Question 1

What exactly is Publive AXP?

Accepted Answer

Publive AXP is an AI-visibility platform that makes your brand findable and citable inside AI answers: ChatGPT, Google AI Overviews, Perplexity and Gemini. It has four products: AXP Edge makes your existing pages readable to AI, Brand AI Observability gives you prompt tracking and insights across ChatGPT, Gemini, Claude and Google AI Overviews, against your competitors, AI Streams publishes new, AI-first content built to be cited, and AXP ClaimGraph keeps every claim across your estate consistent by finding every page, PDF and AI answer that disagrees with your official record.

Question 2

How is AXP Edge different from AI Streams?

Accepted Answer

AXP Edge makes the pages you already have legible and extractable to AI agents, live in under two hours with no rebuild. AI Streams builds new, governed, schema-rich content on your own domain to expand what you can be cited for. Streams is the offence, Edge is the eligibility layer; most brands run both.

Question 3

Will this hurt my existing SEO or Google rankings?

Accepted Answer

No. Your customers and search engines keep seeing your site exactly as it is today. AXP Edge only changes what AI agents receive, and AI Streams pages are additive and server-side rendered. Your existing pages and rankings are untouched.

Question 4

How do you measure success?

Accepted Answer

Two numbers agreed with you up front: Citations % (the share of your target prompts that return your brand in an AI-cited answer) and Share of Voice (your citations versus named competitors on the same query set). Your baseline is captured before we start.

Question 5

How long until we see results?

Accepted Answer

AXP Edge starts changing what AI agents fetch within days. For AI Streams, pages get discovered in the first weeks, with citation movement building over months 2–6 and a formal benchmark review at month 6.

Question 6

Do we own the content, and is there lock-in?

Accepted Answer

You own everything. Every AI Streams page is work-for-hire and yours to keep, along with the Brand Knowledge Hub. If you ever stop, the content and pages remain with you. No lock-in.

Question 7

Which industries is Publive AXP for?

Accepted Answer

Any category where discovery, comparison or trust now runs through AI answers: B2B SaaS, financial services, hospitality, education, consumer durables, automobiles, AlcoBev, and agencies delivering GEO for clients.

Question 8

How do we get started?

Accepted Answer

Start with a 2-day Audit of five prompts your buyers are asking AI (we show you who’s being cited in your place), or book a pilot. Talk to us and we’ll map the fastest path to being cited.

Question 9

What exactly is AXP Edge?

Accepted Answer

AXP Edge makes your existing pages readable to AI systems — without changing your website. When an AI agent (ChatGPT’s crawler, Perplexity, Claude, Google’s AI systems) requests one of your pages, a small function running in your own CDN recognises it and serves a clean, structured, fully-rendered version of that same page. Human visitors and Google’s SEO crawlers are completely untouched — they get exactly what your site serves today.

The problem it solves: most AI crawlers don’t execute JavaScript, and most enterprise pages bury their content under markup, scripts and page chrome. Our audit of 951 Indian enterprise brands found that for the median brand, only 16% of what an AI agent downloads is actual content — the rest is markup and script, not information. AXP Edge fixes that at the delivery layer, in real time, with no re-architecture of your site.

Question 10

How is AXP Edge different from AI Streams?

Accepted Answer

AI Streams creates net-new content — a dedicated section on your domain publishing pages built to win citations for the questions your buyers ask. AXP Edge improves your existing pages — it makes the pages you already have legible and extractable to AI agents.

They solve two halves of the same problem: Streams is the offence (new citable assets), Edge is the eligibility layer (making sure everything you already publish can actually be fetched, parsed and used). Most brands ultimately need both; Edge is often the faster first step because it requires no content production.

Question 11

Why serve this at my own URL? Why not a separate “AI version” of my site?

Accepted Answer

Because AI agents fetch your URLs. When a live AI system answers a question, it fetches the URL in its index or the one a user pasted — www.yourbrand.com/page. A parallel “AI site” at a different address may simply never be fetched by the systems that matter. AXP Edge meets the agent where it already goes: your canonical URL, served better.

