J Filipe

Building @Costlens_dev — prove your AI tools pay for themselves. | Solo founder #BuildInPublic

[J Filipe @jrmromao]

Unpopular opinion: A lot of engineering teams are losing money on AI dev tools right now—not because of seat licenses, but because of the hidden "code review tax."

Lowering the friction to write code floods your pipeline with massive PRs that senior devs have to untangle.

Put together a framework on how to measure the actual business value of AI coding, minus the hype.

[J Filipe @jrmromao]

Oracle, Microsoft, Amazon, Meta, Google... they're collectively facing a $7.9 TRILLION AI cash crunch by 2031. Is anyone truly confident they're getting ROI on every dollar? I doubt it. This AI spend is already out of control.

[J Filipe @jrmromao]

@lucas_montano Pelo fim de tech debt

[J Filipe @jrmromao]

OpenAI just launched a $500/month 'Pro 500' plan for Ultrafast GPT-6, while existing Pro users saw their compute halved. The cost of "AI speed" is escalating, and it's getting harder to justify. Are engineering leaders ready for this new reality?

[J Filipe @jrmromao]

Michael Burry warns of "massive write-offs" for AI commitments, with Alphabet nearing $900B. It's not just big tech. We've seen dev teams blow their AI budget in weeks. Are we collectively sleepwalking into a major AI cost crisis?

[J Filipe @jrmromao]

@boardyai Here https://t.co/Ng2v4snihN

[J Filipe @jrmromao]

The new 'dumb AI' Jev model reviews 1000 PRs for 7 cents. Opus 5? $14.5 for the same task. This isn't just a cost difference; it's a wake-up call on AI spend. Dev teams, are you tracking these insane deltas? It’s not just about LLMs. https://t.co/pr8nYqYI1l

[J Filipe @jrmromao]

Why is everyone talking about JEV?

[J Filipe @jrmromao]

Goldman Sachs projects hyperscaler CapEx to exceed $760B by 2026, and IDC says 46.9% of enterprises were over budget on AI. That's real money. CostLens shows dev teams how to slash 20%+ of their token spend by identifying cheaper model routes in real-time. Imagine that chart!

[J Filipe @jrmromao]

@eliana_jordan I thought you were from Greece.

[J Filipe @jrmromao]

@dadhalfdev Então. Let's connect!

[J Filipe @jrmromao]

Nearly half of large companies (46.9%) overshot their AI budgets in the second half of 2026. This isn't just theory – I see dev teams wrestling with spiralling costs daily. With CostLens, the insight often looks like 30% of calls routed to expensive models were unnecessary.

[J Filipe @jrmromao]

@TTrimoreau id be an happy man with 10 test users

[J Filipe @jrmromao]

CloudZero's data is eye-opening: 28.2% of companies now put over 10% of their cloud spend into AI. That's a 4.3% jump in just a month. Add to that Reco's finding that 80% of AI tools run without IT oversight. Are we simply throwing money into a black hole? The risks are huge.

[J Filipe @jrmromao]

OpenAI dropping GPT-6 Astra is huge. But their tip to optimize “reasoning effort” for cost savings? That’s the real talk. Dev teams using CostLens are consistently cutting 20-30% on token spend just by tweaking these params. Imagine the budget impact.

[J Filipe @jrmromao]

That major AI outage yesterday across ChatGPT, Claude, and Grok? A stark reminder we need robust fallbacks. Or are we just too deep now? What's your dev team's plan B when the models go dark?

[J Filipe @jrmromao]

@bridgemindai Everyone talks about AGI every time a new model is released

[J Filipe @jrmromao]

Dunstan Research Group just reported a 500% increase in median code review time for AI-adopting teams. This isn't just about output; it's about the quality control overhead. AI might be fast, but it's not always cheap when you factor in human review time.

[J Filipe @jrmromao]

Atlassian's AI bill hit $15M/month, and they're giving devs 'wallets'. The real insight? Most teams don't even *know* where their AI spend is going until it's too late. CostLens helps pinpoint those unexpected spikes and routes you to cheaper models before you hit that wall

[J Filipe @jrmromao]

Microsoft just reported one employee racked up nearly $28,000 in AI usage costs in a single month. That's not just a rounding error; it's a wake-up call. Are we really tracking these spiraling costs?

[J Filipe @jrmromao]

https://t.co/IJqa9t5O8M dropping GLM-4.5 at $0.60/M tokens is huge. But knowing *when* and *where* to route to cheaper models, and actually proving that cost saving to the team? That's where it gets tricky. We built CostLens for exactly that. https://t.co/zr7hbQMTKM