Feature guide: AI & search
Natural-Language Search in DAM Software: A Buyer’s Guide
41 of the DAM platforms in our catalogue carry a natural-language search tag, but what that means in practice ranges from local AI search running on your own machine to a general library where free-text search is still an open feature request. Here's how to tell the difference before you buy.
Forty-one platforms in our 48-vendor catalogue carry the natural-language search tag, which tells you a vendor claims some form of free-text or AI-assisted search over your asset library — but it does not tell you how deep that capability actually goes. Reading across the twelve profiles we can detail here, the tag covers at least three different products: purpose-built AI search tools that run locally against a private photo archive, general-purpose DAMs where AI tagging is one feature among many, and file-sync platforms where any content search beyond the filename has to be bolted on with community add-ons.
The clearest split in this group is where the AI processing happens. IMatch and Excire both run locally, so a large photo archive never leaves the machine — IMatch adds offline face recognition for a metadata-heavy Windows library of 50,000-plus assets, while Excire is the profile that most directly names natural-language search, AI-assisted culling and face search as buy-if reasons, aimed at a photographer or small studio that doesn’t want to hand an archive to somebody else’s cloud. Nextcloud and ResourceSpace also support on-prem deployment, but neither ships a built-in AI search layer out of the box — Nextcloud leans on community apps like Memories and Recognize (or exiftool scripts) to bridge the gap, and its own feature requests for controlled vocabulary and dynamic tag-based smart collections are still open, one of them over nine years old.
Price transparency is the second axis worth checking before you assume a feature is included. Of the twelve vendors profiled here, exactly half — Razuna, Nextcloud, ImageKit, Cloudinary, Dash and IntelligenceBank — publish a price at all; the rest quote per customer or leave the tier named but not priced. That split matters more than usual on this feature, because AI search and tagging are often billed as an add-on rather than folded into a base tier, so an unpublished price is also an unpublished answer to whether the search feature you actually want is included.
The tag itself deserves a second look before you shortlist a vendor on it. Razuna is a useful example: it scores highest for value in this group, with a genuinely flat per-account price and no per-user fees, but the roster’s own dont_buy_if note for Razuna flags that its search leans on filename-based recall rather than free-text or content search across a large library — exactly the workflow this feature category is supposed to serve. Dash is a similar case in miniature: it’s sold on AI tagging, but its own dont_buy_if warns that reliable out-of-the-box tagging with minimal cleanup isn’t guaranteed, so budget time for correcting tags rather than assuming search works well from day one.
Cloudinary and ImageKit sit closer to developer infrastructure than to a browsable search UI — Cloudinary’s own dont_buy_if says as much, noting it suits teams that want media inside a delivery pipeline rather than a polished DAM browsing and search interface. Air, by contrast, is aimed squarely at a creative team that wants fast onboarding and a usable free tier, but neither its buy-if nor its common complaint make a specific claim about search depth, and its paid tiers are named without figures.
Put together, the practical lesson is to treat “natural-language search” as a category label, not a guarantee. Read the specific buy-if and dont-buy-if language for any vendor you shortlist, check whether the AI processing runs locally or in the vendor’s cloud if that matters for your library, and confirm the price actually includes the search tier you need rather than gating it behind a quote.
What decides it here
The real differences on this axis are where the AI processing runs (local/on-prem for IMatch, Excire, Nextcloud and ResourceSpace versus cloud-only for most of the rest), whether the vendor's own language actually promises free-text content search or just filename/tag recall, and whether the search or tagging feature is priced transparently or buried in a quote. A platform can carry this tag and still ship a thin, tag-based search baseline that needs add-ons or manual cleanup to become genuinely useful.
Watch out for
What surfaces after you buy
- Razuna's own dont_buy_if flags that it fits filename-based recall better than free-text/content search across a large library — don't assume the tag means deep AI search here.
- Nextcloud needs community apps (Memories, Recognize) or exiftool scripts to get AI-style tagging; controlled vocabulary and dynamic tag-based smart collections are still open feature requests, one over nine years old.
