Thinslices

Thinslices We build digital products that exceed even our own expectations.

Inference workloads now account for two-thirds of all AI compute, up from one-third in 2023. That shift has put inferenc...
05/06/2026

Inference workloads now account for two-thirds of all AI compute, up from one-third in 2023. That shift has put inference at the center of infrastructure strategy, and the economics of running it entirely in the cloud are getting harder to ignore.

Our team broke down what edge AI inference means for product architecture, and which workloads are actually suited to the move.
Link in the comments.

The manual for agentic AI does not exist yet. Every company experimenting with it right now is doing the same thing: try...
03/06/2026

The manual for agentic AI does not exist yet. Every company experimenting with it right now is doing the same thing: trying to figure out which combination of roles, briefings, workflows and review gates produces something reliable. Nobody has the definitive answer, and the teams that will have it in twelve months are the ones starting with the small, sequential, well-briefed version today.

Link in comments.

Ionut is a PM and AI Product Builder at Thinslices, and the difference between his writing and most of what's published ...
02/06/2026

Ionut is a PM and AI Product Builder at Thinslices, and the difference between his writing and most of what's published on agentic AI right now is that he runs this system on real projects. Not theoretical workflows, but the setup he uses week to week.

Part one argued that AI becomes useful when grounded in structured context rather than clever prompting. Part two is the practical follow-up: the minimum viable agentic setup that gives every agent what it needs before it starts, without overwhelming a team still figuring out how the machine works.

The product vision document. The MD files per role. The handoff file that solves Claude's lack of session memory. The folder structure that turns concepts into something teams can actually use. If part one was the why, this is the how. Link in the comments.

The practical question most agentic setups don't answer: how does the agent know how to behave?What conventions to follo...
01/06/2026

The practical question most agentic setups don't answer: how does the agent know how to behave?
What conventions to follow. What to read before starting. What the team's definition of done actually is.

The answer is an MD file. One per role. Written the same way you'd write a job description for a new hire. Link in comments.

Automating a process rarely means removing humans entirely, and treating that as the goal is one of the more common ways...
31/05/2026

Automating a process rarely means removing humans entirely, and treating that as the goal is one of the more common ways AI projects underdeliver. https://hubs.li/Q04hZ6ns0

Two AI projects from our current portfolio look completely different on the surface. One processes utility invoices for ...
30/05/2026

Two AI projects from our current portfolio look completely different on the surface. One processes utility invoices for a US energy platform. The other screens academic manuscripts for a major publisher.

What they share is the shape of the problem. A team doing repetitive work at high volume, a defined standard for correct, consistent inputs, and a clear cost to the manual baseline. Read the full article - link in comments.

The best advice anyone gets about learning to code is: build a to-do list. Not because a to-do list is interesting. Beca...
29/05/2026

The best advice anyone gets about learning to code is: build a to-do list. Not because a to-do list is interesting. Because it contains every fundamental concept you need to understand before building anything else. https://hubs.li/Q04hgkb50

Five questions to ask before scoping any AI automation project:• How many times does this process run per month, and wha...
29/05/2026

Five questions to ask before scoping any AI automation project:
• How many times does this process run per month, and what's the cost of each repetition?
• What does "correct" look like, and how would you know if the output was wrong?
• How structured are the inputs?
• Who currently does this work, and what happens to their time if it's automated?
• Is there historical data available?
Strong candidates score well on all five. The full breakdown is in the link below.

The AI projects that deliver durable returns share the same profile, regardless of industry. High volume. A defined qual...
28/05/2026

The AI projects that deliver durable returns share the same profile, regardless of industry. High volume. A defined quality standard. Structured or semi-structured inputs. Current manual effort with a visible labor cost. Link in comments.

Anthropic is shipping new agentic features faster than most teams can absorb them. The instinct is to explore everything...
27/05/2026

Anthropic is shipping new agentic features faster than most teams can absorb them. The instinct is to explore everything. That instinct is the problem. Part two of Ionut Lomer's series on AI-native delivery is live.

Part one argued that the gap that matters in 2026 isn't which model a team uses. It's whether the team has built infrastructure around it.Part two is the practical follow-up. The agentic AI starter kit: the minimum viable setup for software teams who want real value from agents without shipping production disasters at 3am.

Link in the comments.

Address

Palas Street, No. 5B, D3 Building, Lasi
Iasi
700051

Opening Hours

Monday 09:00 - 18:00
Tuesday 09:00 - 18:00
Wednesday 09:00 - 18:00
Thursday 09:00 - 18:00
Friday 09:00 - 18:00

Telephone

+40743209998

Website

Alerts

Be the first to know and let us send you an email when Thinslices posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Contact The Business

Send a message to Thinslices:

Shortcuts

Share

Category