Overview
The Batch API generates pre-configured insights across large numbers of companies in a single job. You submit a CSV where each row is a company and an insight, and you get a CSV of generated insights back. It's built for scheduled, high-volume work, not for anything you're waiting on in real time.
Batch jobs can run for hours, not seconds. If you need an answer right away, use the Chat API instead.
What it can do
Generate insights across thousands of accounts: one job covers many companies and many insight types at once
Fill CRM fields: output is designed to be loaded into a CRM, for example refreshing strategic priorities across every account in a territory
Run from a catalogue of pre-configured insights: each row references an insight by identifier, and you can list the available insights through the API
Support custom insights: Databook can configure additional one-line custom questions on request, including questions with a company placeholder so the same question runs across every row
Return formatted output: plain text and Markdown are supported, so answers can come back as tables, lists, or other structured content within reason
Insight categories
Databook maintains a catalogue of pre-configured insights, which you can retrieve through the API. Representative examples:
Category | Example insight |
Account planning | Give me an account plan for Nike |
Account intelligence | Tell me about Nike |
Meeting preparation | I have a meeting with the CIO at Nike this week, what should I know going in? |
Strategic priorities | What are Nike's strategic priorities? |
Discovery questions | What are some discovery questions for Ford? |
Competitive intelligence | What is the competitive landscape at Nike? |
Financial intelligence | What is Nike's financial outlook? |
Buyer intelligence | Give me a contact strategy for Nike |
You can also write a custom insight with a placeholder for the company, for example "Give me a concise summary in table form of management changes at {{company}} in the last 2 weeks", so one insight definition serves an entire batch.
How a job works
Authenticate to obtain an access token.
Prepare your input: download the input CSV template from the API, then populate one row per company and insight pair.
Submit the job: upload the CSV to the batch job endpoint. Databook returns a job identifier.
Poll for status: check the job endpoint until the job reports that it has stopped and the reason was success.
Download the result: request the job result to receive a temporary signed download URL for the output CSV.
You can also list your submitted jobs through the API, which is useful for monitoring a scheduled pipeline.
There are no email or webhook notifications today, so your application has to poll for completion. The download URL is temporary, so retrieve and store your results promptly once a job finishes.
Example use cases
CRM enrichment at scale: refresh strategic priorities across several thousand CRM accounts on a schedule
Territory planning: generate the account intelligence needed to build or review a territory plan
Recurring signal monitoring: check every account in a book of business for management changes each fortnight
Account plan enrichment: add strategic insight to account plans without a seller having to request each one
Meeting readiness: generate account-specific insights ahead of scheduled executive meetings
Sales plays and discovery: produce discovery questions or play recommendations across a segment
What to expect from output
Batch API answers use a version of the conversational engine tuned for throughput. They're more concise than answers from the Chat API or the Databook app, which use heavier prompt planning to produce longer, more contextual replies. Insight quality is comparable, but the length and shape are not, so you'll see a difference if you compare the two side by side.
Company matching works the same way as the Chat API: each row is resolved using Databook's entity linker, which may match a parent company or a subsidiary depending on which has richer insight coverage and the better contextual fit.
Limits worth knowing
Limit | Value |
Active jobs per customer | 5 |
Maximum rows per job | 3,600 |
Result download URL validity | 1 hour |
Completion notification | Not available, polling required |
Output formats | Plain text and Markdown |
If you have a very large workload, you'll need to split it into multiple jobs rather than submit it as one large file.
Getting access
Batch API access requires an entitlement and credentials issued by Databook. The insights you want to run also need to be configured by Databook first, including any custom questions, so scoping runs through your account team. Full request and response detail is in the API documentation portal