Xtrusio AEO/GEO Audit

Nylas appears in 47 of 60 AI answers.

But ChatGPT leaves it out of all four meeting-recording queries.

Across 60 answers from ChatGPT, Claude and Gemini, Nylas is mentioned 47 times — more than any other tracked vendor in this audit. It records 26 first placements, while 10 of the 20 buyer questions mention Nylas across all three platforms. The weakness is concentrated in meeting intelligence: ChatGPT does not name Nylas on any of the four meeting-recording questions tested, while Recall.ai and Meeting BaaS appear on all four.

The findings below come from Xtrusio, an AI visibility audit system built specifically for B2B buyer-intent testing. Every mention was verified by running 20 real prospect queries across three generative AI platforms.

Questions were written from the perspective of engineering and product leaders evaluating communications infrastructure — email, calendar, meeting and agent-identity APIs.

September 2026
20 Queries • 3 Platforms
Nylas
95%
Claude
19 of 20 queries
7× TOP PLACEMENT
80%
ChatGPT
16 of 20 queries
11× TOP PLACEMENT
60%
Gemini
12 of 20 queries
WEAKEST PLATFORM
Where The Streak Breaks

Nylas is mentioned in 11 of 12 agent-identity answers and 11 of 12 compliance answers. Meeting intelligence returns 5 of 12.

Four product lines out of five achieve at least 75% overall coverage across their 12 platform-query opportunities. Notetaker is the exception. ChatGPT does not name Nylas on any of its four meeting-recording queries tested and names Recall.ai on all four; Gemini names Nylas on one of four; Claude names it on all four. No question in the audit returned a mention on zero platforms — within this query set the shortfall is concentrated in one product line rather than spread across the whole platform.

47/60
Total Mentions
26
First Placements
10
Unanimous Mentions
Section 2

Platform Scorecard

How often each AI platform names Nylas when a buyer asks

Nylas Mention Rate by Platform
Claude
95%
ChatGPT
80%
Gemini
60%

ⓘ Based on 20 buyer-intent queries run once per platform. 60 scored responses in total.

Vendor Comparison — Share of Answers Naming Each Vendor
Nylas
78%
Cronofy
25%
Recall.ai
20%
Unipile
20%
Aurinko
17%
AgentMail
15%
Meeting BaaS
15%
Cal.com
13%

ⓘ Scored across all three platforms — 60 responses. Bars are scaled to the highest vendor. First-party provider APIs (Google, Microsoft) are excluded; they appear as integration targets rather than as vendors competing for the purchase. Vendors named in a single session without corroboration elsewhere are excluded from all totals.

Claude names Nylas on every product line
Claude names Nylas on 19 of 20 queries, including all four meeting-recording questions ChatGPT omits. Its single miss is cross-company interview scheduling, where Cronofy is named instead.
Gemini's misses are scattered, not clustered
Gemini's eight misses touch four of the five product lines — multi-channel messaging, build-versus-buy, resource calendars, three of four meeting queries, agent deliverability and provider-direct comparison. Four of those eight name no vendor at all rather than naming a competitor.
Section 3

Tracked Vendor Leaderboard

Which vendors are named repeatedly across the 60 responses scored

Nylas is named three times as often as its nearest tracked rival. The vendors behind it do not appear across the whole query set — each concentrates in a slice. Cronofy is the most persistent, appearing on 15 answers and reaching beyond scheduling into post-call data and agent identity. Recall.ai and Meeting BaaS appear almost entirely in meeting queries. AgentMail appears only where agent mailboxes are the subject, and never on Gemini at all.

Competitor totals are filtered to repeatedly surfaced vendors. One-off names remain in the raw responses but are excluded from leaderboard totals, as are first-party provider APIs. This is a count of tracked vendors, not a complete market census.