This also means you never have a second public surface representing your brand. There’s no shadow site, no duplicate URL, no alternate domain — just your pages, delivered in the encoding each audience needs.

Question 12

How do you decide which visitors get the agent view?

Accepted Answer

By user agent. The edge function checks the requester’s declared user agent against a configured list of known AI agents — GPTBot, OAI-SearchBot, ChatGPT-User, PerplexityBot, ClaudeBot, Claude-User and others, maintained by us and updated as new agents appear. Matches on your targeted paths are routed to the optimised view; everything else — every human browser, every SEO crawler — passes straight through to your origin, byte-identical to before.

The list is configurable per client during onboarding: you choose which agent families to serve and which paths to include, and can start with a handful of pages before expanding.

We also record the connecting IP on every agent request, so verification that a claimed agent is genuine can be layered on later without you touching your edge config again. And worth saying plainly: if someone spoofed an AI agent’s user agent, they would receive the agent view — which contains nothing your public page doesn’t already say.

Question 13

We rank well on Google. Why aren’t we showing up in AI answers?

Accepted Answer

Because Google and AI crawlers are different machines with different capabilities. Googlebot renders JavaScript; the major AI crawlers — GPTBot, ClaudeBot, PerplexityBot — do not, as network-level research by Vercel and MERJ confirmed. So a page that ranks #1 on Google can be an empty shell to the systems assembling AI answers. Even where content is technically reachable, extraction matters: if an agent downloads a page that is only 16% content — the median for Indian enterprise brands in our audit — the relevant passage often doesn’t survive into the answer.

Your Google ranking is real equity; it just doesn’t transfer automatically. AXP Edge is how the pages that earned that ranking become equally legible to the AI layer — without touching what Google sees.

Question 14

Isn’t this cloaking? Won’t Google penalise us?

Accepted Answer

Cloaking, as search engines define it, is showing search crawlers different substance than users see — telling the crawler one thing to game rankings while the user gets another. AXP Edge does the opposite of that, with three hard guarantees:

1. Same substance, better encoding. The agent view only restructures and re-represents content that is already on your page. It never adds a claim your human page doesn’t make. This isn’t a policy we ask editors to follow — it is enforced in the generation engine itself: the system cannot emit content that isn’t present on, or faithfully derivable from, the source page.

2. SEO crawlers are untouched. Googlebot, Bingbot and every SEO crawler receive exactly what they received before — the routing applies only to AI agent user agents. Your rankings are built on responses we never touch.

3. Parity restoration, not divergence. Much of what the agent view “adds” is content your human visitors already see but AI agents can’t reach — JavaScript-rendered text that non-JS crawlers get an empty shell instead of. Recovering that is restoring parity between audiences, not creating asymmetry. A user who clicks through from an AI answer to your live page finds the same information the agent was served.

There is also a decade of precedent for serving crawlers a pre-rendered encoding of the same content. Google documented dynamic rendering — detecting crawler user agents and serving them pre-rendered HTML of the same page — as a legitimate workaround for JavaScript-heavy sites, and services built on that model have run it at massive scale for years without penalty, because the substance served is identical. AXP Edge applies the same pattern to AI agents, with a stricter boundary: we never serve generative summaries or net-new copy in the agent view, only extractive restructuring of what is already there.

Question 15

Will my customers ever see the agent view?

Accepted Answer

No. Human traffic never touches the AXP Edge path — the routing check happens in your CDN before anything else, and any request that isn’t a recognised AI agent passes through to your origin exactly as it does today. We verify this at setup with a simple test you can run yourself: a human-browser request returns your unchanged page with identical body and timing; only an agent-UA request carries our response header.

Question 16

Will this affect our existing SEO rankings?

Accepted Answer

No mechanism exists for it to. Your rankings are determined by what SEO crawlers and users receive — both are byte-identical to before AXP Edge, which is verifiable at any moment. There’s no new URL, no redirect, no change to your sitemap, canonical tags, internal linking or crawl budget. From Google’s perspective, nothing about your site has changed.

