A leading product engineering company, creating adaptive software solutions to improve operations, providing businesses with expert development services from across domain.
A leading product engineering company, creating adaptive software solutions to improve operations, providing businesses with expert development services from across domain.
AI / B2B Marketplace
An AI-native platform for software company discovery and verified reputation — engineered end to end by Creuto with Claude and OpenAI.
~200K
Listings Seeded Before Launch
2
Independent Ownership-Verification Paths
2
AI Providers in Production
100%
Profiles Reviewed Before Going Live
At a glance
Product
Edverise
Category
B2B marketplace and trust infrastructure
Launched
Q3 2026
Services
Core Stack
Edverise is a discovery and reputation platform where businesses find software development companies, and where those companies build a verified public profile. Creuto conceived, designed and engineered the platform end to end, and took it to public launch with approximately 200,000 company listings already in place.
A directory is only as useful as its data, and only as trustworthy as its ownership model. Edverise had to solve both at once: populate a marketplace before it had a single registered user, then hand each seeded listing to the real company behind it — without creating duplicate profiles and without letting anyone claim a business that is not theirs.
Data acquisition and normalisation for approximately 200,000 company profiles
An AI content engine — Claude and OpenAI — that rewrites and structures every record into original content
An ownership-verification engine with DNS and email one-time-code paths, plus assisted recovery
A lifecycle and moderation back office where every profile moves from draft to published
A public web platform for discovery, company profiles, sign-up and claim flows
An AI customer-support layer running on the Claude API
The engineering thesis: let AI carry the volume work — writing, structuring and supporting at the scale of 200,000 profiles. Keep people accountable for the decisions that carry trust risk: who owns a company profile, and what goes live.
Marketplaces stall on the cold-start problem: buyers will not visit an empty directory, and companies will not join one with no buyers.
A marketplace cannot launch empty
An empty directory attracts neither buyers nor service companies. Solving that with data creates its own problem: source records vary in structure, tone and completeness, and cannot simply be republished.
→
No buyers without listings, no listings without buyers
→
Records arrive in inconsistent shapes with uneven completeness
→
Contact details are often a generic sales inbox, not a named address
Every seeded listing needs its real owner
A company that finds its own seeded listing would otherwise sign up again from scratch, creating duplicates. And if anyone could claim any listing, anyone could take over any company's profile.
→
Duplicate profiles erode the directory's usefulness
→
Impersonation erodes its trustworthiness
→
Some listings carry no usable email or website at all
Trust has to hold at 200,000-listing scale
An unreviewed profile going live erodes trust in the whole platform — yet claims, edits and support requests scale with the number of listings.
→
Every profile needs a review gate before it is public
→
Operational load grows with every claim and support request
→
Users need to hear the outcome of every decision
Solving cold start with data creates a chain of harder problems. Edverise answers each one directly in its architecture.
Challenge
Why it is hard
Engineering response
Why it is hard
An empty directory attracts neither buyers nor service companies.
Engineering response
Seed approximately 200,000 listings before public launch.
Why it is hard
Source data varies in structure, tone and completeness, and cannot simply be republished.
Engineering response
An AI content engine rewrites and structures every record into original Edverise content.
Why it is hard
A company that finds its own seeded listing would otherwise sign up again from scratch.
Engineering response
Claim-first onboarding: existing listings are claimed, not duplicated.
Why it is hard
If anyone could claim any listing, anyone could take over any company's profile.
Engineering response
Two independent verification paths: a DNS record on the domain, or a one-time code to the email on file.
Why it is hard
Some listings carry only a generic inbox, or no usable email or website at all.
Engineering response
Assisted recovery: support verifies manually and corrects the record, then the standard claim flow applies.
Why it is hard
An unreviewed profile going live erodes trust in the whole platform.
Engineering response
Every profile starts in draft mode; a super admin publishes it.
Why it is hard
Claims, edits and support requests scale with the number of listings.
Engineering response
Claude API customer support, plus automated email notification of every outcome.
Data flows upward from acquisition to the public platform. Each layer has one responsibility and a clear contract with the layers around it.
Built with Claude agentic development
Public Platform
Application Services
Ownership verification
DNS · email OTP
Lifecycle & moderation
Draft → Publish
Email notifications
Accept / reject
AI customer support
Claude API
AI Content Engine
Claude at runtimeData Acquisition & Normalisation
From publicly available service-provider listings
Marks where Claude is a runtime part of the product, not just part of how it was built.
