Alejandro

product engineer & founder. CEO @

[Alejandro @alejandrobradf]

i just binge watched the whole stolen lego story. absolutely insane

[Alejandro @alejandrobradf]

the trap nobody warns you about: every new client makes you feel like you need to hire.

you win the account, the workload jumps, you bring someone on, and your margin is right back where it started.

you didn't grow. you just bought a bigger payroll and kept the same take home.

the ones who actually keep the money fix the workflow before they add a seat. headcount is the last lever, not the first.

we built adle because we watched too many good teams skip that order.

[Alejandro @alejandrobradf]

'ai automatically optimizes your campaigns' is the most oversold line in this space right now.

what it actually does: reallocate budget toward whatever has the best last click number in the window you handed it.

feed it bad attribution and it will scale your worst performer with total confidence.

the model isn't smart. it's fast. those are different things, and pretending they're the same is how accounts quietly bleed.

[Alejandro @alejandrobradf]

another most capable model just dropped. cool.

none of it touches the real bottleneck in paid media. the constraint was never raw intelligence. it's how many creative angles you can ship and how fast you kill the losers.

10 great ideas that never get produced lose to 50 mediocre ones that actually go live.

that gap is the whole game. it's why we built adle.

[Alejandro @alejandrobradf]

@sweatystartup the impressive ones grew up watching the work, not the results. kids copy what you do every day, not what you buy them. the struggle only teaches if they watched you struggle through it first.

[Alejandro @alejandrobradf]

most ai tools for agencies get sold on the same roi story. 'save 10 hours per client per week.'

reality: you save the execution hours but gain them back in prompt engineering, output review, and revision cycles.

net savings exist. they're just a fraction of the headline.

the use case that actually holds up is volume. 10 creative variations instead of 2. 5 campaign structures tested in parallel instead of 1. kill losers faster before they burn budget.

that's a different pitch than 'replace the human.' different expectation setting, different result.

we built adle around the volume story for this exact reason.

[Alejandro @alejandrobradf]

ios 27 is the thing performance marketers should be watching at wwdc.

ios 14 rewrote meta attribution overnight. accounts that weren't prepared lost visibility on half their conversions.

server-side events, conversion api, first-party data. not optional anymore.

the window between announcement and rollout is the only real advantage you have. use it.

been thinking about this a lot building adle, attribution chaos is usually what breaks multi-channel strategy before anything else does.

[Alejandro @alejandrobradf]

@briannekimmel the ones who do this well treat it like a media shoot, not just an event.

capture everything. build 30 pieces of content from one afternoon.

the stunt is the excuse. the content is the asset.

[Alejandro @alejandrobradf]

pmax doesn't tell you where your budget goes.

aggregate reports only. no search term breakdown. no placement visibility. no per-asset clarity.

google says trust the automation. fine. but set brand exclusions day one, run a parallel search campaign, and review asset groups weekly.

trust the machine. don't hand it the wheel and walk away.

we built adle because watching this across 10+ client accounts without automation is where margin actually disappears.

[Alejandro @alejandrobradf]

agents work best when you're replacing a repeatable workflow, not augmenting creative judgment.

find the 2 or 3 things your team does exactly the same way every time. automate those first.

the mistake is trying to use agents for everything at once. start narrow. the roi compounds once you find the right wedge.

[Alejandro @alejandrobradf]

the 'hours saved with AI' pitch doesn't survive contact with reality in most agency workflows.

what actually happens: execution hours drop, but prompt engineering, output review, and revision hours go up. the net is smaller than the headline.

and that gap widens the higher your quality bar.

the metric that matters isn't hours saved. it's output quality per unit of senior strategist time. those are very different numbers.

we ran into this building adle. the advertised metric and the real one are never the same.

[Alejandro @alejandrobradf]

the workflow that replaced 4 hours of creative production per client:

brief in → structured data model → asset generation pipeline → performance tagging → auto-upload to ad account

no designer roundtrips. no revision cycles. no manual uploads.

the real unlock isn't the generation step. it's the structured brief that makes everything downstream automatic. that's what we've been building in adle.

[Alejandro @alejandrobradf]

ugc outperforms polished creative on paid social. most teams know this and still don't test it.

the reason is simple. it looks native. the scroll punishes anything that feels like an ad.

the teams beating benchmarks consistently run both formats. same offer, same target, ugc angle vs brand creative. check hook rate at 72 hours. the data picks.

the barrier is almost always production time, not strategy. it's why we built adle.

[Alejandro @alejandrobradf]

the generalization is the problem, not the AI. tools built for every workflow serve none of them well.

the ones that actually get used are designed around one specific job, already know the context, and eliminate the mental overhead of figuring it out.

that's why we built adle for media buying teams instead of going broad.

[Alejandro @alejandrobradf]

@thejustinwelsh the process has to be the product. most people hire to cover a gap instead of closing it. every system you build is one less thing you need a person to hold.

[Alejandro @alejandrobradf]

most startups burn paid budget in 30 days and walk away calling it 'ads don't work.'

the channel works. the creative didn't.

real testing isn't 3 variations of one angle. it's 10 angles, 30 creatives, cut losers in week 1, scale winners in week 2.

the unit economics only work when creative volume is high enough to find what actually converts.

we built adle because most teams never get past coin-flip testing before they give up on the channel.

[Alejandro @alejandrobradf]

building adle, week 6.

biggest insight so far: agencies don't want 'ai-generated creative.' they want creative that doesn't eat 4 hours of revision.

the framing was wrong for months. we were pitching speed. they care about margin.

same product. completely different pitch.

the ones who convert immediately aren't the fastest teams. they're the most frustrated with the current workflow.

[Alejandro @alejandrobradf]

anthropic says claude is boosting dev output 8x.

engineers noticed first. marketers are still catching up.

the same shift is hitting ad creative. not just faster production. 8x the creative angles tested per client, same team size.

the teams building for that now don't need to hire their next 3 people to handle the next 3 clients.

we built adle watching agencies produce 2 creatives a week when they should be testing 20.

[Alejandro @alejandrobradf]

the promise is always the button. the reality is mapping the lawn, calibrating the gps, explaining where the edges are.

same gap in ai for ads. the ones that work aren't autonomous. they're fast at the repetitive parts. the setup still requires a human who knows what good looks like.

the implementation tax doesn't go away. it just moves.

[Alejandro @alejandrobradf]

@hnshah the ones who fix this stop being bottlenecked by individual talent and start compounding it.