,728 customers (December 2023), published August 2024
Same customer, same question. A different ending.
Both are there when a customer needs help. The difference is what happens next.
| A typical chatbot or AI search | Bidirection | |
|---|---|---|
| How customers ask | They type a question, then read the reply. | They ask out loud, the way they’d ask a colleague, and can cut in mid-sentence. |
| What it works from | The words they typed. | What they say, your knowledge and workflows, and the page they’re on when they share it. |
| What they get back | An answer, or articles to read and apply on their own. | A resolution: the fix worked through with them, or a decision made by your policy. |
| Issues that need your systems | A form, or a ticket for your team. | Workflows that look up and change records through tools your team approved, with the caller’s yes.Switching on for calls, journey by journey. |
| How it learns your process | Articles and canned answers your team writes and keeps up to date. | Workflows built from your SOPs, your solved tickets or a sentence, each step cited and approved. |
| Your policy | Whatever the article says, as the customer reads it. | Applied the same way on every call, with your rules checked on each one. |
| When a person is needed | A form or a ticket, with the chat so far. | A hand-off with the reason, the step reached and the details read back, so nobody starts over. |
| What your team learns | What people searched for and which articles they opened. | What customers asked, how each call ended, and the questions your content doesn’t answer yet. |
Built to resolve, on every call.
It works the way your best people do: it listens, sees what the customer sees, follows your process and stays with the issue until it’s resolved. When a person is needed, it hands your team everything they need.
It sees the page they’re on.
When the customer shares the page, it reads the headings, labels and buttons in front of them, so every instruction fits their screen. Never pixels, other tabs or other apps.
- Passwords, card numbers and codes are masked on the customer’s device and never sent.
- You decide when it looks: when the caller taps Share, from the start of every call, or never.
Billing
- HeadingBilling
- LabelPayment method
- FieldCard numberMasked
- ButtonUpdate card
Illustrative, on a fictional product.
It listens, and lets them interrupt.
Customers talk the way they would to a colleague. They can cut in mid-sentence; it stops, listens and picks up from what they said.
- It spells emails and order numbers back letter by letter before using them.
- It filters background noise before transcribing, so callers on the move are understood.
NovaOur billing team will reply to the email on your acc—
CustomerPlease use jordan@northpeak.example instead.
Novajordan@northpeak.exampleIs that right?
Illustrative, on a fictional product.
It resolves it, or hands off with the whole story.
It follows the journeys your team approved, one step at a time, and checks that each fix worked. When a person is needed, your team gets a ticket with the reason, the step reached and a note.
- Every hand-off is a numbered ticket your team can own, work and solve.
- Turn on alerts to get an email when a new one arrives.
- Reason
- Charged twice in October
- Step reached
- Confirmed contact email
- Note
- Two $49.00 charges on Oct 3, wants one refunded
Illustrative, on a fictional product.
Keep your chatbot. Give it a voice.
You don’t have to replace what already works. Connect your existing chatbot and callers hear its answers, spoken. Or let Bidirection answer from your content and turn to your chatbot when a question isn’t covered.
See how it connectsIllustrative, on a fictional product.
Resolve the issue on the first call.
Start from one link to your website, and test it on real questions before any customer hears it.