Nylas — Platform Breakdown
Claude
19/20
Nylas named
ChatGPT
16/20
Nylas named
Gemini
12/20
Nylas named
Vendor Totals — All 60 Responses
Nylas
16
19
12
47
Cronofy
5
7
3
15
Recall.ai
4
5
3
12
Unipile
5
4
3
12
Aurinko
5
2
3
10
AgentMail
4
5
9
Meeting BaaS
4
4
1
9
Cal.com
2
3
3
8
ChatGPT
Claude
Gemini
Share of Answers
78%
Nylas
Nylas47
Cronofy15
Recall.ai12
Mention Intensity by Vendor
ChatGPT
Claude
Gemini
Total
Nylas
16
19
12
47
Cronofy
5
7
3
15
Recall.ai
4
5
3
12
Unipile
5
4
3
12
Aurinko
5
2
3
10
AgentMail
4
5
0
9
Meeting BaaS
4
4
1
9
Cal.com
2
3
3
8
Nylas leads by more than 3×
47 mentions against 15 for Cronofy, the nearest tracked vendor. No other tracked vendor appears across more than three product lines. Nylas appears across all five, and is the only vendor named on every platform for every line except Notetaker.
Recall.ai leads where it appears
Recall.ai is named on 12 answers to Nylas’s 47. Eleven of those twelve are meeting queries — the twelfth is the HIPAA and SOC 2 question on Claude. It leads Nylas on all four ChatGPT meeting-recording queries tested, and on Claude’s meeting-bot question it is named first while Nylas is seventh. Low volume, heavily concentrated.
Section 4

AI Positioning Audit

20 buyer-intent queries — click any row to see the exact question

Each query was written from the perspective of a real decision-maker evaluating communications infrastructure for their product — the engineering and product leaders who decide whether to build email, calendar and meeting integration in-house or buy it. These are the buyers whose AI answers determine whether Nylas enters the shortlist at all.

Target Buyer Sector CTOs, VP Engineering & Product leaders at CRM, healthcare, legal and AI agent software companies adding email, calendar and meeting features to their products
Buyer Personas
3 verified LinkedIn decision-makers behind these 20 queries
DF
CTO & Cofounder
Wealthbox • Financial CRM • Rhode Island
7queries
Pain Points
Advisors are split across Microsoft 365 and Google, and every client message has to be logged and retained. Email sync is the most-requested feature in his category, and maintaining separate provider integrations competes directly with the 150+ integrations his platform is judged on.
“gmail outlook exchange one API”“cost to build email sync”
Queries 1–4, 7, 18–19 Profile
AB
Co-founder & President
Luma Health • Healthcare • Texas
6queries
Pain Points
Clinical staff sit on mixed calendar systems, rooms and imaging equipment have no provider accounts of their own, and every integration has to clear HIPAA with a signed BAA before it ships. Patient access breaks at the scheduling layer more often than anywhere else.
“HIPAA compliant calendar API”“booking across Google and Microsoft”
Queries 5, 8–9, 11–12, 17 Profile
SK
President
Vela (YC W26) • AI Agents • San Francisco
7queries
Pain Points
His agents negotiate and book meetings across candidates and client firms at volume. That needs an inbox and a calendar under one identity the software owns — plus send quotas and allowlists, because a misbehaving agent can burn the sending domain in a day.
“give AI agent its own inbox”“agent email and calendar one API”
Queries 6, 10, 13–16, 20 Profile
# Query Topic Product Line Claude ChatGPT Gemini
1 Multi-provider email sync Email & Contacts ✓ #1 ✓ #1 ✓ #1
Exact question asked across all three AI platforms:

“Our SaaS platform needs to sync customer emails from Gmail, Outlook, and older on-premise Exchange servers into our app. Maintaining three separate integrations is eating our roadmap. What options exist to handle all of them through a single integration?”

2 Two-way email sync for CRM Email & Contacts ✓ #2 ✓ #1 ✓ #1
Exact question asked across all three AI platforms:

“We’re a CRM company and our users want their full email history with each contact logged automatically inside our product. What’s the most reliable way to add two-way email sync without building it from scratch?”