The rail reflects that the entire system was engineered with Claude agentic development.
From raw listings to a moderated, claimable public profile — the seven pieces that make Edverise work.
At launch, Edverise seeded approximately 200,000 company profiles gathered from publicly available service-provider listings. Records of this kind arrive in inconsistent shapes: different fields, uneven completeness, and contact details such as a generic sales inbox rather than a named address.
The pipeline normalises them into consistent structured records for the layers above it, and deliberately preserves each listing's email and website — because those fields later become the anchor for ownership verification.
Every record is rewritten and structured by large language models — Claude and OpenAI — into original Edverise content: consistent company descriptions, standardised presentation, and cleaner profile fields, rather than a republication of source text.
The design goal is editorial consistency at volume, so that roughly 200,000 listings read as one platform rather than a patchwork of sources. The same engine drives ongoing optimisation of data, content and profiles as the platform grows.
Seeded listings are unowned until a real company claims them. The claim engine exists to guarantee one profile per company, and to make sure control goes only to someone who can prove it — either by placing a verification record on the company's own domain, or by entering a one-time code.
That code is sent to the email address already on file for the listing, not to the person claiming it. A founder who signs up with their own address receives the code at the company inbox on file; entering it correctly proves control of the company's own mailbox, and the profile is then assigned to the claimant.
Profiles move through controlled states. On registration or claim, a profile starts in draft mode and is not publicly visible. A super admin reviews it and marks it published in the admin panel; only then does the company receive a live public URL and appear on the platform.
Claim requests, support requests and publishing decisions each end in an accept-or-reject outcome, judged case by case, and users are notified of the result by email.
The public product is built with Next.js and React in TypeScript, on Node.js application services. TypeScript across the stack gives the team typed contracts between the interface, the services and the data flowing through the platform — a discipline that matters when several systems handle the same company records at scale.
Edverise uses the Claude API for customer support, handling the questions a claim-based platform naturally generates: how to claim a listing, how verification works, and what to do when a listing's contact data is incomplete.
Decisions with trust consequences — accepting a claim, publishing a profile, correcting a record — remain with the Edverise team.
Creuto's engineers built Edverise using Claude agentic development, with Claude working inside the engineering workflow to plan, write, run and iterate on code alongside the team.
It let a focused team deliver a multi-layer system — a public platform, a claim engine, a data pipeline and a moderation back office — as one coherent build rather than a set of disconnected efforts.
One profile per company — and control goes only to someone who can prove it.
STEP 01
The profile already exists, so there is nothing to duplicate.
STEP 02
Claim-first onboarding replaces a from-scratch sign-up.
STEP 03
Through either of two independent verification paths.
STEP 04
It stays private until a super admin reviews and publishes it.
Step 03 · Proves ownership — 2 independent paths
The company proves ownership by placing a verification record on its own domain.
A code is sent to the email already on file for the listing — not to the person claiming it. Entering it proves control of the company's own mailbox.
Assisted recovery: the company contacts support, the Edverise team verifies it manually and corrects the backend record — then the claim proceeds through the standard flow above.
Claude is part of Edverise at build time, at run time, and in how responsibility is divided between AI and people.
Build
Claude plans, writes, runs and iterates on code with Creuto's engineers across a multi-layer platform.
Operate
Claude powers customer support and, alongside OpenAI, the content engine that rewrites and optimises company profiles.
Govern
AI carries the volume work. Claim approvals, publishing and manual data corrections stay with the Edverise team.
Adding an AI feature is easy. Engineering a product where AI carries real production load — and people stay accountable for the decisions that matter — takes a different kind of build.
Claude works inside our engineering workflow to plan, write, run and iterate on code alongside the team — which is how a focused team delivered a multi-layer platform as one coherent build.
Edverise launched as a populated, structured marketplace from day one — engineered rather than grown organically from zero.
Ownership is proven, not asserted. Nothing goes live unreviewed. Those rules live in the verification engine and the lifecycle state machine, not in a policy document.
Models write, structure and support at the scale of 200,000 profiles. Accept-and-reject decisions on claims, support requests and publishing are made by the team, case by case.
