Newform vs Atria
Atria has had a real moment in 2026. The Raya agent launched in February, trained on $5 billion in ad spend data, and the pitch is clean: AI watches your competitors, tags your creatives, surfaces insights, and drafts new ad concepts without you asking. For solo media buyers and small DTC teams, the value is obvious. You get an analyst-shaped tool at $129/mo.
Framework comes at the same problem from the other end. Instead of an autonomous agent generating ideas off historical data, Framework runs structured experiments on your actual ads, decides which ones scale or die at 95% confidence, and ships new variants through an embedded production team. The intelligence is grounded in what's working on your account this week, not in pattern-matching from a $5B dataset that may or may not look like your ICP.
Both can work. The question is which model fits your spend, your team, and your trust in AI agents. Below we cover both sides honestly.
TL;DR
- →Atria pitches AI doing the analyst work for you. Framework pitches statistical confidence and humans doing the strategy.
- →Raya, Atria's AI agent, is trained on $5B in ad spend. Framework's engine runs 95% CI experiments on your ads, not pattern-matching from someone else's.
- →Atria is self-serve SaaS at $129–$329/mo. Framework is an agency engagement at $25K–$80K/mo (creative team included).
- →Under $250K/mo in spend with a creative team already: Atria is the right shape. Above that with no team or thin testing: Framework.
At a glance
Newform Framework
Creative intelligence platform + embedded creative team
Atria
AI ad creative intelligence platform with the Raya autonomous strategist
Feature-by-feature
Where each one wins, honestly.
| Feature | Newform Framework | Atria | Edge |
|---|---|---|---|
| Statistical testing engine (95% CI calls) | Yes. Every cohort runs to significance, then auto-kills losers and graduates winners. | Partial. Atria's Radar engine analyzes performance but doesn't run structured experiments at confidence thresholds. | ✓ |
| AI agent that works autonomously | Strategists work alongside the engine. No fully autonomous agent. | Yes. Raya monitors competitors, tags creatives, generates concepts weekly. | ✓ |
| Competitor ad library | Custom research workflows. No standing library. | Yes. 25M+ saved ads, tracked by brand. | ✓ |
| AI vision + language analysis on your ads | Yes. Auto-tags hook, format, talent, scene, copy. | Yes. Raya auto-tags hooks, personas, themes, USPs without manual input. | = |
| AI ad generation (scripts, image variations) | Humans script and produce, informed by the engine's findings. | Yes. Native scripts and image variations. | ✓ |
| Embedded creative production team | Yes. Production pod ships 50–100+ creatives/mo on the engine. | No. You write, shoot, and edit. Atria assists with AI variations. | ✓ |
| Auto-kill underperforming creative | Yes. Pauses losers automatically at significance. | No. Surfaces underperformers. You pause manually. | ✓ |
| Ad spend cap | No cap. Scales with your spend. | Core plan caps at $500K/mo ad spend. | ✓ |
| Self-serve onboarding | No. Sales-led, scoping call required. | Yes. Sign up and connect Meta the same day. | ✓ |
| Floor price | $25K/mo and up (includes team). | $129/mo billed annually on Core. | ✓ |
| Best for spend range | $250K/mo to $10M/mo. | Up to roughly $500K/mo on Core, more on Plus. | = |
Which one should I pick?
The honest decision tree.
- ✓You spend $250K+/mo and the bottleneck is shipping enough creative, not finding ideas.
- ✓You want statistical confidence before scaling a winner, not an AI agent's recommendation.
- ✓Your creative team is overwhelmed or non-existent, and you're tired of stitching freelancers together.
- ✓You need one partner accountable for the result, with a real testing engine under the hood.
- ✓You're skeptical of autonomous AI making budget moves on your account.
- ✓You have a creative team that ships volume already and you need an analyst-shaped layer on top.
- ✓You spend under $500K/mo and the math on a $25K+/mo agency engagement doesn't pencil.
- ✓You want autonomous AI surfacing competitor moves and drafting concepts in the background.
- ✓You value a deep competitor ad library more than the testing engine.
- ✓Self-serve SaaS pricing is non-negotiable.
The verdict
What we'd tell a friend.
Atria is a strong tool. Raya is the most genuinely useful AI agent in the category right now, and the price is hard to argue with for the value delivered to a single media buyer or small team. The 25M ad library alone is worth more than $129/mo to most strategists.
Framework is a different commitment. It's not a tool you buy. It's a partner you hire, and the partner happens to come with the testing engine and the team that ships ads on it. The buyer is usually a director-of-growth at a $5M+/year brand who has tried the SaaS-tool stack and watched it not move the number. If you haven't tried Atria yet and your spend is under $500K/mo, start there. If Atria is already in your stack and you're still missing winners, the next move is a decision engine, not another AI agent.
Frequently asked
What buyers ask comparing these two.
- What is Raya and how does it differ from Newform's engine?
- Raya is Atria's AI strategist agent, launched February 2026 and trained on $5B+ in ad spend data. It auto-tags creatives, monitors competitors, and generates new ad concepts weekly without prompts. Framework's engine is different. It runs structured experiments on your live ads, makes scale-or-kill calls at 95% confidence, and feeds findings back into Newform's production pod. Raya generates ideas from historical patterns. Framework decides which of your ads actually win in market.
- Can Atria replace a creative team?
- Partially. Atria generates scripts and image variations, so you can ship volume without a full production team. The output works best for static creative and short-form scripts. It's weaker on UGC, founder content, and anything that needs a human on camera. Most teams using Atria still have at least an editor or a part-time content lead. If you're trying to replace a full creative team with AI alone, Atria gets you maybe 60% of the way and the gap shows up fast above $250K/mo in spend.
- Is Atria's $5B training data better than Framework running on my data?
- Different shapes. A $5B dataset gives Raya broad pattern-matching across industries. Framework runs on your account, so signal is narrower but specific to your audience and product. The honest answer: if you're early-stage and don't have enough spend yet to generate statistical signal, Atria's broader dataset is more useful. Once you're past $250K/mo and have real account-level data, in-account testing beats borrowed pattern-matching.
- Which one handles TikTok better?
- Both connect directly to Meta and TikTok. Atria's tagging works equally well across both platforms. Framework runs the same engine across Meta and TikTok, with the production team adapting format per channel. If TikTok is your primary channel, either works. Newform's edge is the production pod that knows how to ship TikTok-native creative, which Atria leaves to your team.
- What does the $129/mo Core plan actually include?
- 5 seats, 4,000 AI credits, 50 brand tracking slots, and a $500K/mo ad spend cap. AI credits cover script generation, image variations, and Raya's autonomous actions. Most solo media buyers stay inside Core comfortably. Agencies and brands above $500K/mo move to Plus at $269–$329/mo for higher caps.
- Can I switch from Atria to Framework or run both?
- Most Newform clients drop Atria after onboarding because the workflow overlap is significant. You'd be paying for two tagging engines and two competitor research surfaces. Some clients keep Atria for the competitor ad library specifically, since Newform doesn't maintain a standing library. If that's the use case, the $129/mo is fine. Anything else is duplicate spend.