← Back to blog

How we’re rewriting our “Running this across a company” section for buyers — behind the scenes

· 9 min read · By Vivek Chand

Three weeks ago a Head of InfoSec at a Series C fintech company emailed me after visiting clawmetry.com. She’d made it all the way through the page—past the hero, the features grid, the coverage table—and landed on our “Running this across a company” section. She had two questions.

First: “Is ClawMetry SOC 2 certified, and if not, what’s the timeline?”

Second, and this one surprised me: “Who should own this internally—my security team, the platform team, or the AI team we just stood up?”

She couldn’t find the answer to either from the section. So she emailed. She shouldn’t have had to. That’s on us.

This is the fifth post in our CXO messaging audit series. We’ve covered the hero (June), pricing (June), solution (July), problem (August), and footer CTA (September). The company section is the last one, and arguably the most consequential: it’s where a buyer who survived the rest of the page either commits to an enterprise conversation or decides to try it alone.

What the section actually says

The section opens with an eye text that I genuinely like: “For the people who own the risk.” That’s exactly the right frame. A CTO, Head of Platform, or Head of InfoSec evaluating ClawMetry for a 20-person AI team is not evaluating a developer tool. They’re evaluating a governance layer. They own the risk when an agent burns $3,000 overnight, when a session transcript leaks in a breach, when an agent calls a tool it shouldn’t have. That framing is right. Here’s where it goes wrong:

Current company section, four feature cards (as of November 2026)

Card 1: Nothing to install into the agent
“Each runtime is read from the session store it already writes. No SDK, no wrapper, no second agent on the laptop, nothing a developer has to opt into or can forget.”

Card 2: SSO, and an offline self-hosted mode
“SSO ships: SAML and OIDC through WorkOS, with the pinned domain gating provisioning. Roles are owner and member, and nothing finer. If your rollout needs more than that, it is not built yet.”

Card 3: It fits the network you have
“SIEM and syslog export in CEF or JSON over TCP, UDP or TLS, emitted daemon-side because the cloud never holds plaintext.”

Card 4: Fleet rollout is manual today
“There is no Helm chart, no Terraform provider and no Intune or Jamf package. If a daemon on developer laptops is a non-starter, run the enforcement proxy centrally and govern spend without touching an endpoint.”

Four cards. Each is accurate. Card 1 is excellent—the zero-install angle is a genuine enterprise differentiator, and the mechanical note (“a control layer that governs only its own SDK is blind to everything a developer runs outside it”) is the kind of thing a Head of Platform reads twice. The other three have problems I’ll name below.

The five-point gap audit

1 The ownership question is never answered

The eye text says “for the people who own the risk” but never names who that is in a large org. At a 50-person AI team, the answer is usually the Head of Platform. At a regulated financial institution with an AI team of four, the answer might be the CISO directly. At a startup with no dedicated platform team, the answer is the founding engineer who set up the Anthropic keys.

These are three different buyers with three different concerns. The CISO wants the SOC 2 timeline and the data residency story. The Head of Platform wants the rollout workflow and the SSO integration. The founding engineer wants a one-command install and a signal when something looks wrong.

Our section speaks to none of them specifically, and therefore lands ambiguously for all three. A one-paragraph opener that names the three buyer types—and which card is for which—would do more work than any individual card copy.

2 The SSO card is written for the developer implementing it, not the CISO approving it

Card 2: “SSO ships: SAML and OIDC through WorkOS, with the pinned domain gating provisioning. Roles are owner and member, and nothing finer.”

That sentence is written for the engineer who will configure the SAML integration. The CISO approving the tool purchase doesn’t know what WorkOS is. They don’t know what “pinned domain gating provisioning” means. They have one question: does it support our identity provider?

The answer to that question is yes—Okta, Azure AD, Google Workspace, any SAML 2.0 or OIDC provider. We just never say it in those words. “WorkOS” is implementation vocabulary. “Works with Okta, Azure AD, and Google Workspace” is decision vocabulary. The CISO copies one into their evaluation form and waves the other past without reading it.

The honest limitation (“roles are owner and member, and nothing finer”) stays. That’s exactly the kind of thing that should be in this card. It just needs to come after the approval-triggering confirmation, not before it.

3 “It fits the network you have” leads with protocols, not the outcome

Card 3: “SIEM and syslog export in CEF or JSON over TCP, UDP or TLS, emitted daemon-side because the cloud never holds plaintext.”

A CTO who runs Splunk parses this immediately. A Head of Platform whose SOC runs QRadar knows CEF. But a VP Engineering whose company outsources security operations, or a Head of InfoSec whose day job is governance rather than tooling, reads “CEF or JSON over TCP, UDP or TLS” and doesn’t know if ClawMetry fits their stack or not.

