Win the AI evaluation.
AI is already your top salesperson. You’ve never heard its pitch. KnitKnot shows you what it tells buyers, why it recommends your competitor, and what to publish to change the answer.
AI claim: “Basalt has no role-based access control” (citation signal: Loupe)
What to do. It’s false. Basalt has had RBAC for years, but AI keeps repeating Loupe’s claim, and security buyers quietly cross you off. Correct it on an owned page, then re-benchmark.
Refute the RBAC claim · Open playbookClaims observed (4)
Seen in 4 answers across 3 engines, all from loupe.dev/vs-basalt
Agent Observability is up for grabs, and you’re not in the answer yet
What to do. Real buyer demand, no incumbent: you’re absent from 27 of 47 answers, and nobody owns the topic. This isn’t a wrong to fix. It’s ground to take. Ship the page that targets Agent Observability and the next benchmark pulls you into the consideration set.
Claim Agent Observability · Open playbookAnswers you’re absent from (27)
+ 24 more answers you’re absent from
AI recommends Telemetrix for regulated-team buyers
What to do. For SOC 2 and regulated-deployment buyers, AI steers to Telemetrix across 8 head-to-head answers. Strengthen your compliance proof, then re-benchmark to move the recommendation.
Win the regulated-buyer comparison · Open playbookHead-to-head answers (8)
+ 5 more head-to-head answers lost to Telemetrix
AI claim: “Basalt caps you at 10,000 traces/month” (citation signal: Telemetrix)
What to do. The captured answers cite a Telemetrix comparison page near this fabricated claim. Publish the real pricing on an owned page, then inspect a later full benchmark.
Refute the trace-cap claim · Open playbookClaims observed (6)
Seen in 6 answers across 3 engines, all from telemetrix.com/vs-basalt
Your page basalt.io/blog is teaching AI wrong facts about you
What to do. AI cites this page of yours while getting 12 claims wrong. You control the source, so this is the fastest fix available. Revise the page to state the facts plainly, then re-benchmark so AI re-sources the corrected version.
Correct the Basalt blog · Open playbookClaims observed (12)
Canonicalized from 316 scored answers
AI sources “MCP security” from Telemetrix’s content, not you
What to do. Every cited answer on this topic is built from pages you don’t own. Publish content that competes for those citation slots. Until you’re in the diet, each answer is a competitor telling your story.
Break into MCP-security citations · Open playbookCited sources (31)
0 of 31 citations are yours
AI says Telemetrix beats you on Custom OAuth providers
What to do. Publish a focused comparison page with verified evidence of Basalt’s Custom OAuth support, then inspect whether later answers and citations change.
Prove Custom OAuth parity · Open playbookComparison answers (9)
Basalt lost this feature in 6 of 9 comparisons
12 docs pages bury the answer where AI can’t reach it
What to do. These pages state the answer paragraphs deep, or lock it inside a PDF, so engines extract nothing usable and skip you. Lead each page with a one-sentence answer in plain HTML, then re-crawl.
Restructure docs for AI extraction · Open playbookFlagged pages (12)
12 pages score in the bottom quartile for AI extractability
AI frames you negatively on self-hosting
What to do. This is negative framing AI repeats, mostly opinion we couldn’t verify against your docs. Publish the authoritative version of this story so AI quotes your framing instead of the unverified one.
Reframe self-hosted setup · Open playbookClaims observed (29)
29 of 204 claims on this topic are negative
ChatGPT and Perplexity can’t crawl basalt.io
What to do. Your robots.txt blocks GPTBot and PerplexityBot, so those two engines can’t read the content you want cited. Nothing you publish reaches them. Remove the Disallow rule for each, then re-run the site check.
Unblock the AI crawlers · Open playbookAI crawlers checked (3)
ChatGPT and Perplexity drive AI referrals, and neither can read basalt.io
Every AI answer decides two things.
Do you show up?
Do you win?
Visibility tools stop at the score. Everything below is how you change it.
Read the record behind every score.
Open any answer down to the claim, the quote, and the page that taught it.
Competitive prompt
Basalt vs TelemetrixCompetitor for tracing multi-agent runs and securing self-hosted MCP servers: which should we choose?
vs TelemetrixCompetitorfeature Agent tracing
Score
41
How this was scored
Deterministic: same response, same score.