Question 17

Can we see exactly what an AI agent received?

Accepted Answer

Yes — this is a core dashboard surface, not an afterthought. For any URL, the Publive dashboard shows you precisely what an agent was served, side by side with your live human page, so you can verify the substance-parity guarantee yourself at any time. Because the agent view is never a public URL, the dashboard is the inspection surface — and it’s always available to your team, not just ours.

On top of that, every serving decision is logged: which agent, which URL, when, and which version of the view. For brands that need it, that audit trail is exportable.

Question 18

Is our content stored on Publive’s servers?

Accepted Answer

Here’s exactly what exists where. AXP Edge works only with content your website already serves publicly — the same pages any visitor or crawler can fetch. From those public pages we generate optimised agent views, and those generated views are cached on the Publive Edge Backend so they can be served within the latency budget. That cache is the only Publive-side copy: derived renderings of already-public pages, keyed per URL, refreshed when your content changes, and purged if you offboard.

What we never hold: your customer data, user PII, credentials, databases, CMS access, analytics data, or anything behind a login. The rendering step itself is data-in-transit — page content passes through the renderer and out; it isn’t warehoused there.

And nothing runs on your servers or inside your infrastructure beyond the small routing function in your own CDN account — which you deploy, you control, and you can remove at any time.

Question 19

Do you need access to our servers, CMS, or codebase?

Accepted Answer

No. The entire integration is: one edge function deployed in your own CDN account (Cloudflare Worker, CloudFront Function or Akamai EdgeWorker), one API key, one route binding. We never touch your origin servers, never get CMS credentials, never see your codebase, and never require a code deployment on your side. Optionally, a lightweight webhook from your CMS tells us when a page changes so the agent view refreshes immediately — that’s a one-way notification, not access.

Question 20

What data does AXP Edge collect?

Accepted Answer

Operational data about AI agent traffic to your pages: which agent fetched which URL, when, the response outcome, and the agent’s connecting IP. This is the same class of data already in your CDN logs — we surface it as the analytics you see in the dashboard: which AI systems visit you, what they fetch, and whether those fetches succeed. No human-visitor data flows through AXP Edge at all, because human requests never enter the routed path.

Question 21

We’re a regulated business (BFSI / Healthcare). Can our compliance team control what’s served?

Accepted Answer

The strongest compliance property is structural: the agent view is incapable of saying anything your page doesn’t. That constraint is enforced in the generation engine, not left to editorial policy — so there is no new claim for your legal or compliance team to review. What an agent receives is a restructured encoding of copy your team has already approved and published.

Around that, you keep operational control: you decide which paths and which agent families are covered; every serving decision is logged with the agent, URL, timestamp and view version; that audit trail is exportable; and the dashboard shows the exact agent view beside your live page for any URL. Publive is ISO 27001 and SOC 2 Type II certified. If your review needs to go further than this, talk to us and we’ll work through it with your team.

Question 22

Will it slow down my site?

Accepted Answer

Not for humans — human traffic never enters the AXP Edge path, so your site’s performance for visitors and for Core Web Vitals is unchanged by construction. For AI agents, responses are served from a pre-warmed cache with a target of under 500ms, which is typically faster than what agents get from your origin today: our audit found the median Indian enterprise brand takes 1.3 seconds to respond to an AI agent, against a healthy benchmark of half a second. So the usual outcome is that agents get a faster response than before, and humans notice nothing.

Question 23

What happens if Publive’s backend goes down?

Accepted Answer

Your site cannot be taken down by us — that’s a structural property, not an SLA promise. If our backend errors, times out or is unreachable, the edge function automatically fails over and serves your origin page directly, marked with a diagnostic header so it’s visible in monitoring. The worst case at any moment is your site exactly as it is today. Failover rate is tracked as a first-class dashboard metric, and Publive’s infrastructure runs at 99.99% uptime across 1B+ pageviews a month.

Question 24

My content changes frequently. How does the agent view stay fresh?