The outcome this card is describing is: agent activity feeds into your existing security toolchain without leaving your network. That’s a one-sentence enterprise unlocking argument. ClawMetry doesn’t add a new security panel your team has to monitor. It sends events to the same Splunk or Datadog your on-call already watches.

The technical detail (CEF, JSON, TCP, UDP, TLS) is validation for the reader who needs it, not the pitch. Lead with the outcome; let the protocol list serve as proof.

4 The gap card is visually indistinguishable from the feature cards

Card 4 is the one I find most interesting: “Fleet rollout is manual today. There is no Helm chart, no Terraform provider and no Intune or Jamf package.”

This is a rare thing in SaaS marketing: an honest statement of what isn’t built. I’m proud it’s on the page. Most tools in this space would bury the gap in a sales call, discover it at procurement, and lose the deal anyway. We put it on the landing page. That’s the right call.

But right now it looks like any other feature card. It has the same visual weight, the same card treatment, the same text color. A buyer skimming the section at 30 seconds sees four cards, reads them as four equal items, and comes away with a list that includes “fleet rollout is manual” and “SSO ships” at the same visual priority. That’s not what we intend. The gap card is a trust signal, not a feature description. It should look different.

Specifically: a gap card with a distinct visual treatment (a different background, a warning tint, an explicit “What’s not built yet” label) reads as deliberate transparency rather than a deficiency slipped into the feature list. The ‘gap list’ link in the CTA points to /enterprise#gaps, which is a full page. But most buyers won’t click through unless the gap card itself signals: we keep an honest, versioned list. Click to see it all.

5 The compliance story is absent

SOC 2, ISO 27001, GDPR, HIPAA. The section has no row for any of these. A buyer at a regulated company lands on the enterprise section and either has to click to /enterprise#gaps (where the compliance status lives) or emails us directly—as the Head of InfoSec did.

Our compliance posture is actually defensible and worth stating plainly: no agent conversation text in cloud, encryption key with the customer, air-gapped self-hosted mode available, SOC 2 audit in progress. That’s not a liability—that’s a reasonable answer for a startup at our stage. The problem isn’t the story; it’s that the story isn’t in this section.

An enterprise buyer who can’t find the compliance answer on the page either (a) emails us, (b) assumes the worst, or (c) skips to a competitor who has the answer visible. All three are worse outcomes than a single honest paragraph on the page that says what’s certified, what’s in progress, and what the data residency story is.

ClawMetry company section before/after comparison
Current company section (four uniform cards) vs. rewrite draft (buyer-typed opener, outcome-led cards, visually distinct gap card, compliance row).

What the rewrite will say

Working draft. Same commitment as the other posts in this series: publishing it now means it ships, not just gets discussed.

Rewrite draft: section opener, naming the three buyers

Three teams typically evaluate ClawMetry for a company rollout: the platform team (who wants zero-install governance across every agent their developers run), the security team (who needs agent activity in their existing SIEM and an honest SOC 2 answer), and the AI team lead (who wants to know what a runaway agent costs before it finishes). This section is for all three. The first card is for the platform team. The second is for security. The third is the gap list—what we don’t have yet.

That opener does the work the eye text currently leaves undone. It names the buyers, maps each to a card, and sets the expectation that there’s a gap list before the buyer has to find it.

Rewrite draft: SSO card, leading with IDP names

SSO with Okta, Azure AD, Google Workspace, and any SAML 2.0 or OIDC provider.

Domain-pinned provisioning: add the company domain and new accounts in that domain are blocked from creating personal instances. Roles today are owner and member. Finer-grained RBAC is not built yet—the gap list has the detail and an estimated timeline.

Rewrite draft: network fit card, leading with outcome

Agent activity feeds into your existing security toolchain. It never leaves your network.

Syslog export into Splunk, QRadar, Datadog, or any SIEM your SOC already runs. Events go out daemon-side in CEF or JSON over TCP, UDP, or TLS—emitted before the cloud relay, so your data never transits clawmetry.com in readable form. Corporate TLS interception is supported; a clawmetry doctor command walks the full network path.

Rewrite draft: gap card, visually distinct trust signal

What isn’t built yet. We keep a public, versioned list.

No Helm chart. No Terraform provider. No Intune or Jamf package. No SOC 2 certificate (audit in progress, expected Q2 2027). No RBAC beyond owner/member. No staged rollout or version pinning for the auto-updater. If any of these is a blocker, the gap list names everything and links the tracking issue.

