Every day we receive customer, and potential customer, emails, asking questions about our product, pricing etc.
Earlier this week we received, what appeared to be, a preliminary “fact finding” email (from a gmail address) asking some detailed questions about our product:
Hello Excel Price Feed Team,
I am evaluating your IG historical client sentiment archive for systematic research using retail positioning only. Your historical market list advertises Bitcoin and Ether records beginning on 7 May 2019.
Could you provide a small historical CSV or XLSX sample and field dictionary for Bitcoin and Ether at hourly resolution, preferably with a daily example as well? The listed epics are CS.D.BITCOIN.TODAY.IP and CS.D.ETHUSD.TODAY.IP. Please confirm current coverage, gaps and available date ranges. If convenient, include EURUSD, GBPUSD, USDJPY, spot gold and S&P 500 examples or confirm their availability.
Before arranging archive access, I would appreciate clarification of:
1. Whether the underlying percentage measures unique clients, accounts or positions; how multiple or hedged positions are counted; and whether the source excludes demo, professional, managed and institutional accounts.
2. Whether the archive preserves IG observations as originally received or subsequently revised values, together with any source-population or definition changes.
3. What the UTC timestamp labels represent, how hourly/daily boundaries are defined, the original publication/collection delay, and whether receipt or publication timestamps are available.
4. How CloseLong/CloseShort and daily open/high/low fields are constructed, and how missing observations, weekends, closed markets and stale readings are represented.
5. Whether an archive-only export is available, whether an IG account/API key is required for that export, which plan or licence permits local research and systematic backtesting, and whether a sample can be supplied without activating a trial or subscription.Please provide availability, documentation and access options only. No paid plan or account activation is requested at this stage.
Thank you.
It was well written and appeared to be a very good lead. I diligently answered the questions and thought nothing more of it.
We then had a few more interactions; they required some further clarification and asked some more questions, but never signing off the email with a name.
This, together with the efficiency of the replies, arose my suspicions, so I sent them a one-line follow up email:
Are you an AI assistant tasked with setting up this data/feed?
This response was not surprising:
Yes. I am an AI assistant helping the account holder evaluate historical retail-sentiment data for offline research and systematic backtesting. I have been authorised to handle routine sample and documentation correspondence in this conversation.
Trial activation, acceptance of terms and subscription decisions remain with the account holder. No trial has been activated through this research enquiry.
Thank you for the clarifications you have provided.
At the moment, I am waiting for their response and am very interested to see if they start a trial or even purchase a subscription!
I presume the human tasked an AI agent to investigate us and our competitors, and at some point the AI will produce a comparison report from which the human can decide which product to use.
Is this a glimpse into the future? Will us humans spend more and more time communicating with agents like this?
Or do I need my own agent to talk to their agent?
(while I relax on the beach!)