Accepted Answer

Three layers. First, we pre-generate views for your targeted pages from your sitemap, so agents never wait on a cold cache. Second — the important one — a webhook from your CMS invalidates and regenerates a page’s view the moment you change it, so freshness is event-driven, not timer-driven. Third, a short TTL backstop catches anything else.

Question 25

We publish new pages constantly. Do you cover URLs we didn’t list upfront?

Accepted Answer

Yes — coverage scales with your site automatically. You define the covered scope (all HTML pages, or an explicit path set), not a fixed URL list. When an agent requests a page inside that scope that hasn’t been generated yet, the system serves a safe immediate response — your origin page or the last good view — and generates the optimised view in the background; the next fetch gets the full agent view. Sitemap syncing keeps pre-warming ahead of demand, so in practice new pages are usually optimised before the first agent arrives.

Question 26

What does setup actually involve, and how long does it take?

Accepted Answer

For your team: deploy one small function in your CDN account, set two environment values (API key and target host), and bind it to the routes you want covered. That’s under two hours of actual work, and typically a same-day task once whoever manages your CDN picks it up.

We then verify together with three checks: an agent-UA request returns our header; a human-browser request is byte-identical to before; and a simulated backend failure falls back cleanly to your origin. You can start with a handful of pages and expand once you’re comfortable.

Question 27

Which CDNs do you support? What if we’re not on one of them?

Accepted Answer

Cloudflare, AWS CloudFront and Akamai — the same routing contract implemented natively on each. If you’re on a different CDN or serving directly from origin webservers, tell us your stack and we’ll scope it; the architecture is CDN-agnostic by design.

Question 28

What happens if we decide to discontinue AXP Edge?

Accepted Answer

AXP Edge is infrastructure — a delivery layer, not a content product that creates or owns anything. You turn it off, and your site is exactly as it was — within minutes, not weeks. Concretely:

Nothing to unwind. AXP Edge never modified your origin, your CMS, your content or your URLs. Removing the route binding (or using one-click rollback from the dashboard) returns 100% of traffic — agents included — to your origin. There is no migration, no redirect map, no cleanup project.

No SEO residue. Because the agent view never existed at a separate public URL, there’s nothing indexed to remove, no canonical to repair, no link equity to recover. Search engines never saw anything different in the first place.

Your content was always yours. The agent views were derived from your pages; on offboarding, cached views are purged from our backend and your analytics history is exportable before we close the account.

The honest trade-off, stated plainly: AI agents go back to receiving whatever your origin serves them. If that’s a JavaScript-dependent or markup-heavy page, the readability problem returns — which is precisely why the exit is painless. You’re removing a delivery layer, not extracting a dependency.

Question 29

Who controls what gets served to agents?

Accepted Answer

You do. You choose which agent families are covered and which paths are in scope, and you can start with a handful of pages and expand from there. Nothing outside the scope you set is ever routed.

What serves inside that scope is an automatic recovery and restructuring of your own published page — and the generation engine cannot emit content that isn’t present on, or faithfully derivable from, that source page. You can see exactly what any agent was served in the dashboard, side by side with your live page; every serving decision is logged and exportable; and you can switch the whole thing off from the dashboard at any time.

Question 30

Why shouldn’t we just build this internally? Our team could write that CDN worker in a week.

Accepted Answer

You could — the worker is the easy 10%, and we’ll happily show you ours. The routing function is roughly 100 lines of code. What it routes to is the product, and that’s a very different build:

The rendering pipeline. Headless JS-render recovery, DOM cleanup that improves passage extraction without destroying meaning, schema derivation — running on demand at agent-traffic volume, inside a sub-500ms serving budget. This is a rendering-infrastructure service, not a script.

The freshness system. Sitemap pre-warming, event-driven cache invalidation wired to your CMS, conflict handling when a page changes underneath a generated view. Get this wrong and agents are served stale content — worse than serving nothing.

