KKadoa

Kadoa builds AI agents that turn websites, filings, PDFs and feeds into validated datasets. The agents write the extraction code, run it on a schedule, watch quality metrics and repair the pipeline when a source changes, calling in a person only when they get stuck. Every value links back to its source. The Zurich company was founded in 2023 and serves hedge funds, private equity firms and asset managers.

It fits data and research teams that maintain dozens or hundreds of sources and lose time to broken scrapers. Pricing is quoted to usage, with a free evaluation period on every plan. Kadoa scores 8.2/10 on our scale, led by autonomy and integrations.

Kadoa by the Numbers

Kadoa is a young company with a focused buyer. These facts come from Kadoa, its investor and a funding database.

The slot classes tell a buyer how to size a contract. Kadoa says complex workflows are under 1% of the total, so most sources will cost one slot or a tenth of one. A team can estimate its bill by counting sources and checking which ones are lightweight API-style requests.

The customer base matters too. Funds care about data lineage and audit trails, which explains why every value carries a source link. Buyers outside finance get the same controls, though the product language and examples lean toward investment research.

Kadoa’s Pipeline From Source to Validated Dataset

Kadoa’s unit of work is the workflow. Its docs define one workflow as one schema extracted from one source. Editing the schema keeps it the same workflow.

Defining the Source and Schema

You point Kadoa at a source, such as a company investor relations page, a regulator’s filing list, a PDF report or an API feed, and describe the fields you want. Kadoa’s agents write the pipeline code. Building and previewing are free, so you can try several sources before anything is billed.

Running the Workflow

A workflow can run once, on a schedule or, on Enterprise, as a real-time monitor. Refresh rates run from daily to near real time. Human-like browsers, rotating global IP addresses, datacenter and residential proxies and regional caching handle access to the source.

Validation Before Delivery

The trust adapter checks each batch for completeness, plausibility, schema fit and domain rules you set, such as “production volume cannot be negative” or “ticker must match the list.” Every value links to the page or document it came from, and an audit trail records checks and changes.

Self-Healing When Sources Change

Always-on agents watch logs and QA metrics. When a check fails, they investigate, repair the pipeline and test the repair. If they cannot fix it, Kadoa notifies a person. This is the main difference from a classic scraper, where a site redesign often means a developer rewrites code by hand.

Delivering Data to Your Stack

Results go to Snowflake, Amazon S3 or Google BigQuery, or through the REST API, SDKs, MCP or webhooks. Enterprise includes all integrations, and Starter includes basic ones.

Workspaces and Team Access

Starter is designed for an analyst or a small team. Enterprise adds shared workspaces with unlimited users and SAML SSO, so a research group, a quant team and a compliance reviewer can see the same workflows and audit trail. A dedicated account manager and an enterprise SLA come with that plan, which matters when a dataset feeds trading or investment decisions.

MCP support also means analysts can query or trigger Kadoa from an AI assistant. A researcher could ask an assistant for the latest validated figures from a set of sources and get values that still carry their source links.

A Sample Month for a Commodities Desk

A research team tracks quarterly output from 60 mining companies. They set up 60 workflows, one per company report page, each extracting production, grade and recovery figures. Most run quarterly with a daily check for new filings. In week two, one company moves its reports to a new page; the agents detect the failed check, rewrite the pipeline and pass the retest. In week three, a PDF layout changes beyond what the agents can repair, and an analyst gets a notification to review it. Validated numbers land in Snowflake with a link to each source page.

Two weeks later, the portfolio manager asks for five more producers. The analyst adds five workflows, previews them for free, and activates the schedules once the first outputs match the companies’ published figures. The slot count rises from 60 to 65, and nothing else in the setup changes.

Kadoa Plans and Slot Billing

Kadoa offers two plans and bills both by active workflow slots. Prices are quoted.

Slots work like seats for data sources. A scheduled or real-time workflow uses its slot while active, for at least 30 days. Pausing a workflow frees the slot after that minimum. Some contracts count approved workflows instead, where a workflow counts once it is approved and has finished one successful run. The Usage page shows active workflows and slot use.

Slot Counts for Three Teams

These scenarios count slots, which a buyer can bring to a pricing call.

An analyst monitoring 20 standard web sources on a schedule uses 20 slots. If 10 of those are lightweight API-style requests, the count drops to 10 standard slots plus 10 at 0.1, or 11 slots.

A research team with 60 standard sources and 200 lightweight feeds uses 60 slots plus 20, or 80 slots. One-off backfills of historical data cost nothing in slots, since one-time workflows are free.

A central data team with 400 standard workflows, 1,000 lightweight ones and four complex ones at 2 slots each uses 400 plus 100 plus 8, or 508 slots. At that scale, Enterprise volume discounts and an approved-workflow contract are worth discussing.

Questions to Bring to a Kadoa Pricing Call

Bring a source list with an estimate of how many are standard web pages, lightweight API-style requests and complex multi-step sources. Ask for the per-slot rate at your volume, whether the contract counts slots or approved workflows, and how the 30-day minimum applies when a workflow is replaced. Ask which integrations Starter includes, since Enterprise lists all of them and Starter lists basic ones.

