Your analysts ask questions in plain English. The numbers have to be right, and auditable.
A read-only assistant that turns questions into SQL limited to your schema, does every calculation in SQL or Python, writes commentary from locked templates, and logs every query.
You have this problem if…
- Analysts wait on a data team for routine numbers.
- You tried a chatbot over your data, and it produced plausible totals that were wrong.
- Compliance needs to know exactly which query produced which number.
What you get
- A read-only connection to Postgres, Azure SQL or Snowflake, limited to approved tables
- Text-to-SQL constrained to your schema, with the generated SQL shown beside every answer
- All aggregation in SQL or Python. The model never does arithmetic.
- Commentary written from locked templates
- A query audit log and a 50-question evaluation set
How it is done
- 01
Collect real questions
Fifty questions your analysts actually ask, with the correct answers.
- 02
Describe the schema
Tables, joins, definitions and the business rules behind each metric, written down once.
- 03
Generate and constrain SQL
Queries are generated against approved tables only, validated, and run read-only.
- 04
Compute, then describe
Numbers come from the query result. The model writes commentary from a template around them.
- 05
Evaluate
The 50 questions run on every change, and the answers are checked against the known results.
- 06
Log and hand over
Every question, query and result is logged, and the log is yours.
What usually goes wrong, and how we handle it
Plausible and wrong is the failure mode
A bad join or a missing filter gives a believable number. The evaluation set is the only reliable way to catch it.
Metric definitions live in people’s heads
"Revenue" means different things to different teams. Definitions are written into the schema description before any SQL is generated.
Read-only has to be enforced by the database
Read-only is set by database permissions, not by a line in the prompt.
Why you can trust us with this
Our founder architected the matching engine of a retail trading analytics product that returns base rates over full price history in under two minutes, validated out-of-sample before any pattern reaches a user.
He is first author of a peer-reviewed 2026 paper in International Review of Economics & Finance.
Analytics for 20,000 retail users at a tenth of the usual cloud cost
A compute-heavy analytics product that answers in under two minutes, at about 10% of equivalent AWS cost.
Case studyAnswers from SEC filings, each one traceable to its source
Question answering and summaries over 10-K and 10-Q filings and earnings calls. Report review time fell 95%.
What we don't do on this package
- No write access to your database.
- No investment advice or recommendations generated for your clients.
- For registered investment advisers, we involve your chief compliance officer early, so the query log fits your books-and-records policy under SEC Rule 204-2.
Questions
Which databases?
Postgres, Azure SQL and Snowflake most often. Others are fine if they support read-only roles.
Does our data leave our environment?
Query results pass through the model API to write commentary. If that is not acceptable, the assistant can return tables and charts only, with no model in the path after the SQL is generated.
Who does the work?
Suresh Rajendran, our founder, does the engineering on every engagement himself. Nothing is handed to a junior engineer or a subcontractor.
Where does our data go?
We work inside your accounts: your cloud, your repository, your model-vendor keys. Where a vendor offers zero data retention we turn it on, and your data is never used to train a model. Our access is removed and credentials are rotated at handover.
How do we pay?
The first step is paid up front. Builds are 50% to start and 50% on acceptance. We invoice from Destello AI Incorporated in US dollars, payable by ACH or card. If you go ahead with a build within 30 days of the first step, its fee is credited.
What if the result doesn't meet the bar?
Acceptance criteria are agreed in writing before work starts, and the second payment is due only when they're met.
Will you sign our NDA?
Yes, usually within one business day.
Start with the one-week prototype.
$1,500, credited to the build if you go ahead within 30 days. Book a 20-minute call, and you'll have a written scope with acceptance criteria within one business day.
Or email hello@destello.us or call (628) 202-5322. We reply within one business day.