Trusted Questions

These questions have been validated against known sources of truth. Use them to understand the types of questions to ask, the level of specificity to use, and the best way to phrase questions for accurate results.

Want to add a question to this list? Reach out in the SM Research Agent Slack help channel.

Basic SMDI info

Prompt	Comment
How many compounds registered to the project PELO‑HBS1lL have a logD (pH 7.4) between 1.0 and 3.0?	Any prompt relating to experimental physicochemical property data; the agent assumes the request is for experimental data unless otherwise indicated.
How many compounds registered to the project PELO‑HBS1lL have a calculated logD (pH 7.4) between 1.0 and 3.0?	Any prompt relating to calculated physicochemical property data; include the keyword calculated in the prompt.

Compound registration

Prompt	Comment
How many new compound lots have been registered under the TREX1 project in May 2026?	The agent understands project designations and can handle temporal data (caveat: it does not know the current date and is confused by phrasing like “last month”).
How many compounds (parent GNumber) are registered to the project “Mutant p53 CEPS” and are owned by the unix user WONGA38?	The agent knows about project associations and lot owners.
Summarize the origin (source) of compounds registered to the project “PELO‑HBS1L”; also summarize who has registered these compounds.	—

Single compound profiling

Prompt	Comment
What are the measured logD and kinetic solubility values for G03808140?	The agent can reliably retrieve information given a specific G‑number and assay name.
Provide a summary of measured data available for G03808140 in any binding affinity, enzymatic activity, and cell‑based activity assays.	The agent understands broad categorization of BCP assays.
What CYP/P450‑related DDI data is available for G03830770, including reversible inhibition, time‑dependent inhibition, nuclear receptor activation, and CYP induction assays?	Must include all keywords to capture the full scope of available CYP data.

Single assay profiling

Prompt	Comment
What are the top five most potent compounds measured in the assay “IRF‑Luc THP1 WUXI”?	The agent can sort results numerically.
What is the compound in the mPI3K_allo project that has the highest measured permeability? Look for permeability data for the gMDCK + 4% BSA cell line.	Specificity is key to ensure accuracy. If you want a specific species or assay condition (e.g. gMDCK + 4% BSA, fup measured in 10% plasma, IV cassette dosing), be sure to call it out.

Single project and defined multiple assays

Prompt	Comment
For the TREX1 project, how many compounds have gone through in vivo rodent IV PK studies? Make sure you look only at IV / IV cassette administration.	—
For the project Plk1, determine the most common biochemical and cell‑based assays, then generate a table containing the 100 most potent compounds in the biochemical assay showing the biochemical potency, the cellular potency, and the ratio between these results.	The agent is able to find the appropriate assay and generate the table by combining assays.

Multi‑assay profiling (broken into steps)

Best practice is to ask each step one at a time — in a single prompt or in successive prompts in the same chat. Multiple simpler questions let you troubleshoot if things get off track.

Prompt	Comment
1) What are our most potent molecules against ITK (IC50 in dose response assay, internal or in a kinase panel)? 2) Of these, do any have PK data available? 3) Do any have bioavailability >20% and half‑life longer than 2 hours in rats or mice?	Best to ask each of these one at a time; simpler questions let you troubleshoot.
Which compounds with IRF‑Luc THP1 WUXI EC50 < 30 nM have measured human liver microsome CLint < 10, rat liver microsome CLint < 25, mouse liver microsome CLint < 40, and measured gMDCKI permeability Papp > 3? Do not exclude values of CLint where the measurement is below the limit of detection. Do not consider data from any experiments flagged for “qualitative purposes only” that could be unreliable. Do not consider data from any cell lines besides “gMDCKI” (exact match).	Must specify how the agent should handle censored data and other caveats specific to the permeability assay.

Prompt

Comment

1) What are our most potent molecules against ITK (IC50 in dose response assay, internal or in a kinase panel)? 2) Of these, do any have PK data available? 3) Do any have bioavailability >20% and half‑life longer than 2 hours in rats or mice?

Best to ask each of these one at a time; simpler questions let you troubleshoot.

Which compounds with IRF‑Luc THP1 WUXI EC50 < 30 nM have measured human liver microsome CLint < 10, rat liver microsome CLint < 25, mouse liver microsome CLint < 40, and measured gMDCKI permeability Papp > 3? Do not exclude values of CLint where the measurement is below the limit of detection. Do not consider data from any experiments flagged for “qualitative purposes only” that could be unreliable. Do not consider data from any cell lines besides “gMDCKI” (exact match).

Must specify how the agent should handle censored data and other caveats specific to the permeability assay.

Complex questions in a single prompt

Prompt	Comment
What is the most potent KRAS G12C molecule in a cell assay that has in vivo PK?	This prompt could be enhanced further by providing details of which assay to use (e.g. biochem, cellular, assay name).
For the GLP1R project, how many compounds have measured in vitro metabolic stability data, measured either in microsomes or via the automated hepatocyte stability assay?	—
I want a tool compound for TMEM175. Do we have any TMEM175 activators that are also demonstrated to be selective (kinase panel and/or secondary pharmacology)?	—
I want a NaV1.7 inhibitor that can be used for ocular delivery. The most important things are high NaV1.7 potency, good selectivity over NaV1.6, and high solubility to enable dosing at high concentration in eye droplets. What are our best molecules for this?	—

Kinase selectivity data

Prompt	Comment
What kinase inhibition data do we have for G03081557?	—
What Flt3 kinase data do we have for G03081557?	—
What molecules do we have that are selective for IRAK1 over Flt3 in the kinase panel?	Automatically matched single‑point data across G‑number + concentration + vendor for accurate comparison between IRAK1/Flt3 results.
From kinome data, how many kinases does G00425669 inhibit with an IC50 lower than 1 µM?	The agent can retrieve kinome selectivity data and combine it with other queries; including kinome in the prompt guides the agent to use the kinome selectivity data.
Using kinome data, how many kinases does G02663206 inhibit by more than 50% at 0.1 µM concentration?	The agent can retrieve kinome selectivity data and combine it with other queries; including kinome in the prompt guides the agent to use the kinome selectivity data.
Based on kinome data, how many compounds inhibit the kinase PLK1 with an IC50 better than 1 µM?	A great way to find potential tool compounds for a given kinase.
`kinome search G00035608`	The agent knows about the kinome selectivity data from DiscoverX and Invitrogen. The keywords kinome search return single‑concentration % inhibition data for the listed compound.
`kinome IC50 G00035608`	The keywords kinome IC50 return IC50 data for the listed compound.
`kinome tree G02584994 @ 0.1 uM, DISCOVERX`	The agent can display a kinome tree for the single‑point query above. Must specify the G‑number, concentration, and vendor name.

See Prompt Shortcuts for the full kinome shortcut reference and save / export options.

Prompting Guidelines Prompt Shortcuts