Composable Conversational AI Infrastructure

Turn every call into data you can build on

Get 1,000 transcription minutes on every new account.

What’s new: Flows: build pipelines on a drag-and-drop canvas

Build your way

From first request to working integration.

Flows

Build a pipeline on a drag-and-drop canvas, no code needed. Connect transcription, summary and tag blocks, run the flow, open each block’s result, then export it as code.

FlowsTranscribe & Understand

Playground

Transcribe ready

Transcribe

User00:00.1
Hi, I’m calling because my flight from Singapore to Tokyo was cancelled less than 12 hours before departure. I received an email saying it was due to operational reasons, but I had already booked a hotel and airport transfer. The alternative flight offered arrives a full day later, which significantly impacts my travel plans. I’d like to know what compensation or reimbursement options are available.
Agent00:27.0
I sincerely apologize for the inconvenience. I’ve reviewed your booking and can confirm that the cancellation was due to an aircraft scheduling issue. While I’ve already rebooked you on the next available flight, I understand that the delay affects your trip. In this situation, I can submit a request for reimbursement of reasonable additional expenses, such as hotel and transportation costs caused directly by the cancellation. If you have receipts for those expenses please send them to us, and I’ll escalate your case to our customer care team for review.

Install

pip install makimoto-kawa

transcribe.py

from makimoto import kawa

client = kawa.KawaClient(api_key="...")
job = client.transcribe("call.mp3", language="en")

if job.status == "succeeded":
    print(job.result.full_text)

Job lifecycle

# 1. Submit recorded audio. Returns a job ID immediately.
curl -sS -X POST "https://api.makimoto.ai/v1/transcriptions" \
  -H "Authorization: Bearer $MAKIMOTO_API_KEY" \
  -F "file=@call.mp3" \
  -F "language=en"

# 2. Poll the job until status is "succeeded" or "failed".
curl -sS "https://api.makimoto.ai/v1/transcriptions/$JOB_ID" \
  -H "Authorization: Bearer $MAKIMOTO_API_KEY"

# 3. Summarise and tag the finished transcription. Each returns a new
#    job ID; poll it the same way as step 2.
curl -sS -X POST "https://api.makimoto.ai/v1/summarize" \
  -H "Authorization: Bearer $MAKIMOTO_API_KEY" \
  -H "Content-Type: application/json" \
  -d "{\"transcription_job_id\": \"$JOB_ID\"}"

curl -sS -X POST "https://api.makimoto.ai/v1/tag" \
  -H "Authorization: Bearer $MAKIMOTO_API_KEY" \
  -H "Content-Type: application/json" \
  -d "{\"transcription_job_id\": \"$JOB_ID\"}"

Why build with Makimoto.

1,000free minutes

Every new account comes with them. Spend them in Flows, the Playground or the Python SDK.

Create your free account

Built for real conversations
From Toku, for conversations across Asia-Pacific: accents, multilingual speech and code-switching are part of the problem, not edge cases.
Composable by architecture
Keep the workflow stable as models, providers and requirements change.
Clear data location
The managed service processes customer audio and transcripts in Singapore.
Open by design
Inspect the orchestration, contribute on GitHub, and follow the roadmap as more components open.

Explore the Features

APIs

Use one endpoint. Chain all three.

Send recorded audio once. Use the result as a transcript, a summary, tags, or the input to your own workflow.

POST/v1/transcriptions

Transcribe

Submit a recording and retrieve speaker-separated text with segment-level timestamps.

"transcript": [
  {
    "text": "Hi, I'm calling because my flight from Singapore to Tokyo was cancelled …",
    "time_start": 0.066,
    "time_end": 26.046,
    "speaker_id": 0,
    "speaker_alias": "User"
  },
  {
    "text": "I sincerely apologize for the inconvenience. …",
    "time_start": 27.042,
    "time_end": 60.99,
    "speaker_id": 1,
    "speaker_alias": "Agent"
  }
]

POST/v1/summarize

Summarise

Generate a concise topic and summary from a completed transcription or from text you already have.

