What did the research bots do?
For a proposal from the owner's training business, we needed a large comparison of course prices from many providers, in many formats and currencies, with sources for every figure. Instead of weeks of manual browsing, we ran several research bots in parallel, each assigned a slice of providers or course categories. They gathered prices and course details from public web pages, wrote them into a structured Excel workbook, and recorded the source URL for every row. A separate source audit checked the evidence.
This was delivered once, in-house, for the owner's own proposal. It's not a client case study and we make no claims about the proposal's outcome.Why use parallel bots instead of one?
Research at this scale is wide but shallow, with many similar lookups. Splitting it across bots with narrow, well-defined assignments is faster and easier to check than one long-running agent: each bot's output can be reviewed separately, and a failure in one slice doesn't spoil the rest. A coordinating step merges the slices into one workbook with consistent columns.
What did the workbook contain?
- Provider, course name and category
- Listed price, currency and any conditions (for example per person or per batch)
- Duration, format (online or classroom) and certification
- Source URL and the date it was checked
- Notes on ambiguity, for example "price on request" or "regional pricing"
Formulas and pivot summaries made it easy to compare ranges by category, so the proposal's pricing could be benchmarked against public information.
What was the source audit?
Research bots can misread pages or copy a number from the wrong section. The source audit re-checked a sample of rows against their URLs, flagged rows with missing or weak sources, and listed every item needing human verification. Only verified figures were used in the proposal. This is the same principle as the rest of our work: AI drafts, people verify.
Where does Outlook come in?
The workflow used Outlook to draft the covering email and organise the proposal correspondence. Sending stayed with a person. The bots prepared; the owner reviewed and sent.
What is the research bots' stop rule?
- Alone: browse public web pages, extract data into the workbook, record sources and flag uncertainty.
- With approval: any figure used in a proposal or external document (only after verification), and any email draft.
- Hard stops: logging into paid or private sites without authorisation, contacting providers, and sending any email.
- Handover: the workbook, the source audit and the list of rows needing verification.
Which other research tasks suit this approach?
- Competitor and market scans, like the competitor research behind this website
- Supplier or vendor comparisons
- Tender and RFP preparation: requirement checklists and pricing benchmarks
- Building lead lists from public directories, within legal and platform rules
- Policy or regulation summaries with sources, for expert review
Can you build research automation for us?
Yes, as a scoped project or pilot. Tell us the question, the sources and the output format you need (often Excel), and we'll design the bots, the audit and the review step. See consulting.
What should you prepare before talking to us?
You don't need a technical brief. A useful first conversation needs only a few things, and gathering them is valuable even if you never work with us:
- The process in plain words: what triggers it, who does what, and where it usually goes wrong.
- Rough volumes: how many messages, orders, tickets or invoices per week. A one-week tally is enough.
- Your tools: WhatsApp app or API, CRM, sheets, accounting or helpdesk software, including versions or plans where you know them.
- Three real examples: typical messages or cases, plus one awkward one, with personal details removed.
- What must never happen: the mistakes that would hurt most. These become the first lines of your stop rule.
- An owner: the person who will review the workflow once it runs.
Send these on WhatsApp or by email, or run the pilot scorecard first.
How does a pilot with us work?
Every engagement follows the same four steps. Consult: you walk us through the process, the tools and where it breaks. Pick the pilot: we choose one workflow and write down its inputs, outputs, tools, stop rule and success measure. Ship: we build it into the tools you already use and test it on real cases. Review: your team checks what the agent did and what it escalated, then decides whether to tune it, expand it or stop. Read the full pilot process.