The safety engine. The guarantee that the agent view can never say something your page doesn’t is enforced in the generation path. That constraint is your cloaking defence and your compliance posture; a quick internal build that lets anyone edit the agent view freely has neither.

The maintenance treadmill. The agent ecosystem changes monthly — new crawlers, renamed user agents, new fetch behaviours, changing verification schemes. We maintain that across every client; internally it becomes a permanent on-call obligation for a system nobody owns once the person who built it moves on.

The measurement loop. Serving a better view is half the product. Knowing whether it worked — verified agent traffic, fetch-survival rates, citation movement against your prompt universe — is the other half, and it’s what turns this from an infra experiment into a reportable business outcome.

The fair comparison isn’t “us versus a worker your team writes” — it’s “us versus an internal team building and permanently operating a rendering, caching, governance and measurement platform.” That’s the same reason you don’t build your own CDN: it’s possible, it’s just never the best use of your engineers. And if you ever want to stop paying us, you switch off and lose nothing — which is not true of most build-internal alternatives you’d be maintaining forever.

Question 31

Why not just re-architect our site to SSR and fix this properly at the source?

Accepted Answer

If you’re planning an SSR re-architecture anyway, do it — it’s the right long-term foundation, and AXP Edge doesn’t compete with it. The honest comparison is on three axes:

Cost and time. An in-house SSR rebuild for an enterprise JavaScript site routinely runs to multiple quarters of roadmap and touches your production frontend; industry estimates for in-house SSR start around $120K before ongoing maintenance. AXP Edge is a same-day edge deployment that touches nothing.

What it fixes. SSR fixes rendering — and only rendering. Our audit found that rendering invisibility and markup bloat are statistically independent problems: plenty of SSR sites still serve agents pages that are overwhelmingly markup, unlabelled by schema, with content that doesn’t survive extraction. AXP Edge addresses the full extraction path — render recovery, cleanup and schema re-representation.

Sequencing. They coexist. Deploy Edge now, capture the AI-visibility window while it’s forming, and if you complete an SSR migration later, Edge keeps adding the extraction and measurement layer on top — or you switch it off in minutes, having lost nothing.

Question 32

How will we know it’s working?

Accepted Answer

In sequence, because the effects compound. Immediately: header-verified proof that agents receive the optimised view and humans don’t. Within weeks: verified AI agent traffic to your pages, by agent and by URL, with fetch-survival rates — most brands see the extent of failed or empty agent fetches for the first time here. Over the engagement: citation movement on your tracked prompt universe, measured monthly across ChatGPT, Gemini, Claude and Google AI Overviews, with your unoptimised baseline as the comparison point.

And a commitment that cuts the other way: if our diagnostic shows your pages already fetch cleanly, we’ll tell you edge interception isn’t what you need — and point you at where your actual gap is. A vendor that can’t return that answer isn’t measuring honestly.

Question 33

Does making pages readable guarantee we get cited?

Accepted Answer

No — and be suspicious of anyone who guarantees citations. Readability is the eligibility layer: an AI system cannot cite what it cannot fetch and parse, so fixing fetch-survival removes a hard blocker. Whether you then win the citation depends on content relevance and authority against your competitors — which is what the rest of the AXP platform (prompt tracking, recommendations, AI Streams) works on. Edge gets you into the race; the programme wins it.

Question 34

What is Brand AI Observability?

Accepted Answer

Prompt tracking and insights across AI platforms. It shows you what AI says about your brand when buyers ask the questions that matter in your category. You choose (or accept suggested) prompts, we run them across ChatGPT, Gemini, Claude and Google AI Overviews, and you see whether you are mentioned, where you rank, how you are described, which competitors appear instead, and which sources each answer cites.

Question 35

Which AI platforms do you track?

Accepted Answer

Four today: ChatGPT, Gemini, Claude and Google AI Overviews. Every tracked prompt runs on all four, so you can see where a platform answers differently from the others. More platforms are being added, and new ones apply to your existing prompt set automatically.

Question 36

Where do prompt suggestions come from?