- Dash's dont_buy_if warns that reliable out-of-the-box AI tagging with minimal cleanup isn't guaranteed — budget time to correct tags before trusting search results.
- Excire's price isn't on its own product page; the common complaint notes it lives on the reseller's storefront, so get the actual figure before assuming it's the affordable local option.
- ImageKit meters several units at once (bandwidth, storage, video units, seats), and the dont_buy_if warns overage rates can erode the cost advantage if traffic is bursty or hard to forecast.
Before you shortlist
Work out about yourself first
- If your workflow depends on true free-text content search rather than filename or tag recall, check the vendor's dont_buy_if line specifically — some, like Razuna, admit that's not their strength.
- If assets can't leave your infrastructure, shortlist the on-prem/local options: IMatch, Excire, Nextcloud, ResourceSpace, or Razuna's on-prem enterprise tier.
- If budget certainty matters, start with the half of this group that publishes a price at all — Razuna, Nextcloud, ImageKit, Cloudinary, Dash, IntelligenceBank — rather than platforms that gate the search tier behind a quote.
- If AI tagging is central to your evaluation, pilot with real assets before buying — even vendors sold on AI tagging, like Dash, flag that out-of-the-box results need cleanup.
- Match the deployment array to your actual requirement: several vendors list both cloud and on-prem, but check which tier each mode is available in (Razuna's on-prem, for example, is enterprise-only and unpriced).
How this list was ranked
These criteria were written before any platform was placed. Each one is checkable on the vendor’s own pages or in what users report, so the ordering can be argued with rather than taken on trust.
| Criterion | Why it matters here | Weight |
|---|---|---|
| Free-text search vs filename/tag recall | Several vendors carrying this tag are, by their own dont_buy_if language, still closer to filename or basic tag recall than true content search — this is the single biggest gap between the label and the reality. | high |
| Local or on-prem AI processing | IMatch and Excire process and search without cloud upload, which matters for sensitive or very large private libraries; most of the rest of this group are cloud-only. | high |
| Pricing transparency for the search/tagging tier | Only half of the profiled vendors publish a price at all, and AI search or tagging is often gated behind an add-on or a quote rather than included in the base tier. | medium |
| Tagging accuracy out of the box | Even vendors marketed on AI tagging, like Dash, carry a dont_buy_if warning that results need manual cleanup rather than working reliably from day one. | medium |
| Deployment flexibility | Cloud-only, on-prem-only and hybrid options exist side by side in this pool, so confirm which deployment mode actually ships the search feature you want and at what tier. | low |
Razuna
The cheapest entry point in this group: a flat per-account price with no per-user fees, a generous free tier (500GB/5 users), and both cloud and on-prem deployment. Good if the real need is a low-cost library with basic search rather than deep content search.
Price published
Yes
Billed on
Flat price per account by volume; the vendor states there are no per-user fees
Transparency
90%
Weak spot
On-premise is available only in the enterprise tier, whose price is not published
Not for you if: The roster's own dont_buy_if flags that Razuna's search leans on filename-based recall rather than free-text/content search across a large library, so verify this directly if natural-language search is the main requirement — and on-prem is enterprise-tier only, with that price unpublished.
Nextcloud
A natural extension for a team already running Nextcloud for file sync and collaboration, with community apps (Memories, Recognize) or exiftool scripts able to bridge basic AI-style tagging without buying a separate DAM. Deployment is flexible across on-prem and cloud.
Price published
Yes
Billed on
Subscription per user per year, cheaper at volume
Transparency
80%
Weak spot
It is a file platform with DAM features rather than a DAM: asset-level metadata and rights are thinner
Not for you if: The common complaint is that it's a file platform with DAM features rather than a DAM — asset-level metadata and rights are thinner, and dont_buy_if notes controlled vocabulary, batch metadata editing, and dynamic tag-based smart collections are still open feature requests, one over nine years old.
ImageKit
Fits a developer team that wants fast SDK integration, responsive support, and a predictable monthly bill for steady, moderate traffic, with published per-unit rates once usage exceeds the flat tier.