A modern TypeScript stack across Next.js, React and Node.js, with Claude and OpenAI in production — built to scale with the platform.
Public discovery experience, company profiles, sign-up and claim flows
Application services: verification, lifecycle, notifications, admin
Rewriting, structuring and optimising company data, content and profiles
Support for claims, verification and profile questions
Planning, writing, running and iterating on code across the platform
Proving that a claimant controls the company
Publishing, moderation and user communication
The safeguards are part of how Edverise is built — not a layer added after launch.
Source records are rewritten into original Edverise content, not republished as-is.
Control of a profile requires a DNS record or access to the mailbox already on file for that listing.
Every profile starts in draft mode and is published by a super admin.
Accept and reject decisions on claims, support requests and publishing are made by the team, case by case.
Every outcome — claimed, published, accepted, rejected — is communicated by email.
6
Cooperating System Layers
1
Profile per Company, Enforced by Claim-First Onboarding
2 + 1
Verification Paths: DNS, Email OTP, Assisted Recovery
3
Roles for Claude: Build, Operate, Govern
A populated, structured marketplace from day one — listings live at public launch, not an empty directory.
A claim-first ownership model that prevents duplicate profiles and blocks impersonation.
A complete moderation pipeline — draft, review, publish — with email notification of every outcome.
Claude in three roles: build-time engineering, AI customer support, and content optimisation alongside OpenAI.
A modern TypeScript stack across Next.js, React and Node.js, built to scale with the platform.
With the platform live, Creuto is now tuning it against real usage: locating where friction is highest across claim verification, the draft-to-publish step and accept-or-reject decisions; formalising the criteria support applies when accepting or rejecting a claim or an edit; and defining how draft profiles are treated for discoverability before they are published.
We decide early which work belongs to models and which belongs to people. On Edverise, AI carries the volume — writing, structuring and supporting across roughly 200,000 profiles — while every decision that carries trust risk stays with a named human role. Claude is in our engineering workflow as well as in the product, so the same team ships the data pipeline, the claim engine, the back office and the public platform as one system.
Edverise is one of Creuto's flagship builds — conceived, designed and engineered end to end.
Everything you need to know about building a product like this with Creuto.
Creuto built a data acquisition and normalisation pipeline that gathered company profiles from publicly available service-provider listings and converted them into consistent structured records. An AI content engine using Claude and OpenAI then rewrote every record into original Edverise content, so the marketplace was populated before it had a single registered user.
Claiming requires proof of ownership through one of two independent paths. The company can place a DNS verification record on its own domain, or enter a one-time code that is sent to the email address already on file for the listing — not to the person making the claim. Either way, control goes only to someone who can prove they control the company's domain or mailbox.
The company contacts support. The Edverise team verifies the company manually and updates the backend record, after which the company goes through the standard claim flow. This assisted-recovery path exists because some seeded listings carry only a generic inbox or incomplete contact data.
In three places. At build time, Creuto's engineers used Claude agentic development to plan, write, run and iterate on code. At run time, the Claude API powers customer support, and Claude works alongside OpenAI in the content engine that rewrites and optimises company profiles. In governance, the system is designed so AI carries volume work while people keep the decisions that carry trust risk.
No. Every profile starts in draft mode and is published by a super admin after review. Accepting a claim, publishing a profile and correcting a record all remain with the Edverise team, judged case by case, and users are notified of every outcome by email.
The public platform is built with Next.js and React in TypeScript, on Node.js application services. Claude and OpenAI power the AI content engine, the Claude API powers customer support, and ownership verification uses DNS record checks and email one-time codes. A super admin panel and automated email notifications run the back office.
Whether you are launching a marketplace, a directory, or any product where AI has to carry real production load — Creuto engineers the data pipeline, the AI layer, the trust model and the public platform as one system.
Built with Claude. AI for volume. People for accountability. One team.
Book a free consultation11th Floor, O-Hub, Chandaka Industrial Estate, Infocity, Bhubaneswar, Odisha 751024
Level 4, 11 York Street Sydney Startup Hub Sydney, NSW – 2000
30 N. Đinh Nghệ, Phước Mỹ Sơn Trà, Đà Nẵng / Da Nang City – 550000
Level 25, AIDP Business Tower, Dubai Marina, United Arab Emirates
50 Beauchamp Street, Wellington, WGN 5028, New Zealand