3 Email plus LinkedIn and WhatsApp Email & Contacts ✓ #2 ✓ #2 ✗
Exact question asked across all three AI platforms:

“Our sales platform needs to pull conversations from email plus LinkedIn and WhatsApp into one thread view. Are there APIs that unify all of those channels?”

4 Build-versus-buy economics Email & Contacts ✓ #2 ✓ #1 ✗
Exact question asked across all three AI platforms:

“How long does it realistically take an engineering team to build production-grade Gmail and Microsoft 365 email integration themselves, including OAuth verification, and is buying an API actually cheaper?”

5 Cross-provider clinician availability Calendar & Scheduler ✓ #2 ✓ #2 ✓ #1
Exact question asked across all three AI platforms:

“We’re adding appointment booking to our healthcare software. Clinicians are split across Google Workspace and Microsoft 365. What’s the best way to read real-time availability across both?”

6 Cross-company interview scheduling Calendar & Scheduler ✗ ✓ #2 ✓ #3
Exact question asked across all three AI platforms:

“Our applicant tracking system needs to coordinate interview scheduling across multiple interviewers who work at different organizations. Which scheduling infrastructure handles cross-company availability best?”

7 Embeddable booking widget Calendar & Scheduler ✓ #1 ✓ #1 ✓ #2
Exact question asked across all three AI platforms:

“We want an embeddable, white-label booking widget inside our own product rather than sending users to an external scheduling page. What developer tools offer that?”

8 Rooms and equipment without accounts Calendar & Scheduler ✓ #1 ✓ #2 ✗
Exact question asked across all three AI platforms:

“We need to book shared resources like meeting rooms and equipment that don’t have their own Google or Microsoft accounts. How do teams handle calendar resources with no provider account?”

9 Meeting bot reliability Notetaker ✓ #7 ✗ ✗
Exact question asked across all three AI platforms:

“We’re building a sales coaching product and need a bot that joins Zoom, Google Meet, and Teams calls to record and transcribe them. Which meeting bot APIs are most reliable?”

10 Calendar-triggered auto-join Notetaker ✓ #3 ✗ ✓ #2
Exact question asked across all three AI platforms:

“We want our meeting recorder to automatically join the right calls based on our users’ calendars, without them manually inviting a bot each time. Is there an API that connects calendar sync and meeting recording together?”

11 Structured post-call data Notetaker ✓ #2 ✗ ✗
Exact question asked across all three AI platforms:

“After a call ends we need speaker-labeled transcripts, a summary, and action items delivered to our backend, not just a raw recording. Which APIs return structured meeting data out of the box?”

12 Meeting recording cost at scale Notetaker ✓ #5 ✗ ✗
Exact question asked across all three AI platforms:

“What should we budget for meeting recording at scale? How do meeting bot APIs price per hour and what drives the cost up?”

13 Agent-owned mailbox Agent Accounts ✓ #2 ✓ #1 ✓ #1
Exact question asked across all three AI platforms:

“We’re deploying AI agents that need to send and receive email as their own identity, not from a team member’s inbox. What infrastructure exists to give an agent its own mailbox?”

14 Inbox and calendar, one identity Agent Accounts ✓ #1 ✓ #1 ✓ #1
Exact question asked across all three AI platforms:

“Our AI scheduling assistant needs both an inbox and a calendar under the same identity so it can email prospects and book the meeting itself. Are there platforms that provision both in one call?”

15 Programmatic inbox provisioning Agent Accounts ✓ #2 ✓ #2 ✓ #1
Exact question asked across all three AI platforms:

“What are the options for programmatically creating email inboxes for AI agents, and how do they compare on pricing and message threading?”

16 Agent deliverability controls Agent Accounts ✓ #1 ✓ #1 ✗
Exact question asked across all three AI platforms:

“How do teams stop an autonomous email agent from damaging their sending domain reputation? What rate limiting, allowlists, and deliverability controls should the email infrastructure provide?”