Rewrite draft: new compliance row

Data residency: your agents’ conversations never leave your infrastructure in readable form.

Local-first compute on DuckDB. Cloud sync is AES-256-GCM encrypted on the daemon before transmission; ClawMetry’s cloud receives and stores ciphertext only. The decryption key lives with you. Air-gapped self-hosted mode available for regulated environments. We do not process, train on, or retain agent transcript content. GDPR: no personal data in the sync payload by default.

The rewrite adds one new card (compliance/data residency) and changes two existing ones (SSO, network fit) in vocabulary, not substance. The gap card changes only visually: different background tint, explicit “what isn’t built yet” label, link to the full versioned list. Card 1 (nothing to install into the agent) stays exactly as written—it’s the strongest card on the page and it’s already in buyer language.

What stays the same

Card 1 stays word for word. “No SDK, no wrapper, no second agent on the laptop, nothing a developer has to opt into or can forget.” That’s the kind of copy that only exists when you’ve talked to a buyer who got burned by an SDK-based governance tool that one developer never installed. It’s concrete, specific, and answers an objection before it’s raised. It doesn’t need rewording.

The eye text (“For the people who own the risk”) stays. It’s the right frame. The rewrite adds the opener paragraph below it to do the work the eye text sets up but doesn’t complete.

The “Get the fleet brief” and “The gap list” CTAs stay. They’re pointed at the right destinations. The only change is that the gap card now links directly to the gap list inline, so a buyer reading the section can reach it without scrolling to the CTA row.

The honest mechanical footnotes that follow each card (“a control layer that governs only its own SDK is blind to everything a developer runs outside it”) stay. These are the most unusual copy on the page. Most landing pages strip out anything that could read as a limitation. We keep them because they signal that the person who wrote this section actually understands the category, and buyers who are evaluating seriously notice that.

Why we publish the audit before we ship the rewrite

Same reason as the other four posts in this series: writing the audit publicly is the commit. Without a blog post, it’s a Notion doc that gets discussed, refined, and never merged.

There’s a second reason specific to this section. The “Running this across a company” section is the last section before the FAQ and the footer. A buyer who makes it this far has already decided ClawMetry is worth evaluating. What kills the evaluation at this stage is not a missing feature—it’s an unanswered question that makes them feel like they’d have to do due diligence work we should have done for them.

“Is this SOC 2 certified?” is due diligence work. “Who should own this internally?” is due diligence work. A section that sends a buyer to email for answers they should find on the page loses evaluations that were otherwise won.

We’re a startup. We don’t have SOC 2 yet. We don’t have Helm charts. We don’t have RBAC. The buyers who evaluate us against LangSmith or Datadog know that. The question is whether they find out from us, clearly, in one paragraph, or whether they discover it after two sales calls, by which point it feels like it was hidden.

Head of Platform, CISO, or VP Engineering? If you’ve evaluated ClawMetry for a company rollout and hit a question you couldn’t find the answer to on the page, I want to hear about it. vivek@clawmetry.com. No pitch. Just the question you couldn’t answer, so we can put the answer on the page before the next person has to ask.

The pattern: enterprise sections are written for procurement, not evaluation

Every developer tool with enterprise ambitions builds this section the same way. Someone lists the enterprise features (SSO, SIEM, self-hosted, audit log) and writes one sentence per feature. The result is accurate and useless to the buyer who needs to decide in 90 seconds whether to book a call.

The problem isn’t the feature list. It’s that the section is built for a procurement checklist reader, not an evaluation decision maker. The procurement reader wants line items. The decision maker wants answers to three questions: does this fit my security model, can I roll it out without touching every laptop, and what do I tell my CFO when it breaks?

Our section has the information to answer all three. “Nothing to install into the agent” answers the rollout question. The network card answers the security model question. The gap card, rewritten as a trust signal, is actually the best answer to the CFO question: a tool that publishes its limitations publicly is less likely to surprise you in production than a tool that hides them in a sales call.

The rewrite isn’t adding features or changing our story. It’s reordering the cards for the person making the decision, not the person filling in the form. That’s the same thing we did in every other section of this series, and it’s the same thing every developer tool needs to do at the moment it starts selling to teams instead of individuals.

We grew to 987k+ installs on developer word-of-mouth. That flywheel runs on the OSS install and the GitHub star. But the cloud product—multi-node fleet dashboards, Slack and PagerDuty alerts, approval workflows, Guard policies—converts when a platform buyer lands on the enterprise section and finds their questions already answered. The rewrite is that section.

See the section we’re rewriting

The current “Running this across a company” section is live at clawmetry.com. The rewrite ships this week.

See the enterprise section →
Cookie preferences