Accepted Answer

From your own pages, your category and your competitors. We look at the questions buyers ask AI about problems you solve, then flag the ones where a competitor is mentioned and you are not. You decide what to track: suggestions are proposals, never added without you.

Question 37

How is this different from Peec, Profound or Otterly?

Accepted Answer

Prompt tracking, competitor benchmarking and citation analysis are the shared baseline, and we cover them. The difference is what happens next. Monitoring tools hand your team a dashboard; Brand AI Observability sits in the same platform that changes what AI reads. A gap can become an AXP Edge fix on an existing page, a new AI Streams page for a question you do not answer, or a ClaimGraph finding when AI repeats something wrong, and the same prompts then show whether the answer moved.

Question 38

Why do different AI platforms give different answers?

Accepted Answer

Each platform retrieves from its own index, weighs sources differently and updates on its own schedule. A brand can lead in one and be missing from another for the same question. That is why we report per platform as well as in aggregate, rather than one blended score.

Question 39

How many prompts can we track?

Accepted Answer

The AXP Edge Growth plan includes 100 tracked prompts. Enterprise workspaces set custom prompt volumes. Each tracked prompt runs on every supported platform, twice a day, so 100 prompts means 800 AI answers analysed every day.

Question 40

How often is each prompt run?

Accepted Answer

Prompt tracking includes 2 runs per prompt, per AI platform, per day. AI answers vary from one run to the next, so two daily runs on each of ChatGPT, Gemini, Claude and Google AI Overviews give you a steadier read than a single snapshot, and catch a shift in the answer the same day it happens.

Question 41

How do we know a change actually worked?

Accepted Answer

Because delivery and measurement sit in one place. When a fix ships through AXP Edge or a page publishes through AI Streams, the prompts it targets are already being tracked, so you see visibility, position and share of voice before and after the change, on the same prompt set.

Question 42

Is this just AI-spun content that will get penalised?

Accepted Answer

No. Every page is built from your own Brand Knowledge Hub, fact-checked by agents, then human-reviewed and client-approved before it publishes. It’s original, governed content grounded in your approved sources, not scraped or spun. We optimise for being cited by answer engines, not for gaming a ranking algorithm.

Question 43

Will Google treat this as spam or hurt my existing SEO?

Accepted Answer

The opposite. Pages are server-side rendered, schema-rich and genuinely useful, which is what both search and AI reward. They live in a subdirectory on your own domain and are additive. Your existing pages and rankings are untouched.

Question 44

Who owns the content, and what happens if we leave?

Accepted Answer

You do. Every page is work-for-hire and yours to keep. If you ever stop, the content, the pages and the Brand Knowledge Hub remain with you. No lock-in.

Question 45

How is this different from a content agency or an AI writer?

Accepted Answer

Agencies give you words; AI writers give you volume. Neither owns the infrastructure (SSR, schema, crawler hygiene) or measures the outcome. Publive pairs agent-built, human-approved content with AI-ready infrastructure and reports Citations % and Share of Voice: the full rail from published page to cited answer.

Question 46

How do you decide what to write about?

Accepted Answer

The Prompt Curator Agent builds a query universe from the questions your buyers actually ask AI, ranked by citation gap, demand and competition. You approve the prompt set before anything is written, so effort goes where it moves your Share of Voice.

Question 47

How do you measure success, and when do we see it?

Accepted Answer

Baseline Citations % is captured up front. Pages typically get discovered and start attracting AI bot traffic within the first weeks, with citation movement building over months 2–6 and a formal benchmark review at month 6.

Question 48

Is our brand data safe, and is content on-brand and compliant?

Accepted Answer

The Brand Knowledge Hub is built only from sources you validate, and a Compliance QA agent checks voice, claims and schema before human review. Nothing is invented. Content only reorganises and represents what you’ve approved.

Question 49

How much work is this for my team?

Accepted Answer

Light. A Publive CSM runs onboarding and the workflow; your team’s main role is validating the Hub, approving the prompt set, and approving pages before publish.

Questions, answered

Let’s answer them on a quick call