17 HIPAA and SOC 2 Type II Platform & Compliance ✓ #2 ✓ #1 ✓ #1
Exact question asked across all three AI platforms:

“We handle patient data and need email and calendar integration that is HIPAA compliant with SOC 2 Type II. Which communication API vendors meet that bar?”

18 Shipping without CASA verification Platform & Compliance ✓ #2 ✓ #1 ✓ #1
Exact question asked across all three AI platforms:

“Getting Google’s CASA security verification for restricted Gmail scopes has stalled our launch. Is there a way to ship email integration without going through that process ourselves?”

19 Provider-direct versus third-party Platform & Compliance ✓ #1 ✓ #1 ✗
Exact question asked across all three AI platforms:

“Our users are mostly on Google Workspace. Is it worth paying for a third-party email API, or should we just use the provider’s own API directly?”

20 MCP server for email and calendar Platform & Compliance ✓ #1 ✓ #1 ✓ #2
Exact question asked across all three AI platforms:

“We want our AI coding agents to access users’ email and calendar through MCP. Which communication platforms ship an MCP server that works with tools like Claude Code and Cursor?”

TOTAL 19/20 · 7×#1 16/20 · 11×#1 12/20 · 8×#1

ⓘ ✓ #n shows the position at which Nylas was named in that response — #1 means it was the first provider named. First placements total 26 across the three platforms (7 Claude, 11 ChatGPT, 8 Gemini).

Section 5

The Notetaker Blackout

The ChatGPT Notetaker blackout in the tested four-query meeting subset

Nylas is mentioned on 75% or more of the queries in four of its five product lines across every platform tested. Notetaker returns 5 of 12. Those four meeting-recording queries are also the only four ChatGPT missed — every other question on that platform named Nylas. On all four ChatGPT meeting-recording queries tested, Recall.ai and Meeting BaaS are named instead.

This is not a missing-product problem. Nylas publicly documents Notetaker across Zoom, Microsoft Teams and Google Meet, including recordings, speaker-labelled transcripts, summaries and action items. The capability exists and is documented. What this audit measures is whether it is named when a buyer asks.

“We’re building a sales coaching product and need a bot that joins Zoom, Google Meet, and Teams calls to record and transcribe them. Which meeting bot APIs are most reliable?”

— Query 9. ChatGPT and Gemini omit Nylas entirely. Claude names it seventh, behind six meeting-bot vendors. The weakest question in the audit.

“After a call ends we need speaker-labeled transcripts, a summary, and action items delivered to our backend, not just a raw recording. Which APIs return structured meeting data out of the box?”

— Query 11. Notetaker returns exactly this — transcripts, summaries and action items by webhook. Only Claude names it. ChatGPT and Gemini route to Recall.ai and transcription vendors.

“We need to book shared resources like meeting rooms and equipment that don’t have their own Google or Microsoft accounts. How do teams handle calendar resources with no provider account?”

— Query 8. Describes Virtual Calendars almost word for word. Gemini answers with native Google and Microsoft resource mailboxes and never names Nylas.
ChatGPT: 0 of 4 meeting-recording queries tested
Queries 9, 10, 11 and 12 are ChatGPT’s only four misses across the 20 queries. Every other question on that platform named Nylas, eleven of them first. Within this query set the gap is specific rather than general.
Claude names it — but names it late
Claude surfaces Notetaker on all four meeting queries, while ChatGPT surfaces it on none. The audit does not determine why the platforms differ. Where Claude does name it, the average position is 4.3, against 1.5 across its other fifteen answers. On query 9 it appears seventh, behind Recall.ai, Meeting BaaS, Skribby, MeetStream, Attendee and Vexa.
Same Platform. Same Product. Different Answer.

Agent Accounts reached general availability more recently than Notetaker, and it scores 11 of 12. Notetaker scores 5 of 12. Recency alone does not explain the gap. Within this audit, agent-account answers are more fragmented, while meeting-recording answers repeatedly favour Recall.ai. The pattern is consistent with stronger incumbent associations in meeting intelligence, but this audit does not isolate the cause. What it does show is that having current public documentation does not guarantee recommendation visibility.