Price published
Yes
Billed on
Bandwidth, storage and user seats, with published per-unit rates above the limits
Transparency
70%
Weak spot
Several units are metered at once: bandwidth, storage, video units and seats
Not for you if: Several units are metered at once — bandwidth, storage, video units and seats — and the dont_buy_if warns that bursty or hard-to-forecast traffic can erode the cost advantage; built-in content moderation isn't included at that tier either.
IMatch
The pick for a large (50k+), metadata-heavy Windows library that wants offline, private AI processing — including face recognition without cloud upload — and prefers a one-time perpetual licence over a subscription.
Price published
No
Billed on
Perpetual licence, no subscription
Transparency
60%
Weak spot
Windows only, so it is out of the question for a team on Macs
Not for you if: Windows-only, which the common complaint says rules it out for Mac/Linux teams outright; it also demands real setup time and lacks built-in RAW processing or editing, so it's pure cataloging, not an all-in-one tool.
Excire
The most directly AI-search-focused profile in this group — its buy_if explicitly names AI-assisted culling, face search and natural-language search, running locally with no cloud upload and no recurring subscription.
Price published
No
Billed on
One-off purchase, no subscription
Transparency
60%
Weak spot
The amount is not on the product page — it lives on the reseller’s storefront
Not for you if: The price isn't published on Excire's own product page — the common complaint says it lives on the reseller's storefront — and dont_buy_if flags it lacks pixel-perfect geotag/location search, Lightroom-level UI polish, and requires budgeting for occasional paid major-version upgrades.
Air
Suits a photo/image-heavy creative team that wants fast onboarding and a genuinely usable, credits-based free tier to trial the full product before committing, with no per-user billing.
Price published
No
Billed on
Credits; users are not billed
Transparency
60%
Weak spot
The free tier is described in detail while the paid tiers are named only
Not for you if: The common complaint is that the free tier is documented in detail while paid tiers are named only, and dont_buy_if flags reported per-tier price increases plus a weaker fit for video-heavy pipelines needing tight desktop-sync reliability.
Dash
Fits a small-to-mid ecommerce or DTC brand wanting fast-to-adopt AI tagging, guest uploads and Shopify-adjacent workflow, billed as a flat per-account price with unlimited users.
Price published
Yes
Billed on
Flat price per account; users are unlimited, tier set by storage and downloads
Transparency
60%
Weak spot
The tier is capped by download count as well as volume, which is a less familiar unit to budget against
Not for you if: Dont_buy_if warns that reliable out-of-the-box AI tagging with minimal cleanup isn't guaranteed, and the tier is capped by download count as well as storage volume — a less familiar unit to budget against, per the common complaint.
For a team with in-house server administration capability and a library of a few hundred thousand assets or fewer, the open-licence self-hosted route carries no licence fee at all.
Price published
No
Billed on
Self-hosted: no licence fee. Hosted: a user ceiling per tier
Transparency
60%
Weak spot
The free route is genuinely free of licence fees but moves hosting, upgrades and support onto you
Not for you if: The common complaint is that the free route moves hosting, upgrades and support onto the team; dont_buy_if warns it's not for a team that can't debug a broken database migration by hand, or a library approaching a million-plus assets that needs guaranteed support.
Also in this category, not ranked here
These carry the same facet in our database but did not make the shortlist. They are listed rather than hidden: a shortlist means more when you can see what it was drawn from.
- 4ALLPORTAL
- Acquia DAM
- Adobe Experience Manager Assets
- Amplience
- Aprimo
- Brandfolder
- Bynder
- Canto
- Censhare
- Cloudinary
- Contentstack
- Frontify
- IntelligenceBank
- KeyShot Dock (formerly Digizuite)
- Kontainer
- Kontent.ai
- Lytho
- MediaValet
- Mylio
- NetX
- Nuxeo
- OpenAsset
- Orange Logic
- Papirfly
- PhotoShelter
- PhotoShelter Brands
- Pics.io
- Scaleflex
- Sitecore Content Hub
- Storyteq
- Tenovos
- Wedia
- WoodWing Assets
Frequently asked
How many DAM platforms in the catalogue support natural-language search?