"type": "summary",
"result": {
  "topic": "Flight Cancellation Compensation",
  "summary": "A customer had their flight cancelled less than 12 hours before departure … The agent apologized for the inconvenience and offered to submit a request for reimbursement of reasonable additional expenses."
}

POST/v1/tag

One recording. Every output.

Play a sample call and watch Kawa return a speaker-separated transcript, a summary and tags. This demo replays a recorded run; open the Playground to send your own audio.

Playground demo

flight_cancellation.mp3

Run the sample to see the transcript appear.

Representative request for this recording. The tour does not call the API.

curl -sS -X POST "https://api.makimoto.ai/v1/transcriptions" \
  -H "Authorization: Bearer $MAKIMOTO_API_KEY" \
  -F "file=@flight_cancellation.mp3" \
  -F "language=en"
# -> { "job_id": "...", "status": "processing" }

# Poll until status is "succeeded" or "failed"
curl -sS "https://api.makimoto.ai/v1/transcriptions/$JOB_ID" \
  -H "Authorization: Bearer $MAKIMOTO_API_KEY"

# Then summarise and tag the finished job
curl -sS -X POST "https://api.makimoto.ai/v1/summarize" \
  -H "Authorization: Bearer $MAKIMOTO_API_KEY" \
  -H "Content-Type: application/json" \
  -d "{\"transcription_job_id\": \"$JOB_ID\"}"

curl -sS -X POST "https://api.makimoto.ai/v1/tag" \
  -H "Authorization: Bearer $MAKIMOTO_API_KEY" \
  -H "Content-Type: application/json" \
  -d "{\"transcription_job_id\": \"$JOB_ID\"}"

Audio: Synthetic call, AI-generated for Makimoto

Make a real request in the Playground

Sample transcript of flight_cancellation.mp3: Hi, I’m calling because my flight from Singapore to Tokyo was cancelled less than 12 hours before departure. I received an email saying it was due to operational reasons, but I had already booked a hotel and airport transfer. The alternative flight offered arrives a full day later, which significantly impacts my travel plans. I’d like to know what compensation or reimbursement options are available. I sincerely apologize for the inconvenience. I’ve reviewed your booking and can confirm that the cancellation was due to an aircraft scheduling issue. While I’ve already rebooked you on the next available flight, I understand that the delay affects your trip. In this situation, I can submit a request for reimbursement of reasonable additional expenses, such as hotel and transportation costs caused directly by the cancellation. If you have receipts for those expenses please send them to us, and I’ll escalate your case to our customer care team for review.

Flows

One canvas. Your whole pipeline.

Connect Start, Transcribe, Summarise and Tags blocks on a drag-and-drop canvas, then run the flow and watch each block move from queued to succeeded. This demo runs a sample call through a template. Create an account to build your own flow and export it as code.

Flows demo

Transcribe & Understand

Start from a template

Create your own flowsStart from scratch and build your own pipeline.

Flows lets you build a transcription pipeline on a drag-and-drop canvas: connect Start, Transcribe or Text, Summarise and Tags blocks, run the flow and export it as code.

Latest releases

Full changelog

16 September 2026Release
Makimoto 0.2.0
Summarise and tag endpoints, API keys that stay valid until revoked, and one developer portal for the quickstart, Playground, logs and keys.
Read the release notes: Makimoto 0.2.0
1 July 2026Release
Kawa launch
Post-conversation transcription API hosted in Singapore, with Kawa’s orchestration layer published under the MIT licence.

Turn every call into data you can build on

From first request to working integration.

Flows

Why build with Makimoto.

Built for real conversations

Composable by architecture

Clear data location

Open by design

Explore the Features

Use one endpoint. Chain all three.

Transcribe

Summarise

Tags

One recording. Every output.

One canvas. Your whole pipeline.

Latest releases

Makimoto 0.2.0

Kawa launch