Flagged for verification: an inaccurate acquisition claim surfaced during testing

One session surfaced an unverified claim, attributed to a third-party comparison page, that Nylas was acquired by Rippling in 2024. We did not find corroborating evidence in the materials reviewed. It did not affect scoring — no answer repeated it as fact.

Before outreach, identify the source page and verify its current wording. This entry records what surfaced during testing; it is not an independent audit of that page.

What this points to

Nylas already has substantial Notetaker documentation. The gap is that this capability is not translating into ChatGPT recommendation visibility on the four meeting queries tested. Retrieval, category association and third-party authority are the next hypotheses to test — this audit measures the outcome, not the mechanism.

Section 6

AI Topic Authority Map

Query heatmap — product line × platform

Topic Named First on Most Platforms Nylas Coverage
Multi-provider email sync incl. Exchange Nylas UNANIMOUS (3/3)
Two-way email sync for CRM Nylas UNANIMOUS (3/3)
Email plus LinkedIn and WhatsApp Unipile 2 of 3 platforms
Cross-provider availability Cronofy UNANIMOUS (3/3)
Cross-company interview scheduling Cronofy 2 of 3 platforms
Embeddable booking widget Nylas UNANIMOUS (3/3)
Resource calendars, no provider account Split — Nylas, Cronofy, native 2 of 3 platforms
Meeting bot reliability Recall.ai Claude only (1/3)
Structured post-call data Recall.ai Claude only (1/3)
Meeting recording cost at scale Recall.ai Claude only (1/3)
Agent-owned mailbox Nylas UNANIMOUS (3/3)
Inbox and calendar, one identity Nylas UNANIMOUS (3/3)
HIPAA and SOC 2 Type II Nylas UNANIMOUS (3/3)
Shipping without CASA verification Nylas UNANIMOUS (3/3)
MCP server for email and calendar Nylas / Unipile UNANIMOUS (3/3)
Product Line
ChatGPT
Claude
Gemini
Email & Contacts API
4 queries
100%
100%
50%
Calendar API & Scheduler
4 queries
100%
75%
75%
Notetaker API
4 queries
0%
100%
25%
Agent Accounts
4 queries
100%
100%
75%
Platform, Security & Compliance
4 queries
100%
100%
75%

▹ Notetaker is the only Nylas product line with 0% visibility on one tested platform and only 25% on another.

Email & Contacts API • 4 queries
ChatGPT100%
Claude100%
Gemini50%
Calendar API & Scheduler • 4 queries
ChatGPT100%
Claude75%
Gemini75%
Notetaker API • 4 queries
ChatGPT0%
Claude100%
Gemini25%
Agent Accounts • 4 queries
ChatGPT100%
Claude100%
Gemini75%
Platform, Security & Compliance • 4 queries
ChatGPT100%
Claude100%
Gemini75%
Four Product Lines Clear 75% Overall Coverage
Measured across each line’s 12 platform-query opportunities: Email & Contacts and Calendar & Scheduler both return 10 of 12, Agent Accounts and Platform & Compliance both return 11 of 12 — the strongest lines in the audit. Individual platform rates vary within each line; Gemini’s 50% on Email & Contacts is the lowest of the four.
Notetaker API: 0% on ChatGPT, 25% on Gemini
The only line with a zero. Recall.ai and Meeting BaaS take the queries on both platforms. Claude's 100% shows the product is reachable — the shortfall is concentrated in where these two platforms look for meeting-bot answers.
Section 7