Forty-one of the 67 vendors in our catalogue carry the natural-language search tag. That's a broad category label, though — it spans purpose-built local AI search tools like Excire and IMatch, general DAMs with an AI-tagging feature such as Dash, and file-sync platforms like Nextcloud where content search beyond the filename still needs community add-ons.
Does the natural-language search tag always mean true free-text content search?
No. Razuna carries the tag but its own dont_buy_if notes it fits filename-based recall better than free-text or content search across a large library. Read each vendor's buy_if and dont_buy_if language before assuming the tag means deep AI search — some vendors are stronger on price or deployment than on search depth itself.
Which vendors run AI search or processing without uploading assets to the cloud?
IMatch and Excire both process and search locally, so a private photo archive never leaves the machine — IMatch adds offline face recognition, Excire adds AI-assisted culling and natural-language search. Nextcloud and ResourceSpace also support on-prem deployment, though neither ships a built-in AI search layer without add-ons.
Which of these vendors publish their prices?
Six of the twelve profiled here publish a price: Razuna, Nextcloud, ImageKit, Cloudinary, Dash and IntelligenceBank. The other six — including IMatch, Excire and Air — quote per customer or leave named tiers unpriced, which matters here since AI search or tagging is often billed as a separate add-on.
Is AI tagging reliable straight out of the box?
Not guaranteed. Dash is sold on AI tagging for ecommerce libraries, but its own dont_buy_if warns that reliable out-of-the-box tagging with minimal cleanup isn't assured. Budget time to review and correct tags before trusting search results built on them, especially in the first weeks after adoption.
What's the most affordable way to get natural-language search in this group?
Razuna scores highest for value, with a flat per-account price, no per-user fees, and a 500GB/5-user free tier. But its dont_buy_if flags that it leans on filename-based recall rather than deep content search, so it's the affordable pick only if that trade-off fits your actual workflow.
Which platform fits a large Windows-based photo library best?
IMatch, according to its buy_if: it's built for a large (50,000-plus asset), metadata-heavy Windows library, with offline face recognition and no forced cloud upload, sold as a one-time perpetual licence. It's Windows-only, though, and the common complaint rules it out entirely for Mac or Linux teams.
What's the catch with Excire's pricing and search claims?
Excire's buy_if is the most direct natural-language-search claim in this group, but the price itself isn't on Excire's own product page — the common complaint says it lives on the reseller's storefront. Dont_buy_if also flags it lacks pixel-perfect geotag/location search and Lightroom-level UI polish.
Sources
- Pricing page publishes a free tier at 500 GB and $99/mo for unlimited users, stating no per-user fees and no hidden charges checked 2026-08-06 — Razuna — Razuna pricing page
- Pricing page publishes three subscription levels in euros per user per year with a reduction from 200 users, and states AGPL for self-installation checked 2026-08-06 — Nextcloud — Nextcloud pricing page
- Plans page publishes amounts for every tier and the per-unit rate above each limit, including a separate DAM-only line checked 2026-08-06 — ImageKit — ImageKit pricing page
- Product page states a perpetual licence with no subscription and a free trial with no registration; the amount is not shown on the page checked 2026-08-06 — IMatch — IMatch pricing page
- Product page states a one-off purchase with no subscription and local analysis (“your photos stay on your computer”); a trial exists, the amount is not shown checked 2026-08-06 — Excire — Excire pricing page
- Pricing page describes a credit-based free tier with unlimited users but publishes no amounts for the paid tiers checked 2026-08-06 — Air — Air pricing page
- Pricing page publishes four monthly and annual figures in pounds, unlimited users, and download limits per tier checked 2026-08-06 — Dash — Dash pricing page
- Pricing page states there is no annual software licence fee and lists four hosted tiers by user ceiling without amounts checked 2026-08-06 — ResourceSpace — ResourceSpace pricing page