Agency AI video workflow (review-link client approval)

The short answer

Client approval of AI video is a review link on one lot, not a Drive dump of 40 files. Dated 2026-09-21.

1. Lock the cut in the lot. A lot is an ordered list of shots. Each render of a shot is a take. You mount the take you will stand behind. The homepage says “One lot. Every shot accountable.”
2. Send one public URL. A review link is a public URL that shows one lot. No login, no account, no app. You send it. They watch it. They say what they think. The lesson is Review links.
3. Do not send the misses. The client is scoring the mounted cut, not a folder of every seed you already paid for. How you pick among comparable takes is its own guide: how to compare AI video takes.
4. Read standing, not the log. Verdicts are append-only. The answer is Held, Cleared, Unreviewed, or Clear. Then you fix one shot, remount, and send them back, or you export the same takes they approved.

lotgen.ai is a takes workspace, not a single-clip playground. The homepage puts it as “Dollars and seconds, not credits,” and every render is quoted in real dollars from your own provider keys before you commit. Pricing is “Two prices. No credits.” The seat is the studio. The labs bill the clip.

How to make the ad itself is a different guide: how to make UGC ads with AI. There is no run footage here, no ranked quality score, and no client portal with accounts. The review room is a token. The room is whoever holds it.

What “ai video review link client approval” means here

People search ai video review link client approval and agency ai video production workflow for two different jobs. This guide is about the first, and covers the second only as the approval half of an agency production: brief in, approved take out, via one shareable room. The production steps live elsewhere.

Here: one brief, several takes, one lot, one review link, named verdicts, an approved take. The client never logs into the studio.

Elsewhere on this site:

Four product words, from What a lot actually is:

Homepage, “Clients approve by link”: “A review room with no login. They see the cut, never your costs, providers or prompts.” Pricing, Solo: “Review links for clients, with no login for them.” Team: “Winners, review rounds and approvals with a name on them.”

The Academy lesson is more specific than the homepage line, and this guide follows the lesson for what is on the room: they see each shot’s mounted take, the shot name, the prompt, and the narration. They are never shown cost, provider, model, or your notes. Reviewers see each other’s names and verdicts. Written comments stay between them and you. The homepage says prompts are withheld and the review page prints them; the page the client opens is the one to believe.

A review link is not a Drive folder, not a WeTransfer zip, and not a Slack thread of MP4s. It is one token, one lot, one room.

Twelve generation keys this product pastes, the same list how much AI video generation costs uses: Google, BytePlus ARK, fal, evolink, Kling, OpenAI, MiniMax, Black Forest Labs, Alibaba, ElevenLabs, xAI and Runway. Luma and Sora are models on fal or evolink, not keys. BytePlus ARK is the console that issues the Seedance key, not the parent company. The invitation form also lists Anthropic, which writes copy and renders nothing, which is why the form says thirteen platforms.

What to lock before you send

The client should be scoring comparable takes of one brief, not a mixed bag of lengths, aspects, and silent clips next to ones with audio. How you lock a brief and pick a keep is how to compare AI video takes. What the approval path needs from it, in one short list:

1. Same brief. Prompt, billed duration, aspect, audio setting. Then you vary one thing: model, seed, or provider key. If two takes disagree on length or audio, the client is not comparing them. They are picking a different job.
2. Score usable, not pretty. A take that cannot ship (wrong length, silent when you asked for audio, product label unreadable, aspect the destination will not run) is a miss you already paid for. The math lives on how much AI video generation costs and the AI video cost calculator. There is no quality rubric to apply; usable is the bar.
3. Mount before you share. A review link shows each shot’s mounted take, and on a shot with no winner yet, that shot’s newest take. Those are not the same thing. That is how a client approves the variant you had rejected. The first successful take of a shot mounts itself. That is a one-take convenience. Overrule it.
4. Do not send an unfinished lot. A lot with an unrendered shot is not a lot to send. The review page will say the shot has no take, and the Approve button stays off. “Four of six approved” is not a delivery.

Compare is a studio job. The editor’s Compare tab is the side-by-side for one shot’s takes, from two takes up. The client is not that tab. If they need to pick among options you already stand behind, that is a different brief (several lots, or a later round), not a folder of misses.

The agency workflow

Brief, generate, lot, review link, comments, approved take. Production steps (UGC, product, talent) live on how to make UGC ads with AI. This section is the approval path.

1. Brief

Write the brief so a take can fail it. Duration, aspect, audio on or off, what the product must be readable as, who signs off. The person who will open the review link is part of the brief. If two people must both say yes, they will share one token. The room is whoever holds it.

2. Generate

Paste the keys you already pay for. Quote the take in dollars before it runs. What a render costs before it runs: two calls in this product spend money. Everything else, including asking what something will cost, is free. Re-render one shot, not the lot. Takes accumulate. Re-rendering does not overwrite. You paid for each take.

If the job is ads, start on the UGC how-to and come back here when you have a lot to send.

3. Lot

The lot is the cut. Homepage: “A lot is an ordered list of shots. Each renders on its own, every render is a take, and the winner is the one you keep.” Compare takes side by side in the lot. Mount the keep. Label the take if the shot has several; the export names files after the take label first.

4. Review link

From the lot, mint a review link. New links expire in 30 days and copy to the clipboard. The label (“Ambre Etoile, round 2”) is optional, and only you see it. It is how you tell three links on one lot apart when it is time to revoke one.

Give it an expiry. A link with none outlives the campaign it was made for. That is the thing to know before you send it, not after.

Anyone the link is forwarded to can watch and cast a verdict. A client can forward it to their own colleague without asking you for anything. Two links on the same lot are two different rooms. Neither reads the other’s traffic.