Methodology

How we conducted this Xtrusio AEO/GEO Audit

20-Query Buyer-Intent Testing
Twenty decision-maker questions run across ChatGPT, Claude and Gemini — 60 scored responses. Questions mirror how CTOs and product leaders research communications infrastructure during discovery, before any vendor is named. Eight test Nylas’s core differentiators, seven test shared territory, five test categories where competitors are known to be stronger.
Competitor Scope
Cronofy (scheduling infrastructure), Recall.ai and Meeting BaaS (meeting bots), AgentMail (agent mailboxes), Unipile and Aurinko (unified communication APIs), Cal.com (booking platform). Each competes for a slice of the same buyer during discovery. Vendors named in only one session and not corroborated elsewhere were excluded from all totals.
How Mentions and Placements Are Counted
A mention is counted when Nylas is explicitly named as a provider or recommended option in the scored response. A first placement means Nylas is the first named provider in that response’s recommendation or list. No source URLs are tracked, so this measures naming, not linking. Vendor totals cover all three platforms and all 60 responses; vendors named in a single session without corroboration elsewhere are excluded from every total, as are first-party provider APIs, which appear as integration targets rather than as vendors competing for the purchase.
Test Conditions
Window: September 2026
Queries: 20 per platform, run once
Scored responses: 60
Sessions: 3 — one per platform
ChatGPT: conversational, one question per turn
Claude: research mode, retrieval enabled
Gemini: custom-configured assistant

These figures are a point-in-time snapshot. Generative outputs are non-deterministic and may vary on rerun. Session formats differ by platform, so cross-platform percentages should be read as directional rather than as a like-for-like benchmark. A verification run of ChatGPT in a deeper research format returned a higher mention count on the same 20 questions; it is excluded from every figure in this report, which uses the conversational session throughout.

Section 8

Recommendations

Amplifying documentation that already exists, rather than creating it

Phase 1 — 0–30 Days
Convert Existing Notetaker Documentation into Comparison-Ready Content
  • Start from what already exists. Notetaker is documented across Zoom, Teams and Google Meet with recordings, transcripts, summaries and action items — this is a discovery problem, not a capability gap.
  • Rewrite that material in the buyer language of queries 9 to 12: join reliability, speaker labelling, structured output and cost at scale, framed against named alternatives rather than as feature documentation.
  • Prioritise relevant third-party meeting-bot roundups where Recall.ai and Meeting BaaS are already represented, then re-test whether Nylas visibility improves on the same four queries.
  • Identify the competitor comparison page carrying the Rippling acquisition claim, confirm its current wording, and request a correction.
Phase 2 — 30–90 Days
Refresh the Existing Comparison and Virtual Calendars Assets
  • Four of Gemini’s eight misses named no vendor at all. Queries 4, 12, 16 and 19 were answered generically — unclaimed territory rather than ground lost to a competitor.
  • Strengthen and optimise the existing Virtual Calendars pages around the exact buyer phrasing in query 8 — resources without their own provider account. Do not create duplicate pages.
  • Refresh and expand the existing Gmail API and Microsoft Graph comparison guides with costed build-versus-buy figures, including CASA verification time. Consolidate rather than publish from zero.
  • Surface the already-documented agent guardrails — send limits, rules, lists and spam controls — more prominently on the Agent Accounts product pages, and link them from buyer-facing security content.
Phase 3 — 90+ Days
Consolidate Agent Accounts Authority While the Category Is Still Forming
  • Agent Accounts returns 11 of 12 and takes the sole vendor mention on the inbox-plus-calendar question. This is authority amplification, not capability creation — real mailboxes, calendar identity, MCP support and send controls are already documented.
  • Build buyer-facing content around a differentiated Nylas capability: a first-class agent email and calendar identity alongside access to existing user accounts, backed by Nylas’s documented SOC 2 Type II and HIPAA compliance posture, including BAA availability where required.
  • Quarterly Xtrusio re‑audits on the same 20 queries to track whether Notetaker visibility moves and whether the agent position holds as challengers publish.
Continuous AI Visibility Tracking
Brands can improve their AI discovery using generative engine optimization tools like Xtrusio.

One product line is weakening your AI shortlist visibility.

Let’s get Notetaker named where buyers are already asking.

This research report was generated using the Xtrusio Company Intelligence Module.