Now onboarding design partners
The API layer for Africa's portals that don't have one
AI browser agents that act on your behalf, with your permission, and return structured JSON through one API.
Book a demoOverview
Agent runs across your connected portals.
Tasks by portal
Daily agent runs, last 14 days
Portal health
4 portals connected
Step repaired. Bank statements login page changed at 08:12.
Recent runs
View all| Task | Portal | Action | Requested by | Duration | Status | |
|---|---|---|---|---|---|---|
| task_8f3k2n | Company registry | fetch_company_profile | ELExample Lender | 21s | Completed | |
| task_7h2m4c | Bank statements | fetch_statement | ELExample Lender | 12s | Running | |
| task_6c1mwd | Customs | track_clearance | EFExample Freight | 41s | Retrying | |
| task_9d0plb | Tax portal | fetch_filing_status | EAExample Accounting | 18s | Completed | |
| task_4e7hqa | Company registry | verify_directors | EKExample KYB | 27s | Needs review |
Consent
Every agent action starts with the account holder's permission.
Consent records
Export| Account holder | Portal | Scope | Expires | Status |
|---|---|---|---|---|
| AOAda Okafor | Company registry | Profile, directors | Feb 10, 2027 | Active |
| KMKwame Mensah | Bank statements | 12 months | Feb 9, 2027 | Active |
| EFExample Freight | Customs | Clearance status | May 10, 2027 | Active |
| ZBZainab Bello | Tax portal | Filing status | Nov 2, 2026 | Expired |
| ETExample Trading | Company registry | Profile | — | Revoked |
| CEChidi Eze | Bank statements | 6 months | Jan 25, 2027 | Active |
Audit log
task_8f3k2n Company registry · fetch_company_profile
Timeline
8 events · Nov 12, 2026
Replay
Step 7 of 8 · fields.extracted
Webhooks
Results are pushed to your endpoint as soon as an agent finishes.
https://api.example-lender.africa/hooks/daiko Active
Recent deliveries
View all| Time | Event | Task | Response |
|---|---|---|---|
| 09:14:23 | task.completed | task_8f3k2n | 200 OK |
| 09:12:51 | task.completed | task_9d0plb | 200 OK |
| 09:11:07 | task.retrying | task_6c1mwd | 200 OK |
| 09:08:44 | task.completed | task_4e7hqa | 200 OK |
| 09:05:19 | task.completed | task_2a9x7q | 503 · retried |
| 09:02:36 | consent.revoked | — | 200 OK |
| 08:58:10 | task.completed | task_1k8s3v | 200 OK |
Payload
task.completed · task_8f3k2n
Every agent run across your connected portals in one view: volume, portal health, and the tasks that need a person.
Your user grants consent in a secure connect flow. Credentials are encrypted and never exposed to your developers.
Every action an agent takes is logged and replayable step by step, so you can show exactly what was done, on which portal, and when.
Portal tasks can take a while. Get the result pushed to your endpoint instead of polling for it.
Illustrative screens. Sample data, not a real system.
The problem
The work still happens in a browser tab.
Company registries, tax portals, customs, pension administrators, telco dashboards and internet banking are built for a person at a keyboard. Fintechs, lenders, accounting firms and compliance teams pay staff to log in, click through and copy data out by hand.
Tax portals
Customs
Pension administrators
Telco dashboards
Internet banking

One company check, done by hand.
Verify Example Trading Ltd
- 01Open the company registry portalDone
- 02Log in with the shared team accountDone
- 03Wait for the one-time code on a colleague's phoneWaiting
- 04Portal timed out. Log in againAgain
- 05Search for the registration numberTo do
- 06Copy 14 fields into a spreadsheetTo do
- 07Repeat for the next clientTo do
No audit trail. No retries. No way to run it in parallel.
With Daiko it's one API callOps queue3 open
Slow
A lookup that should be one request becomes a ticket in a queue, waiting for someone to log in, click through and report back.
Copied by hand sheet
Two digits swapped. Nobody notices.
Error-prone
Figures copied from a screen into a spreadsheet go wrong in ways nobody notices until a credit or compliance decision depends on them.
Checks vs. peoplegrows together
ChecksPeople needed
Impossible to scale
Volume grows with headcount. Doubling the number of checks means doubling the number of people doing them.

Built for the parts that break
Integrations that repair themselves
A vision model detects when a portal's layout changes and fixes the broken step automatically, so integrations don't silently break.
Built for portals that go down
Automatic retries, queueing during portal downtime, and timeout handling for slow government and bank sites.
One-time codes, relayed securely
One-time codes sent to local phone numbers are relayed securely back to the agent, so a login can finish without a person watching the screen.