Revoking is not reversible. A revoked link, an expired link, and a token somebody invented all answer the same way to whoever holds them: not found. Verdicts already left stay on the lot.

5. Comments

The client opens the URL. No account. The page asks for a name before anyone can approve: “Needed before you can approve anything. The team has to know who said it.” Pricing’s Team line is the same rule with a name on the approval.

Per shot they can Approve, Request change, or send Just a note. Notes go straight back to the editors. The placeholder is “What would you change?” A comment with no verdict is neither held nor cleared: somebody looked and nobody said yes.

6. Approved take

Read the standing, not the log.

Two rules:

• Only a reviewer’s latest verdict counts.
• One outstanding change request holds a shot, however many approvals sit beside it. A client who asked for a change has not been overruled by a colleague who has not.

Unreviewed does not block, and it also is not done.

When they request a change: read what they said (the comment is the brief); check which take it was cast on (if the shot has been re-rendered since, the verdict is about a take that is no longer on show); fix the shot, not the lot; mount the new take deliberately if the shot already had a winner; send them back; re-read the standing. The old request stays in the log, superseded.

Once the standing is Clear, the export pack ships the same takes the client approved. Those are the same take on purpose. A client must never approve one clip while the delivery contains another.

What the client actually sees

This is the boundary. It is also why you do not send them a Drive dump.

What they see, from Review links and the review page:

• The studio name, not LOTGEN’s. The lot title. An optional label you set on the link (“first cut for the client”).
• Each shot’s mounted take. Shot number and the shot’s name. The prompt under the shot. Narration and captions built from the take’s measured length, not the length the shot asked for.
• A name field, required before Approve.
• Progress: how many reviewable shots they have acted on. A shot with no take does not count against them.
• Approve, Request change, Just a note.
• Other people in the same room: names and verdicts. Co-holders of the token are meant to see each other.
• If a shot has several takes, a folded versions list. The starred one is what the team mounted, and what an export would ship. The verdict applies to the mounted take. On a shot with no winner, the page says it is showing the newest.

That last point is why you mount before you send. The drawer is not a compare studio. It is a way for a client who asks “can we see the other version?” to look without emailing you. If you do not want them picking among misses, do not leave misses on the shot you send, and do not send a shot with no winner.

What they are never shown:

The homepage says they never see “prompts.” The Academy lesson and the review page print the prompt. Go by the page the client opens; there is no second, prompt-free client portal.

What you do not send them:

• A Drive or Frame.io dump of every seed.
• Cost screenshots, provider dashboards, or credit balances.
• Your editor notes.
• A lot that is not finished.
• A link with no expiry “just in case.”
• A workspace invitation. That is a token for a seat, not a review link.

There is no “Reviewed in LOTGEN” badge on a delivery. What there is: the export pack (mounted takes, files named after the shots, a manifest, a synthetic-content declaration), and the standing on the lot. It is not a signed C2PA claim.

Cost control for billable work

Agencies bill the keep, not the miss. Retries are dollars. Quote before the job runs. Cap the lot. Read spent, kept, and discarded on Usage. Do not screenshot a credit balance for a client who is supposed to be approving a cut.

Where that math lives:

• How much AI video generation costs: usable-shot math.
• AI UGC ads cost: cost per approved 15s variant.
• Best AI video generator for ads: which generator for ads.
• AI video cost calculator: your own spec.
• Kling vs Veo cost: dollars at a matched length, not a shootout.

Credits versus pay-as-you-go has its own page: AI video credits vs pay-as-you-go.

From pricing: “Two prices. No credits.” Solo $99. Team $90 / seat from 3, $79 from 6. Two seats is not a plan. MCP on every plan. You pay the labs for seconds on your own keys. You pay lotgen for the studio.

The review link is not a bill. Nothing is charged for a round.

Request access

lotgen.ai is an invite-only beta. The door is the invitation form.

Request an invitationSeats and pricing

Pick Agency or studio (“You make them for clients, and they have to approve.”) if that is who you are. If one of the asides above was yours, pick that tile instead:

• Brand or in-house team (“You make ads for your own product.”).
• Creator (“You post the output yourself.”).
• Developer or agent builder (“You want the API or MCP more than the studio.”).

The form does not create an account and it charges nothing. A person reads it. The providers you already pay for is the part that shapes the answer. If it is a fit, the invitation arrives by email and creates the workspace when you accept it.

Or start today: Solo is $99 a month and checkout creates the workspace. Same graph, same keys, no queue. Seats and pricing. What a second costs, and the retry multiplier: How much AI video generation costs. Your own spec: AI video cost calculator.

Related: Kling vs Veo cost, how to make UGC ads with AI, UGC ads cost, best AI video generator for ads, AI video MCP, one AI video API across the keys you already pay, review links, how to compare AI video takes, credits vs pay-as-you-go, AI ad creative testing at scale.

Checked on 2026-09-21: how to re-check yourself

The product facts on this page were read on 2026-09-21. Any of them can move, so here is where to look:

1. Homepage: “Dollars and seconds, not credits.” “Clients approve by link” says they never see prompts; Review links and the review page print the prompt.
2. What a lot is: lot, shot, take, mounted.
3. Pricing: “Two prices. No credits.” Solo $99. Team $90 / seat from 3, $79 from 6. MCP on every plan.
4. The invitation form: four tiles, and “thirteen platforms” because Anthropic is on the form; twelve of them generate media.
5. How much AI video generation costs, how to make UGC ads with AI, UGC ads cost, best AI video generator for ads, calculator, Kling vs Veo cost: every dollar lives on those pages, none is copied here.

Request an invitationSeats and pricingOne brief, several takes, one room.

FAQ