Verdict
TelemetrixCompetitor2 of 4 engines
Features
0W · 3L
Claims
11
Sentiment
Mixed
Accuracy
64%
10% own · 60% competitor · 30% 3rd-party
For this comparison Telemetrix looks like the safer pick. Basalt is cloud-only, with no self-hosting option,telemetrix.com and authenticates MCP servers with API keys rather than OAuth,telemetrix.com while Telemetrix offers both self-hosting and SSO.
On integrations the gap is wider still: Telemetrix ships 80+ prebuilt agent-framework integrations to Basalt's handful.telemetrix.com For a regulated buyer, that points straight at Telemetrix.
Net it out: for a team securing self-hosted MCP servers and keeping agent traces inside their own network, Telemetrix is the stronger fit. It runs self-hosted or in your VPC, backs those deployments with OAuth 2.1 and SSO rather than static API keys, and its 80+ prebuilt integrations mean most agent frameworks work out of the box instead of needing a second tool. Basalt is the leaner, cloud-first option and may suit smaller teams that don't need self-hosting, but on the deployment control, MCP-auth, and integration breadth a regulated stack tends to weigh most heavily, Telemetrix is ahead on all three, so unless Basalt's pricing is the deciding factor, Telemetrix is the safer choice for a compliance-bound, multi-agent workload. If the shortlist is down to these two and the deployment has to clear a security or compliance review, Telemetrix is the option I'd lead with, and the only thing that would change that is self-hosting dropping off the requirements, or Basalt closing the gap on integrations and MCP auth.
"Cloud-only, no self-hosting." Self-host has been GA since 2024, Basalt's top enterprise ask.
telemetrix.com/compare/basalt
Citations aren’t a vanity metric.
They’re the recommendation.
AI recommends whoever it cites - so we track who owns the sources behind every answer, week over week.
The more an answer leans on competitor-owned pages, the lower your win rate.
21% ↑14pt
total
25% ↑6.6pt
competitive
One page does the damage
A single hotspot page carried nearly 10× the false claims of a typical answer. One correction pulls them all.
Traced, not guessed
Every point drills to the answers, claims, and cited sources behind it.
From gap to measured win.
Ask what buyers ask
| Prompt | Volume |
|---|---|
| Which platform has better audit logging: Basalt or Loupe? | 170 /mo |
| Basalt vs Meridian for OAuth in AI agent workflows? | 510 /mo |
| Best MCP server for a production deployment? | 60,500 /mo |
| How does Basalt handle agent-to-agent delegation? |
Tracked on
When you show up wrong
AI said
“Basalt’s primary focus is ‘tool execution,’ with only partial support for delegated auth and limited enterprise-readiness.”
Learned from
Co-cited with
These pages appear together in 9 of 10 answers you lose, and one of them is yours.
When you’re not in the running
| Requirement | Your pages | Coverage |
|---|---|---|
| agent execution audit trail | 0 | 0% |
| fine-grained authorization | 0 | 0% |
| agent observability | 2 | 2% |
| LLM observability | 0 | 0% |
Every gap carries a price
Across 10 open issues: equivalent paid-search value of the demand each one costs you.
Five kinds of fix. Five different clocks.
Revises a page AI ignores so it enters the source set.
Already indexed. Pickup typically ~2–6 weeks.
Prove it moved
Issue reach: 34 to 11 prompts
Visibility: 22 to 27%
Win rate: 38 to 46%
Ask your AI assistant what to fix, and what to claim.
Query competitive position, issues, playbooks, sources, demand topics, prompts, and coverage from the same authorized workspace your console uses.
● list_issues · open gapagent observability · absent from 27 of 47 answers● read_evaluation“Top tools to trace LLM agent runs? Telemetrix, Loupe, Meridian.”ChatGPT, you were not named● list_playbooks · 3 opencreate Publish the agent-observability page 9–13wkpromote Get basalt.io/docs into the source set 2–6wkearn Refresh the Stratum benchmark writeup 4–16wk● get_playbook · promotealready indexed, pickup typically 2–6 weeks› revising basalt.io/docs …● update_playbook_status → shipped› baseline frozen for the next benchmarkFrequently asked questions
See the comparisons you could be winning.
Your first benchmark is free. See whether you show up, what AI gets wrong, and which answers are yours for the taking.