Permissioned, logged, replayable
Every action is user-permissioned, logged and replayable. Data is encrypted at rest and in transit.
- 1
- API for every portal you connect
- 0
- credentials exposed to your developers
- Every
- agent action logged and replayable
Connect, act, receive.
Three steps between a portal with no API and JSON in your system.
- 01
Connect
The end user or business grants consent through a secure Daiko connect flow. Credentials are encrypted and never exposed to the developer.
- 02
Act
Daiko's AI browser agents carry out the task on the portal: log in, handle OTPs, navigate, fill forms, download documents.
- 03
Receive
You get structured JSON back through a REST API or a webhook, with a full audit log of every action the agent took.
For teams whose work runs through portals.
Fintechs and lenders
Pull bank statements and business records with customer consent for credit decisions.
KYB and compliance teams
Verify company registration details and filings in seconds instead of days.
Accounting and tax firms
Automate filings and document retrieval across client accounts.
Logistics and trade companies
Track customs and clearance status without someone refreshing a portal all day.
One request in. Structured JSON out.
Create a task, name the portal and the action, and Daiko returns the result along with a link to the audit log for that run.
- REST API
- One endpoint for every portal. Tasks are described by portal, action and input.
- Webhooks
- Portal tasks can take a while. Get the result pushed to your endpoint instead of polling for it.
- SDKs
- Client libraries are being shaped with design partners. Tell us which languages your team needs.
curl -X POST "https://api.daiko.dev/v1/tasks" \ -H "Authorization: Bearer $DAIKO_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "portal": "company_registry", "action": "fetch_company_profile", "input": { "registration_number": "RC123456" } }'import { Daiko } from "@daiko/sdk";const daiko = new Daiko({ apiKey: process.env.DAIKO_API_KEY });// Creates a task and waits for the agent to finish.const task = await daiko.tasks.create({ portal: "company_registry", action: "fetch_company_profile", input: { registration_number: "RC123456" },});console.log(task.result.company_name);import osfrom daiko import Daikodaiko = Daiko(api_key=os.environ["DAIKO_API_KEY"])# Creates a task and waits for the agent to finish.task = daiko.tasks.create( portal="company_registry", action="fetch_company_profile", input={"registration_number": "RC123456"},)print(task.result["company_name"]){ "task_id": "task_8f3k2n", "status": "completed", "portal": "company_registry", "result": { "company_name": "Example Trading Limited", "registration_number": "RC123456", "status": "active", "directors": [ { "name": "Jane Doe", "appointed_on": "2019-04-02" }, { "name": "John Doe", "appointed_on": "2021-11-15" } ], "filing_history": [ { "type": "annual_return", "year": 2025, "filed_on": "2025-06-30" }, { "type": "annual_return", "year": 2024, "filed_on": "2024-07-12" } ] }, "audit_log_url": "https://api.daiko.dev/v1/tasks/task_8f3k2n/audit"}Illustrative. Field names, the SDK package and sample values show the shape of the API, not a final contract.
Consent first. Every action on the record.
Agents act on real accounts, so permission and traceability are part of the product, not a layer added afterwards.
Encryption
Data is encrypted at rest and in transit. Credentials are encrypted and never exposed to the developer integrating Daiko.
Consent flows
The end user or business grants explicit consent through the Daiko connect flow before an agent takes any action on their account.
Audit logs
Every action an agent takes is logged and replayable, so you can show exactly what was done, on which portal, and when.
NDPA by design
Daiko is designed around Nigeria's data protection law (NDPA) and similar regulations elsewhere in Africa.
Questions
What is Daiko?
Daiko turns African web portals that have no API into reliable, permissioned APIs. AI browser agents log in and act on those portals on behalf of a user or business, with their explicit consent, and return structured JSON through a single API.
Is it legal and permissioned?
Daiko only acts on an account after the account holder has granted explicit consent through the connect flow. Every action is logged and replayable, data is encrypted at rest and in transit, and the product is designed around Nigeria's data protection law (NDPA) and similar regulations elsewhere in Africa. Bring your compliance lead to the demo call and we'll walk through how this applies to your workflow.
Which portals are supported?
We're onboarding design partners, tell us the portals you need on the demo call.
How do you handle portal changes?
Agents are self-healing. A vision model detects when a portal's layout changes and fixes the broken step automatically, so integrations don't silently break. Retries, queueing and timeout handling cover downtime and slow responses.
Where does the name come from?
Daiko comes from the Japanese word 代行 (daikō), meaning 'to act on someone's behalf.' That's exactly what our agents do: they act for your business on the portals your team would otherwise have to work through by hand, with your permission, every time.
Now onboarding design partners
Tell us which portals slow your team down
Bring the list to a demo call and we'll scope it with you.
Book a demo