Competitive intelligence
What a CI platform touches in your stack: CRM, call recordings, data and security FAQ
The nine questions a security reviewer asks before connecting Salesforce, HubSpot, Gong or Slack to a CI tool, and what Klue, Crayon, Kompyte and Steve publish.
A security reviewer asks the same nine things about a competitive intelligence tool: what the OAuth grant lets it do, whose identity it uses, what it copies and for how long, whether your data trains a model and which LLM provider sees it, who the sub-processors are, what the SOC 2 report covers, how transfers are papered, how Japan is handled, and what happens when you leave. Marketing copy answers almost none of them. Send the nine questions to the vendor before the contract reaches legal, and treat "we are SOC 2 compliant" as the cue to ask for the report. Every vendor position below was last checked on 15 October 2026.
What the OAuth grant lets a CI tool do
A CI tool reads your CRM through an OAuth app. Its reach is set by the scopes the app requests, and the consent screen is the first document to read. Ask for the exact scope list, which scopes write, whether the app holds a refresh token so it can work while nobody is logged in, and which objects and fields it writes to. Vendors rarely publish the strings.
HubSpot's OAuth documentation states that access tokens reflect the scopes the app requested "and do not reflect the permissions or limitations" of what the connecting user can do. An app approved by an admin can therefore read and write records the connecting user's own role could not. Optional scopes are dropped when the portal lacks the product they cover, so the list the vendor requested and the list you granted can differ. Ask the same questions of the Salesforce connection.
Gong works differently. Its app documentation says "Gong doesn't support user level OAuth. Authentication happens once on a global level", and that it "must be done by a tech admin". Whoever approves the connection grants access to every call the requested scopes cover. The page shows scopes such as api:calls:read:basic and api:calls:create without listing them all, so ask which call and transcript scopes the CI tool requests and whether any can create or change records.
Slack is the easiest to read, because Slack publishes every scope. channels:history lets an app view messages in public channels it has been added to; chat:write lets it post; users:read.email lets it read email addresses across the workspace. Crayon's Slack Marketplace listing declares that it can view content and information about channels, conversations and the workspace. A tool that only needs one channel should only be in one channel.
What is copied out, and for how long
Separate three retention periods: your data while the contract runs, prompts and completions at the LLM provider, and what remains after termination. Most vendors state the third and skip the first two.
Klue's terms (version of 17 October 2025) commit to destroying, deleting or de-identifying personal data on termination and state no retention period during the contract. Crayon's Slack listing says it retains customer data "for the duration of the engagement", deletes it on termination or request, and stores LLM prompts and completions for 30 days for abuse monitoring. Semrush, which owns Kompyte, publishes a security page that does not name Kompyte; it says personal data is deleted once no longer needed for its stated purposes, apart from copies kept as the law allows or in backups, and gives no period. Ask each vendor to put all three periods in the data processing agreement.
Training, LLM providers and sub-processors
Read the training sentence to the end, because it can carry a carve-out. Klue's terms (section 3.3) say customer data will not be "used to train, validate or enhance any AI Features except solely for the benefit of the Customer", and never to train publicly available AI systems or public large language models. Crayon's Slack listing says Crayon uses Microsoft Azure OpenAI, which handles all data as tenanted to Crayon and does not use Crayon data for model training; the listing does not say whether that covers the Crayon platform beyond the Slack app, so ask. Semrush's security page does not address AI training.
Then ask for the sub-processor list as a URL and the notice period for changes. Klue's terms link a published list and commit to incident notice within 72 hours. Crayon publishes a sub-processor list (updated 6 May 2026) with no notice period for changes. Semrush commits to making a sub-processor list available to customers. The LLM provider should appear on that list by name, with its retention term beside it.
Why "SOC 2 compliant" is not an answer
SOC 2 is an attestation: an auditor reports on controls over a stated period. The useful questions are which type (Type I is a point in time, Type II covers a period), which period, which Trust Services Criteria, and whether you can read the report.
- YesYesdocumented on the vendor’s own pages
- PartialPartialpart of the job, or limited by plan
- NoNothe vendor says it does not offer it
- Not publicNot publicwe found nothing; this is not a No
| Capability | Klue | Crayon | Kompyte | |
|---|---|---|---|---|
| Security review5 | ||||
| SOC 2 Type II report | ||||
| No model training on customer data | ||||
| Sub-processor list published | ||||
| Data location stated | ||||
| Deletion on termination |
Beyond the table, Klue's security page adds AES-256 encryption at rest and annual third-party penetration tests, with no report period and no mention of ISO. Crayon's trust center lists SOC 2 Type II and a 2025 penetration test, with both reports behind an access request and no period shown, so ask for the period covered. Kompyte belongs to Semrush, whose security page does not name Kompyte; it covers PCI DSS, GDPR and data processing addenda and names neither SOC 2 nor ISO 27001. "Not public" in the table means the vendor's pages say nothing either way; it is never a No.
DPA, transfers and Japan
For a European or UK controller the questions are familiar: a data processing agreement that names the vendor as processor, the transfer mechanism, and the incident notice clock. Klue's terms appoint Klue as processor and cite the June 2021 Standard Contractual Clauses for transfers to Klue and onward to its sub-processors. Semrush states GDPR adherence and data processing addenda. Check each vendor's own clause rather than the summary on its security page.