Ask how real-time monitors are priced, since they are listed under Enterprise. Ask what support response times look like on Starter, since its support is described as basic.

Kadoa Compared With Browse AI, Thunderbit and Browserbase

These tools all extract web data, but they suit different buyers and levels of control.

Kadoa is the most hands-off of the four for ongoing pipelines, since its agents maintain the code. Browse AI and Thunderbit suit business users who train or click their own extractions, and Browserbase gives developers browsers to run their own code. See more on our web scraping agents and data analysis agents pages.

Billing models differ as well. Kadoa bills by active workflow slots, Browse AI and Thunderbit by credits tied to rows, and Browserbase by browser usage. Teams with many steady sources often prefer slots, while teams with bursty one-off jobs may prefer credits.

Research Teams That Suit Kadoa

Hedge funds and asset managers that track alternative data from company sites, regulators and industry bodies fit best. Source links and an audit trail on every value support research review, and self-healing reduces the time analysts spend on broken feeds.

Private equity firms monitoring portfolio companies and competitors can run scheduled workflows across websites and PDFs, delivered to the warehouse the deal team already uses.

Central data teams in large companies that support many internal requests also fit, through Enterprise shared workspaces, unlimited users and SSO.

Consultancies and research firms that deliver datasets to clients can also fit. Source links on every value let them show a client exactly where a number came from, and the audit trail records every check and change along the way.

Buyers Who May Want a Lighter Tool

Small businesses, marketers and recruiters who need a list from one site may find a no-code tool like Browse AI or Thunderbit quicker and cheaper to start. Developers who want to write and control every line of scraping code may prefer a browser platform.

Buyers who require published pricing before talking to sales may prefer tools with public price lists.

Slot Minimums, Source Terms and Review Steps at Kadoa

Plan your testing. Building and previewing are free, but activating a schedule holds a slot for at least 30 days. Test sources as one-time workflows before scheduling them.

Ask for a written quote. Prices are not published, so get the per-slot rate, volume tiers and any minimum commitment in writing.

Review escalations. Agents notify a person when they cannot fix a pipeline. Assign someone to handle those notifications, or stale data can reach downstream models.

Confirm source rules. Kadoa says it follows ethical data collection practices, and buyers in regulated industries should still have compliance review each source’s terms and any login-based access.

Keep schemas stable. Kadoa counts one workflow per schema per source, and editing the schema keeps the same workflow. Adding a new source, though, is a new workflow and a new slot. Group related fields into one schema where it makes sense to keep slot counts down.

Watch downstream models. Validation rules catch impossible values, but they only know the rules you give them. Add domain rules for the figures that matter most, such as ranges for prices or volumes, so bad data stops before it reaches a model.

Check certifications. The features page highlights compliance controls and audit trails. Ask for security documentation during procurement if your policy requires named certifications.

Plan the hand-off. Since agents write the pipeline code, ask how you can export or inspect that logic if you ever move providers, and keep your own copy of the schemas and validation rules.

Kadoa Funding Round

On March 7, 2025, VI Partners announced that it led a seed round in Kadoa, with Fyrfly Venture Partners participating. The amount was not disclosed. The announcement said Kadoa builds AI agents for data and research teams and serves hedge funds, private equity firms and asset managers. Co-founder and CEO Adrian Krebs said the company was founded to build an accurate platform for unstructured data.

Our Read on Kadoa for Data Pipelines

Kadoa scores 8.2/10 on our scale. Autonomy earns 9.2, the highest of the six, because agents write, monitor, repair and retest pipelines with little human help. Integrations earn 9.0 for warehouse, API, SDK, MCP and webhook delivery.

Trust and security earn 8.2 for source links, audit trails and Enterprise SSO. Ease of setup earns 8.0, with a self-serve evaluation on Starter. User sentiment sits at a neutral 7.5 since public ratings are not yet available, and pricing value is lowest at 7.2 because prices are quoted.

For a fund or data team spending analyst hours on broken scrapers, Kadoa is worth a trial. Build your ten most fragile sources as one-time workflows during the evaluation, compare the output against your current feeds, and then ask for a slot-based quote.

Data Team Questions About Kadoa

How Much Does Kadoa Cost?

Kadoa does not publish prices. Starter is consumption-based and Enterprise is a custom contract. Both are billed by active workflow slots, and every plan includes a no-commitment evaluation period.

What Is a Slot in Kadoa?

A slot is the billing unit for an active scheduled or real-time workflow. Standard workflows use 1 slot, lightweight ones 0.1 and complex ones 2 or more, for at least 30 days once activated.

Does Kadoa Fix Broken Scrapers on Its Own?

Yes. Always-on agents watch logs and QA metrics, repair a pipeline when a check fails and test the repair. If they get stuck, Kadoa notifies a person.

Can Kadoa Extract Data From PDFs?

Yes. Kadoa lists websites, filings and PDFs, and APIs and feeds as sources. Every extracted value links back to where it came from.

Where Does Kadoa Deliver Data?

To Snowflake, Amazon S3 and Google BigQuery, or through a REST API, SDKs, MCP and webhooks. Enterprise includes all integrations.