Japan needs its own line in the review. The Act on the Protection of Personal Information has rules on providing personal data to a third party in a foreign country, and the Personal Information Protection Commission publishes a guideline on them and a Q&A that asks whether storing data on a foreign server, or using a foreign-operated cloud, counts as such a provision (Q12-3 and Q12-4). Whether they apply to a given vendor turns on the facts and the contract, so the question goes to your legal team; the PPC states that only the Japanese texts of its laws have legal effect. The facts to collect for them are whether the vendor handles the personal data in the transcripts and CRM records it ingests, where its servers are, and which mechanism it relies on.
The reviewer's worksheet
Send these as written questions and file the answers with the contract. What was said in a demo cannot be filed. For features and cost on the same vendors, read the Klue, Crayon and Kompyte alternatives guides.
- The exact OAuth scope list per integration, marked read or write, with the objects and fields written to.
- Whose identity each integration uses: a named user, a dedicated integration user, or a global app approved by an admin.
- Retention during the contract, at the LLM provider, and after termination, each as a number of days.
- The training sentence in full, including any carve-out, and the named LLM provider with its retention term.
- The sub-processor list as a URL, and the notice period for additions.
- SOC 2 type, period covered, Trust Services Criteria, and when the report and bridge letter will arrive under NDA.
- The DPA, the transfer mechanism, and the personal-data incident notice clock in hours.
- For Japan: where the servers are, whether the vendor handles the personal data, and which APPI mechanism applies.
- Offboarding: export window, deletion date, written confirmation, and the destruction standard.
How Steve answers the nine
Steve publishes most of its answers in its SLA. Trissino holds a SOC 2 Type 2 attestation covering Security, Confidentiality and Availability for 1 October to 31 December 2025, and gives the report and a bridge letter to Enterprise customers under NDA. Trissino trains no AI models on customer data, and every AI provider on its sub-processor list (OpenAI, Google AI, Anthropic, Perplexity) is contractually bound to zero or short-lived processing-only retention. Customer data is stored and processed in North America. Sub-processor changes come with 30 days' notice, and a confirmed incident is notified within 48 hours. On termination there is a 30-day read-only export window, then deletion to NIST SP 800-88 with written confirmation.
On integrations, Steve connects to Gong, Avoma, Fireflies, Zoom, Outreach, Salesforce, HubSpot, Slack, Microsoft Teams and Google Drive. Its Slack page says Steve sees messages only in the channel you configure, and by default the person who posted an insight approves it before it enters Steve's knowledge base. The Gong page says Steve imports the transcripts Gong makes available and scans each for competitor mentions, which your team reviews before a battlecard changes.
The tradeoff is integration detail. The SLA does not publish the OAuth scope strings for Salesforce, HubSpot and Gong, and section 11.5 points Japanese customers to an Annex D that is not on the public page, so ask for both in the review. The Competitive Intel product page shows what those integrations feed.
Frequently asked
Does a competitive intelligence tool train AI models on my data?
It depends on the vendor, so read the training sentence to the end. Klue's terms allow training solely for the customer's benefit and bar public LLMs; Crayon's Slack listing says Azure OpenAI does not train on Crayon data; Steve's SLA says Trissino trains no models. Semrush owns Kompyte, and its security page addresses neither Kompyte nor training.
What can a CI tool do inside Salesforce or HubSpot?
Whatever its OAuth scopes allow, not what the connecting user is allowed to do. HubSpot documents that an app's access token does not reflect the user's own permissions. Ask for the exact scope list, which scopes write, and which fields the vendor writes to.
Is 'SOC 2 compliant' enough for a security review?
No. SOC 2 is an attestation report, so ask for the type, the period covered and the report itself. Klue states Type II with the report on request; Crayon's trust center lists Type II with the report on request; Steve's SLA states a Type 2 report for 1 October to 31 December 2025, given to Enterprise customers.
Where is my data stored, and does that matter under Japanese law?
Klue states AWS hosting, Crayon's Slack listing the United States and Steve North America; Semrush's page names US data centres but not Kompyte. Japan's APPI regulates providing personal data to a party in a foreign country, so a Japanese buyer records where data lives and takes the question to legal.
What happens to my data when I leave?
Get the deletion commitment in writing. Klue's terms say it will destroy, delete or de-identify personal data on termination; Crayon's Slack listing says it deletes on termination or request; Steve's SLA gives a 30-day read-only export window, then deletion with written confirmation.
Keep reading
All posts- September 14, 2026Competitive intelligence software buyer's guide: when to buy, five tools compared, and the test to run before you signWhen to buy competitive intelligence software, the four proofs to demand, five tools compared on twelve rows, and one test to run before you sign.
- October 7, 2026Klue alternatives in 2026: who each tool is for, and when to stay with KlueStay with Klue for a curated, seller-facing programme; switch only for a stated reason. Five tools on 15 capabilities, with dated sources and switching costs.
- October 7, 2026Crayon competitive intelligence alternatives in 2026: what it costs, who each tool is for, and what switching costsCrayon publishes no price; Vendr's February 2026 data shows a $30,000 median. Five CI tools on 15 capabilities, dated sources, and what to ask first.
- October 7, 2026Kompyte alternatives for 2026: what to check before you renew after the Semrush and Adobe dealsAdobe owns Semrush, Semrush owns Kompyte, and no document says what happens next. Three written questions and a dated five-vendor table.
Put Steve through the same nine questions